scientific writing tips in research

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• Prev Chronic Dis

Successful Scientific Writing and Publishing: A Step-by-Step Approach

John k. iskander.

1 Centers for Disease Control and Prevention, Atlanta, Georgia

Sara Beth Wolicki

2 Association of Schools and Programs of Public Health, Washington, District of Columbia

Rebecca T. Leeb

Paul z. siegel.

Scientific writing and publication are essential to advancing knowledge and practice in public health, but prospective authors face substantial challenges. Authors can overcome barriers, such as lack of understanding about scientific writing and the publishing process, with training and resources. The objective of this article is to provide guidance and practical recommendations to help both inexperienced and experienced authors working in public health settings to more efficiently publish the results of their work in the peer-reviewed literature. We include an overview of basic scientific writing principles, a detailed description of the sections of an original research article, and practical recommendations for selecting a journal and responding to peer review comments. The overall approach and strategies presented are intended to contribute to individual career development while also increasing the external validity of published literature and promoting quality public health science.

Introduction

Publishing in the peer-reviewed literature is essential to advancing science and its translation to practice in public health ( 1 , 2 ). The public health workforce is diverse and practices in a variety of settings ( 3 ). For some public health professionals, writing and publishing the results of their work is a requirement. Others, such as program managers, policy makers, or health educators, may see publishing as being outside the scope of their responsibilities ( 4 ).

Disseminating new knowledge via writing and publishing is vital both to authors and to the field of public health ( 5 ). On an individual level, publishing is associated with professional development and career advancement ( 6 ). Publications share new research, results, and methods in a trusted format and advance scientific knowledge and practice ( 1 , 7 ). As more public health professionals are empowered to publish, the science and practice of public health will advance ( 1 ).

Unfortunately, prospective authors face barriers to publishing their work, including navigating the process of scientific writing and publishing, which can be time-consuming and cumbersome. Often, public health professionals lack both training opportunities and understanding of the process ( 8 ). To address these barriers and encourage public health professionals to publish their findings, the senior author (P.Z.S.) and others developed Successful Scientific Writing (SSW), a course about scientific writing and publishing. Over the past 30 years, this course has been taught to thousands of public health professionals, as well as hundreds of students at multiple graduate schools of public health. An unpublished longitudinal survey of course participants indicated that two-thirds agreed that SSW had helped them to publish a scientific manuscript or have a conference abstract accepted. The course content has been translated into this manuscript. The objective of this article is to provide prospective authors with the tools needed to write original research articles of high quality that have a good chance of being published.

Basic Recommendations for Scientific Writing

Prospective authors need to know and tailor their writing to the audience. When writing for scientific journals, 4 fundamental recommendations are: clearly stating the usefulness of the study, formulating a key message, limiting unnecessary words, and using strategic sentence structure.

To demonstrate usefulness, focus on how the study addresses a meaningful gap in current knowledge or understanding. What critical piece of information does the study provide that will help solve an important public health problem? For example, if a particular group of people is at higher risk for a specific condition, but the magnitude of that risk is unknown, a study to quantify the risk could be important for measuring the population’s burden of disease.

Scientific articles should have a clear and concise take-home message. Typically, this is expressed in 1 to 2 sentences that summarize the main point of the paper. This message can be used to focus the presentation of background information, results, and discussion of findings. As an early step in the drafting of an article, we recommend writing out the take-home message and sharing it with co-authors for their review and comment. Authors who know their key point are better able to keep their writing within the scope of the article and present information more succinctly. Once an initial draft of the manuscript is complete, the take-home message can be used to review the content and remove needless words, sentences, or paragraphs.

Concise writing improves the clarity of an article. Including additional words or clauses can divert from the main message and confuse the reader. Additionally, journal articles are typically limited by word count. The most important words and phrases to eliminate are those that do not add meaning, or are duplicative. Often, cutting adjectives or parenthetical statements results in a more concise paper that is also easier to read.

Sentence structure strongly influences the readability and comprehension of journal articles. Twenty to 25 words is a reasonable range for maximum sentence length. Limit the number of clauses per sentence, and place the most important or relevant clause at the end of the sentence ( 9 ). Consider the sentences:

• By using these tips and tricks, an author may write and publish an additional 2 articles a year.
• An author may write and publish an additional 2 articles a year by using these tips and tricks.

The focus of the first sentence is on the impact of using the tips and tricks, that is, 2 more articles published per year. In contrast, the second sentence focuses on the tips and tricks themselves.

Authors should use the active voice whenever possible. Consider the following example:

• Active voice: Authors who use the active voice write more clearly.
• Passive voice: Clarity of writing is promoted by the use of the active voice.

The active voice specifies who is doing the action described in the sentence. Using the active voice improves clarity and understanding, and generally uses fewer words. Scientific writing includes both active and passive voice, but authors should be intentional with their use of either one.

Sections of an Original Research Article

Original research articles make up most of the peer-reviewed literature ( 10 ), follow a standardized format, and are the focus of this article. The 4 main sections are the introduction, methods, results, and discussion, sometimes referred to by the initialism, IMRAD. These 4 sections are referred to as the body of an article. Two additional components of all peer-reviewed articles are the title and the abstract. Each section’s purpose and key components, along with specific recommendations for writing each section, are listed below.

Title. The purpose of a title is twofold: to provide an accurate and informative summary and to attract the target audience. Both prospective readers and database search engines use the title to screen articles for relevance ( 2 ). All titles should clearly state the topic being studied. The topic includes the who, what, when, and where of the study. Along with the topic, select 1 or 2 of the following items to include within the title: methods, results, conclusions, or named data set or study. The items chosen should emphasize what is new and useful about the study. Some sources recommend limiting the title to less than 150 characters ( 2 ). Articles with shorter titles are more frequently cited than articles with longer titles ( 11 ). Several title options are possible for the same study ( Figure ).

Two examples of title options for a single study.

Abstract . The abstract serves 2 key functions. Journals may screen articles for potential publication by using the abstract alone ( 12 ), and readers may use the abstract to decide whether to read further. Therefore, it is critical to produce an accurate and clear abstract that highlights the major purpose of the study, basic procedures, main findings, and principal conclusions ( 12 ). Most abstracts have a word limit and can be either structured following IMRAD, or unstructured. The abstract needs to stand alone from the article and tell the most important parts of the scientific story up front.

Introduction . The purpose of the introduction is to explain how the study sought to create knowledge that is new and useful. The introduction section may often require only 3 paragraphs. First, describe the scope, nature, or magnitude of the problem being addressed. Next, clearly articulate why better understanding this problem is useful, including what is currently known and the limitations of relevant previous studies. Finally, explain what the present study adds to the knowledge base. Explicitly state whether data were collected in a unique way or obtained from a previously unstudied data set or population. Presenting both the usefulness and novelty of the approach taken will prepare the reader for the remaining sections of the article.

Methods . The methods section provides the information necessary to allow others, given the same data, to recreate the analysis. It describes exactly how data relevant to the study purpose were collected, organized, and analyzed. The methods section describes the process of conducting the study — from how the sample was selected to which statistical methods were used to analyze the data. Authors should clearly name, define, and describe each study variable. Some journals allow detailed methods to be included in an appendix or supplementary document. If the analysis involves a commonly used public health data set, such as the Behavioral Risk Factor Surveillance System ( 13 ), general aspects of the data set can be provided to readers by using references. Because what was done is typically more important than who did it, use of the passive voice is often appropriate when describing methods. For example, “The study was a group randomized, controlled trial. A coin was tossed to select an intervention group and a control group.”

Results . The results section describes the main outcomes of the study or analysis but does not interpret the findings or place them in the context of previous research. It is important that the results be logically organized. Suggested organization strategies include presenting results pertaining to the entire population first, and then subgroup analyses, or presenting results according to increasing complexity of analysis, starting with demographic results before proceeding to univariate and multivariate analyses. Authors wishing to draw special attention to novel or unexpected results can present them first.

One strategy for writing the results section is to start by first drafting the figures and tables. Figures, which typically show trends or relationships, and tables, which show specific data points, should each support a main outcome of the study. Identify the figures and tables that best describe the findings and relate to the study’s purpose, and then develop 1 to 2 sentences summarizing each one. Data not relevant to the study purpose may be excluded, summarized briefly in the text, or included in supplemental data sets. When finalizing figures, ensure that axes are labeled and that readers can understand figures without having to refer to accompanying text.

Discussion . In the discussion section, authors interpret the results of their study within the context of both the related literature and the specific scientific gap the study was intended to fill. The discussion does not introduce results that were not presented in the results section. One way authors can focus their discussion is to limit this section to 4 paragraphs: start by reinforcing the study’s take-home message(s), contextualize key results within the relevant literature, state the study limitations, and lastly, make recommendations for further research or policy and practice changes. Authors can support assertions made in the discussion with either their own findings or by referencing related research. By interpreting their own study results and comparing them to others in the literature, authors can emphasize findings that are unique, useful, and relevant. Present study limitations clearly and without apology. Finally, state the implications of the study and provide recommendations or next steps, for example, further research into remaining gaps or changes to practice or policy. Statements or recommendations regarding policy may use the passive voice, especially in instances where the action to be taken is more important than who will implement the action.

Beginning the Writing Process

The process of writing a scientific article occurs before, during, and after conducting the study or analyses. Conducting a literature review is crucial to confirm the existence of the evidence gap that the planned analysis seeks to fill. Because literature searches are often part of applying for research funding or developing a study protocol, the citations used in the grant application or study proposal can also be used in subsequent manuscripts. Full-text databases such as PubMed Central ( 14 ), NIH RePORT ( 15 ), and CDC Stacks ( 16 ) can be useful when performing literature reviews. Authors should familiarize themselves with databases that are accessible through their institution and any assistance that may be available from reference librarians or interlibrary loan systems. Using citation management software is one way to establish and maintain a working reference list. Authors should clearly understand the distinction between primary and secondary references, and ensure that they are knowledgeable about the content of any primary or secondary reference that they cite.

Review of the literature may continue while organizing the material and writing begins. One way to organize material is to create an outline for the paper. Another way is to begin drafting small sections of the article such as the introduction. Starting a preliminary draft forces authors to establish the scope of their analysis and clearly articulate what is new and novel about the study. Furthermore, using information from the study protocol or proposal allows authors to draft the methods and part of the results sections while the study is in progress. Planning potential data comparisons or drafting “table shells” will help to ensure that the study team has collected all the necessary data. Drafting these preliminary sections early during the writing process and seeking feedback from co-authors and colleagues may help authors avoid potential pitfalls, including misunderstandings about study objectives.

The next step is to conduct the study or analyses and use the resulting data to fill in the draft table shells. The initial results will most likely require secondary analyses, that is, exploring the data in ways in addition to those originally planned. Authors should ensure that they regularly update their methods section to describe all changes to data analysis.

After completing table shells, authors should summarize the key finding of each table or figure in a sentence or two. Presenting preliminary results at meetings, conferences, and internal seminars is an established way to solicit feedback. Authors should pay close attention to questions asked by the audience, treating them as an informal opportunity for peer review. On the basis of the questions and feedback received, authors can incorporate revisions and improvements into subsequent drafts of the manuscript.

The relevant literature should be revisited periodically while writing to ensure knowledge of the most recent publications about the manuscript topic. Authors should focus on content and key message during the process of writing the first draft and should not spend too much time on issues of grammar or style. Drafts, or portions of drafts, should be shared frequently with trusted colleagues. Their recommendations should be reviewed and incorporated when they will improve the manuscript’s overall clarity.

For most authors, revising drafts of the manuscript will be the most time-consuming task involved in writing a paper. By regularly checking in with coauthors and colleagues, authors can adopt a systematic approach to rewriting. When the author has completed a draft of the manuscript, he or she should revisit the key take-home message to ensure that it still matches the final data and analysis. At this point, final comments and approval of the manuscript by coauthors can be sought.

Authors should then seek to identify journals most likely to be interested in considering the study for publication. Initial questions to consider when selecting a journal include:

• Which audience is most interested in the paper’s message?
• Would clinicians, public health practitioners, policy makers, scientists, or a broader audience find this useful in their field or practice?
• Do colleagues have prior experience submitting a manuscript to this journal?
• Is the journal indexed and peer-reviewed?
• Is the journal subscription or open-access and are there any processing fees?
• How competitive is the journal?

Authors should seek to balance the desire to be published in a top-tier journal (eg, Journal of the American Medical Association, BMJ, or Lancet) against the statistical likelihood of rejection. Submitting the paper initially to a journal more focused on the paper’s target audience may result in a greater chance of acceptance, as well as more timely dissemination of findings that can be translated into practice. Most of the 50 to 75 manuscripts published each week by authors from the Centers for Disease Control and Prevention (CDC) are published in specialty and subspecialty journals, rather than in top-tier journals ( 17 ).

The target journal’s website will include author guidelines, which will contain specific information about format requirements (eg, font, line spacing, section order, reference style and limit, table and figure formatting), authorship criteria, article types, and word limits for articles and abstracts.

We recommend returning to the previously drafted abstract and ensuring that it complies with the journal’s format and word limit. Authors should also verify that any changes made to the methods or results sections during the article’s drafting are reflected in the final version of the abstract. The abstract should not be written hurriedly just before submitting the manuscript; it is often apparent to editors and reviewers when this has happened. A cover letter to accompany the submission should be drafted; new and useful findings and the key message should be included.

Before submitting the manuscript and cover letter, authors should perform a final check to ensure that their paper complies with all journal requirements. Journals may elect to reject certain submissions on the basis of review of the abstract, or may send them to peer reviewers (typically 2 or 3) for consultation. Occasionally, on the basis of peer reviews, the journal will request only minor changes before accepting the paper for publication. Much more frequently, authors will receive a request to revise and resubmit their manuscript, taking into account peer review comments. Authors should recognize that while revise-and-resubmit requests may state that the manuscript is not acceptable in its current form, this does not constitute a rejection of the article. Authors have several options in responding to peer review comments:

• Performing additional analyses and updating the article appropriately
• Declining to perform additional analyses, but providing an explanation (eg, because the requested analysis goes beyond the scope of the article)
• Providing updated references
• Acknowledging reviewer comments that are simply comments without making changes

In addition to submitting a revised manuscript, authors should include a cover letter in which they list peer reviewer comments, along with the revisions they have made to the manuscript and their reply to the comment. The tone of such letters should be thankful and polite, but authors should make clear areas of disagreement with peer reviewers, and explain why they disagree. During the peer review process, authors should continue to consult with colleagues, especially ones who have more experience with the specific journal or with the peer review process.

There is no secret to successful scientific writing and publishing. By adopting a systematic approach and by regularly seeking feedback from trusted colleagues throughout the study, writing, and article submission process, authors can increase their likelihood of not only publishing original research articles of high quality but also becoming more scientifically productive overall.

Acknowledgments

The authors acknowledge PCD ’s former Associate Editor, Richard A. Goodman, MD, MPH, who, while serving as Editor in Chief of CDC’s Morbidity and Mortality Weekly Report Series, initiated a curriculum on scientific writing for training CDC’s Epidemic Intelligence Service Officers and other CDC public health professionals, and with whom the senior author of this article (P.Z.S.) collaborated in expanding training methods and contents, some of which are contained in this article. The authors acknowledge Juan Carlos Zevallos, MD, for his thoughtful critique and careful editing of previous Successful Scientific Writing materials. We also thank Shira Eisenberg for editorial assistance with the manuscript. This publication was supported by the Cooperative Agreement no. 1U360E000002 from CDC and the Association of Schools and Programs of Public Health. The findings and conclusions of this article do not necessarily represent the official views of CDC or the Association of Schools and Programs of Public Health. Names of journals and citation databases are provided for identification purposes only and do not constitute any endorsement by CDC.

The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors' affiliated institutions.

Suggested citation for this article: Iskander JK, Wolicki SB, Leeb RT, Siegel PZ. Successful Scientific Writing and Publishing: A Step-by-Step Approach. Prev Chronic Dis 2018;15:180085. DOI: https://doi.org/10.5888/pcd15.180085 .

When you choose to publish with PLOS, your research makes an impact. Make your work accessible to all, without restrictions, and accelerate scientific discovery with options like preprints and published peer review that make your work more Open.

• PLOS Biology
• PLOS Climate
• PLOS Complex Systems
• PLOS Computational Biology
• PLOS Digital Health
• PLOS Genetics
• PLOS Global Public Health
• PLOS Medicine
• PLOS Mental Health
• PLOS Neglected Tropical Diseases
• PLOS Pathogens
• PLOS Sustainability and Transformation
• PLOS Collections

Welcome to the PLOS Writing Center

Your source for scientific writing & publishing essentials.

A collection of free, practical guides and hands-on resources for authors looking to improve their scientific publishing skillset.

ARTICLE-WRITING ESSENTIALS

Your title is the first thing anyone who reads your article is going to see, and for many it will be where they stop reading. Learn how to write a title that helps readers find your article, draws your audience in and sets the stage for your research!

The abstract is your chance to let your readers know what they can expect from your article. Learn how to write a clear, and concise abstract that will keep your audience reading.

A clear methods section impacts editorial evaluation and readers’ understanding, and is also the backbone of transparency and replicability. Learn what to include in your methods section, and how much detail is appropriate.

In many fields, a statistical analysis forms the heart of both the methods and results sections of a manuscript. Learn how to report statistical analyses, and what other context is important for publication success and future reproducibility.

The discussion section contains the results and outcomes of a study. An effective discussion informs readers what can be learned from your experiment and provides context for the results.

Ensuring your manuscript is well-written makes it easier for editors, reviewers and readers to understand your work. Avoiding language errors can help accelerate review and minimize delays in the publication of your research.

The PLOS Writing Toolbox

Delivered to your inbox every two weeks, the Writing Toolbox features practical advice and tools you can use to prepare a research manuscript for submission success and build your scientific writing skillset. 

Discover how to navigate the peer review and publishing process, beyond writing your article.

The path to publication can be unsettling when you’re unsure what’s happening with your paper. Learn about staple journal workflows to see the detailed steps required for ensuring a rigorous and ethical publication.

Reputable journals screen for ethics at submission—and inability to pass ethics checks is one of the most common reasons for rejection. Unfortunately, once a study has begun, it’s often too late to secure the requisite ethical reviews and clearances. Learn how to prepare for publication success by ensuring your study meets all ethical requirements before work begins.

From preregistration, to preprints, to publication—learn how and when to share your study.

How you store your data matters. Even after you publish your article, your data needs to be accessible and useable for the long term so that other researchers can continue building on your work. Good data management practices make your data discoverable and easy to use, promote a strong foundation for reproducibility and increase your likelihood of citations.

You’ve just spent months completing your study, writing up the results and submitting to your top-choice journal. Now the feedback is in and it’s time to revise. Set out a clear plan for your response to keep yourself on-track and ensure edits don’t fall through the cracks.

There’s a lot to consider when deciding where to submit your work. Learn how to choose a journal that will help your study reach its audience, while reflecting your values as a researcher.

Are you actively preparing a submission for a PLOS journal? Select the relevant journal below for more detailed guidelines. 

How to Write an Article  

Share the lessons of the Writing Center in a live, interactive training.

Access tried-and-tested training modules, complete with slides and talking points, workshop activities, and more.

• The Scientist University

The Fundamentals of Academic Science Writing

Writing is an essential skill for scientists, and learning how to write effectively starts with good fundamentals and lots of practice..

Nathan Ni holds a PhD from Queens University. He is a science editor for The Scientist’s Creative Services Team who strives to better understand and communicate the relationships between health and disease.

View full profile.

Learn about our editorial policies.

Writing is a big part of being a scientist, whether in the form of manuscripts, grants, reports, protocols, presentations, or even emails. However, many people look at writing as separate from science—a scientist writes, but scientists are not regarded as writers. 1 This outdated assertion means that writing and communication has been historically marginalized when it comes to training and educating new scientists. In truth, being a professional writer is part of being a scientist . 1 In today’s hypercompetitive academic environment, scientists need to be as proficient with the pen as they are with the pipette in order to showcase their work. 

Using the Active Voice

Stereotypical academic writing is rigid, dry, and mechanical, delivering prose that evokes memories of high school and undergraduate laboratory reports. The hallmark of this stereotype is passive voice overuse. In writing, the passive voice is when the action comes at the end of a clause—for example, “the book was opened”. In scientific writing, it is particularly prevalent when detailing methodologies and results. How many times have we seen something like “citric acid was added to the solution, resulting in a two-fold reduction in pH” rather than “adding citric acid to the solution reduced the pH two-fold”?

Scientists should write in the active voice as much as possible. However, the active voice tends to place much more onus on the writer’s perspective, something that scientists have historically been instructed to stay away from. For example, “we treated the cells with phenylephrine” places much more emphasis on the operator than “the cells were treated with phenylephrine.” Furthermore, pronoun usage in academic writing is traditionally discouraged, but it is much harder, especially for those with non-native English proficiency, to properly use active voice without them. 

Things are changing though, and scientists are recognizing the importance of giving themselves credit. Many major journals, including Nature , Science , PLoS One , and PNAS allow pronouns in their manuscripts, and prominent style guides such as APA even recommend using first-person pronouns, as traditional third-person writing can be ambiguous. 2 It is vital that a manuscript clearly and definitively highlights and states what the authors specifically did that was so important or novel, in contrast to what was already known. A simple “we found…” statement in the abstract and the introduction goes a long way towards giving readers the hook that they need to read further.

Keeping Sentences Simple

Writing in the active voice also makes it easier to organize manuscripts and construct arguments. Active voice uses fewer words than passive voice to explain the same concept. It also introduces argument components sequentially—subject, claim, and then evidence—whereas passive voice introduces claim and evidence before the subject. Compare, for example, “T cell abundance did not differ between wildtype and mutant mice” versus “there was no difference between wildtype and mutant mice in terms of T cell abundance.” T cell abundance, as the measured parameter, is the most important part of the sentence, but it is only introduced at the very end of the latter example.

The sequential nature of active voice therefore makes it easier to not get bogged down in overloading the reader with clauses and adhering to a general principle of “one sentence, one concept (or idea, or argument).” Consider the following sentence: 

Research on CysLT 2 R , expressed in humans in umbilical vein endothelial cells, macrophages, platelets, the cardiac Purkinje system, and coronary endothelial cells , had been hampered by a lack of selective pharmacological agents , the majority of work instead using the nonselective cysLT antagonist/partial agonist Bay-u9773 or genetic models of CysLT 2 R expression modulation) .

The core message of this sentence is that CysLT 2 R research is hampered by a lack of selective pharmacological agents, but that message is muddled by the presence of two other major pieces of information: where CysLT 2 R is expressed and what researchers used to study CysLT 2 R instead of selective pharmacological agents. Because this sentence contains three main pieces of information, it is better to break it up into three separate sentences for clarity.

In humans, CysLT 2 R is expressed in umbilical vein endothelial cells, macrophages, platelets, the cardiac Purkinje system, and coronary endothelial cells . CysLT 2 R research has been hampered by a lack of selective pharmacological agents . Instead, the majority of work investigating the receptor has used either the nonselective cysLT antagonist/partial agonist Bay-u9773 or genetic models of CysLT 2 R expression modulation.

The Right Way to Apply Jargon

There is another key advantage to organizing sentences in this simple manner: it lets scientists manage how jargon is introduced to the reader. Jargon—special words used within a specific field or on a specific topic—is necessary in scientific writing. It is critical for succinctly describing key elements and explaining key concepts. But too much jargon can make a manuscript unreadable, either because the reader does not understand the terminology or because they are bogged down in reading all of the definitions. 

The key to using jargon is to make it as easy as possible for the audience. General guidelines instruct writers to define new terms only when they are first used. However, it is cumbersome for a reader to backtrack considerable distances in a manuscript to look up a definition. If a term is first introduced in the introduction but not mentioned again until the discussion, the writer should re-define the term in a more casual manner. For example: “PI3K can be reversibly inhibited by LY294002 and irreversibly inhibited by wortmannin” in the introduction, accompanied by “when we applied the PI3K inhibitor LY294002” for the discussion. This not only makes things easier for the reader, but it also re-emphasizes what the scientist did and the results they obtained.

Practice Makes Better

Finally, the most important fundamental for science writing is to not treat it like a chore or a nuisance. Just as a scientist optimizes a bench assay through repeated trial and error, combined with literature reviews on what steps others have implemented, a scientist should practice, nurture, and hone their writing skills through repeated drafting, editing, and consultation. Do not be afraid to write. Putting pen to paper can help organize one’s thoughts, expose next steps for exploration, or even highlight additional experiments required to patch knowledge or logic gaps in existing studies. 

Looking for more information on scientific writing? Check out The Scientist’s TS SciComm  section. Looking for some help putting together a manuscript, a figure, a poster, or anything else? The Scientist’s Scientific Services  may have the professional help that you need.

• Schimel J. Writing Science: How to Write Papers That Get Cited And Proposals That Get Funded . Oxford University Press; 2012.
• First-person pronouns. American Psychological Association. Updated July 2022. Accessed March 2024. https://apastyle.apa.org/style-grammar-guidelines/grammar/first-person-pronouns  

Related community Research Resources

How Cloud Labs and Remote Research Shape Science 

Artificial Neural Networks: Learning by Doing

Peer Profile Program

What this handout is about

Nearly every element of style that is accepted and encouraged in general academic writing is also considered good practice in scientific writing. The major difference between science writing and writing in other academic fields is the relative importance placed on certain stylistic elements. This handout details the most critical aspects of scientific writing and provides some strategies for evaluating and improving your scientific prose. Readers of this handout may also find our handout on scientific reports useful.

What is scientific writing?

There are several different kinds of writing that fall under the umbrella of scientific writing. Scientific writing can include:

• Peer-reviewed journal articles (presenting primary research)
• Grant proposals (you can’t do science without funding)
• Literature review articles (summarizing and synthesizing research that has already been carried out)

As a student in the sciences, you are likely to spend some time writing lab reports, which often follow the format of peer-reviewed articles and literature reviews. Regardless of the genre, though, all scientific writing has the same goal: to present data and/or ideas with a level of detail that allows a reader to evaluate the validity of the results and conclusions based only on the facts presented. The reader should be able to easily follow both the methods used to generate the data (if it’s a primary research paper) and the chain of logic used to draw conclusions from the data. Several key elements allow scientific writers to achieve these goals:

• Precision: ambiguities in writing cause confusion and may prevent a reader from grasping crucial aspects of the methodology and synthesis
• Clarity: concepts and methods in the sciences can often be complex; writing that is difficult to follow greatly amplifies any confusion on the part of the reader
• Objectivity: any claims that you make need to be based on facts, not intuition or emotion

How can I make my writing more precise?

Theories in the sciences are based upon precise mathematical models, specific empirical (primary) data sets, or some combination of the two. Therefore, scientists must use precise, concrete language to evaluate and explain such theories, whether mathematical or conceptual. There are a few strategies for avoiding ambiguous, imprecise writing.

Word and phrasing choice

Often several words may convey similar meaning, but usually only one word is most appropriate in a given context. Here’s an example:

• Word choice 1: “population density is positively correlated with disease transmission rate”
• Word choice 2: “population density is positively related to disease transmission rate”

In some contexts, “correlated” and “related” have similar meanings. But in scientific writing, “correlated” conveys a precise statistical relationship between two variables. In scientific writing, it is typically not enough to simply point out that two variables are related: the reader will expect you to explain the precise nature of the relationship (note: when using “correlation,” you must explain somewhere in the paper how the correlation was estimated). If you mean “correlated,” then use the word “correlated”; avoid substituting a less precise term when a more precise term is available.

This same idea also applies to choice of phrasing. For example, the phrase “writing of an investigative nature” could refer to writing in the sciences, but might also refer to a police report. When presented with a choice, a more specific and less ambiguous phraseology is always preferable. This applies even when you must be repetitive to maintain precision: repetition is preferable to ambiguity. Although repetition of words or phrases often happens out of necessity, it can actually be beneficial by placing special emphasis on key concepts.

Figurative language

Figurative language can make for interesting and engaging casual reading but is by definition imprecise. Writing “experimental subjects were assaulted with a wall of sound” does not convey the precise meaning of “experimental subjects were presented with 20 second pulses of conspecific mating calls.” It’s difficult for a reader to objectively evaluate your research if details are left to the imagination, so exclude similes and metaphors from your scientific writing.

Level of detail

Include as much detail as is necessary, but exclude extraneous information. The reader should be able to easily follow your methodology, results, and logic without being distracted by irrelevant facts and descriptions. Ask yourself the following questions when you evaluate the level of detail in a paper:

• Is the rationale for performing the experiment clear (i.e., have you shown that the question you are addressing is important and interesting)?
• Are the materials and procedures used to generate the results described at a level of detail that would allow the experiment to be repeated?
• Is the rationale behind the choice of experimental methods clear? Will the reader understand why those particular methods are appropriate for answering the question your research is addressing?
• Will the reader be able to follow the chain of logic used to draw conclusions from the data?

Any information that enhances the reader’s understanding of the rationale, methodology, and logic should be included, but information in excess of this (or information that is redundant) will only confuse and distract the reader.

Whenever possible, use quantitative rather than qualitative descriptions. A phrase that uses definite quantities such as “development rate in the 30°C temperature treatment was ten percent faster than development rate in the 20°C temperature treatment” is much more precise than the more qualitative phrase “development rate was fastest in the higher temperature treatment.”

How can I make my writing clearer?

When you’re writing about complex ideas and concepts, it’s easy to get sucked into complex writing. Distilling complicated ideas into simple explanations is challenging, but you’ll need to acquire this valuable skill to be an effective communicator in the sciences. Complexities in language use and sentence structure are perhaps the most common issues specific to writing in the sciences.

Language use

When given a choice between a familiar and a technical or obscure term, the more familiar term is preferable if it doesn’t reduce precision. Here are a just a few examples of complex words and their simple alternatives:

In these examples, the term on the right conveys the same meaning as the word on the left but is more familiar and straightforward, and is often shorter as well.

There are some situations where the use of a technical or obscure term is justified. For example, in a paper comparing two different viral strains, the author might repeatedly use the word “enveloped” rather than the phrase “surrounded by a membrane.” The key word here is “repeatedly”: only choose the less familiar term if you’ll be using it more than once. If you choose to go with the technical term, however, make sure you clearly define it, as early in the paper as possible. You can use this same strategy to determine whether or not to use abbreviations, but again you must be careful to define the abbreviation early on.

Sentence structure

Science writing must be precise, and precision often requires a fine level of detail. Careful description of objects, forces, organisms, methodology, etc., can easily lead to complex sentences that express too many ideas without a break point. Here’s an example:

The osmoregulatory organ, which is located at the base of the third dorsal spine on the outer margin of the terminal papillae and functions by expelling excess sodium ions, activates only under hypertonic conditions.

Several things make this sentence complex. First, the action of the sentence (activates) is far removed from the subject (the osmoregulatory organ) so that the reader has to wait a long time to get the main idea of the sentence. Second, the verbs “functions,” “activates,” and “expelling” are somewhat redundant. Consider this revision:

Located on the outer margin of the terminal papillae at the base of the third dorsal spine, the osmoregulatory organ expels excess sodium ions under hypertonic conditions.

This sentence is slightly shorter, conveys the same information, and is much easier to follow. The subject and the action are now close together, and the redundant verbs have been eliminated. You may have noticed that even the simpler version of this sentence contains two prepositional phrases strung together (“on the outer margin of…” and “at the base of…”). Prepositional phrases themselves are not a problem; in fact, they are usually required to achieve an adequate level of detail in science writing. However, long strings of prepositional phrases can cause sentences to wander. Here’s an example of what not to do from Alley (1996):

“…to confirm the nature of electrical breakdown of nitrogen in uniform fields at relatively high pressures and interelectrode gaps that approach those obtained in engineering practice, prior to the determination of the processes that set the criterion for breakdown in the above-mentioned gases and mixtures in uniform and non-uniform fields of engineering significance.”

The use of eleven (yes, eleven!) prepositional phrases in this sentence is excessive, and renders the sentence nearly unintelligible. Judging when a string of prepositional phrases is too long is somewhat subjective, but as a general rule of thumb, a single prepositional phrase is always preferable, and anything more than two strung together can be problematic.

Nearly every form of scientific communication is space-limited. Grant proposals, journal articles, and abstracts all have word or page limits, so there’s a premium on concise writing. Furthermore, adding unnecessary words or phrases distracts rather than engages the reader. Avoid generic phrases that contribute no novel information. Common phrases such as “the fact that,” “it should be noted that,” and “it is interesting that” are cumbersome and unnecessary. Your reader will decide whether or not your paper is interesting based on the content. In any case, if information is not interesting or noteworthy it should probably be excluded.

How can I make my writing more objective?

The objective tone used in conventional scientific writing reflects the philosophy of the scientific method: if results are not repeatable, then they are not valid. In other words, your results will only be considered valid if any researcher performing the same experimental tests and analyses that you describe would be able to produce the same results. Thus, scientific writers try to adopt a tone that removes the focus from the researcher and puts it only on the research itself. Here are several stylistic conventions that enhance objectivity:

Passive voice

You may have been told at some point in your academic career that the use of the passive voice is almost always bad, except in the sciences. The passive voice is a sentence structure where the subject who performs the action is ambiguous (e.g., “you may have been told,” as seen in the first sentence of this paragraph; see our handout on passive voice and this 2-minute video on passive voice for a more complete discussion).

The rationale behind using the passive voice in scientific writing is that it enhances objectivity, taking the actor (i.e., the researcher) out of the action (i.e., the research). Unfortunately, the passive voice can also lead to awkward and confusing sentence structures and is generally considered less engaging (i.e., more boring) than the active voice. This is why most general style guides recommend only sparing use of the passive voice.

Currently, the active voice is preferred in most scientific fields, even when it necessitates the use of “I” or “we.” It’s perfectly reasonable (and more simple) to say “We performed a two-tailed t-test” rather than to say “a two-tailed t-test was performed,” or “in this paper we present results” rather than “results are presented in this paper.” Nearly every current edition of scientific style guides recommends the active voice, but different instructors (or journal editors) may have different opinions on this topic. If you are unsure, check with the instructor or editor who will review your paper to see whether or not to use the passive voice. If you choose to use the active voice with “I” or “we,” there are a few guidelines to follow:

• Avoid starting sentences with “I” or “we”: this pulls focus away from the scientific topic at hand.
• Avoid using “I” or “we” when you’re making a conjecture, whether it’s substantiated or not. Everything you say should follow from logic, not from personal bias or subjectivity. Never use any emotive words in conjunction with “I” or “we” (e.g., “I believe,” “we feel,” etc.).
• Never use “we” in a way that includes the reader (e.g., “here we see trait evolution in action”); the use of “we” in this context sets a condescending tone.

Acknowledging your limitations

Your conclusions should be directly supported by the data that you present. Avoid making sweeping conclusions that rest on assumptions that have not been substantiated by your or others’ research. For example, if you discover a correlation between fur thickness and basal metabolic rate in rats and mice you would not necessarily conclude that fur thickness and basal metabolic rate are correlated in all mammals. You might draw this conclusion, however, if you cited evidence that correlations between fur thickness and basal metabolic rate are also found in twenty other mammalian species. Assess the generality of the available data before you commit to an overly general conclusion.

Works consulted

We consulted these works while writing this handout. This is not a comprehensive list of resources on the handout’s topic, and we encourage you to do your own research to find additional publications. Please do not use this list as a model for the format of your own reference list, as it may not match the citation style you are using. For guidance on formatting citations, please see the UNC Libraries citation tutorial . We revise these tips periodically and welcome feedback.

Alley, Michael. 1996. The Craft of Scientific Writing , 3rd ed. New York: Springer.

Council of Science Editors. 2014. Scientific Style and Format: The CSE Manual for Authors, Editors, and Publishers , 8th ed. Chicago & London: University of Chicago Press.

Day, Robert A. 1994. How to Write and Publish a Scientific Paper , 4th ed. Phoenix: Oryx Press.

Day, Robert, and Nancy Sakaduski. 2011. Scientific English: A Guide for Scientists and Other Professionals , 3rd ed. Santa Barbara: Greenwood.

Gartland, John J. 1993. Medical Writing and Communicating . Frederick, MD: University Publishing Group.

Williams, Joseph M., and Joseph Bizup. 2016. Style: Ten Lessons in Clarity and Grace , 12th ed. New York: Pearson.

You may reproduce it for non-commercial use if you use the entire handout and attribute the source: The Writing Center, University of North Carolina at Chapel Hill

Make a Gift

Introduction

Why is writing important in science.

Writing is the most common form of scientific communication, yet scientists have a reputation for being poor writers. Why? One reason could be that writing is never really taught to scientists. Better writing will benefit your science career in several ways. Within the scientific community, improved communication leads to improved collaboration, easier access to cross-disciplinary knowledge, and faster, less painful training. Besides this, you will be able to communicate better not only with other researchers, but with the public, who funds your research. If scientists were better writers, the gap between the public and academy would shrink.

How can I use this resource?

The Scientific Writing Resource is online course material that teaches how to write effectively. The material is not about correctness (grammar, punctuation, etc), but about communicating what you intend to the reader . It can be used either in a science class or by individuals. It is intended for science students at the graduate level.

Instructors: Use the resource in a science class to dedicate a lecture or two to writing skills. For each lesson, there are 3 resources: principles (for a lecture), examples, and a worksheet (for assignment).

Individuals: You can go through the resource on your own. The lessons can be done in about 45 minutes each. Just read through the lessons and examples, and then try your hand on the worksheets.

Communicating Effectively

The goal of writing is communication. These lessons do not put forth absolute rules . If the intent of the writer is communicated, the writing was effective, regardless of rules kept or broken. Instead of rules, these lessons provide principles of communication and writing from the reader’s persective. The ideas presented here are derived from many sources, notably including the philosophy of Joseph Williams and George Gopen . If you know what readers expect , then you can fill that expectation . Approaching writing like this will help you improve your written communication.

The key to success in writing lies in smart revision . At first, treat these as principles of revision rather than principles of writing. As you practice, they will naturally become integrated as you write.

This resource focuses on scientific writing, not science writing. What’s the difference? Find out here: What is scientific writing?

Start with lesson 1

Writing the Scientific Paper

When you write about scientific topics to specialists in a particular scientific field, we call that scientific writing. (When you write to non-specialists about scientific topics, we call that science writing.)

The scientific paper has developed over the past three centuries into a tool to communicate the results of scientific inquiry. The main audience for scientific papers is extremely specialized. The purpose of these papers is twofold: to present information so that it is easy to retrieve, and to present enough information that the reader can duplicate the scientific study. A standard format with six main part helps readers to find expected information and analysis:

• Title--subject and what aspect of the subject was studied.
• Abstract--summary of paper: The main reason for the study, the primary results, the main conclusions
• Introduction-- why the study was undertaken
• Methods and Materials-- how the study was undertaken
• Results-- what was found
• Discussion-- why these results could be significant (what the reasons might be for the patterns found or not found)

There are many ways to approach the writing of a scientific paper, and no one way is right. Many people, however, find that drafting chunks in this order works best: Results, Discussion, Introduction, Materials & Methods, Abstract, and, finally, Title.

The title should be very limited and specific. Really, it should be a pithy summary of the article's main focus.

• "Renal disease susceptibility and hypertension are under independent genetic control in the fawn hooded rat"
• "Territory size in Lincoln's Sparrows ( Melospiza lincolnii )"
• "Replacement of deciduous first premolars and dental eruption in archaeocete whales"
• "The Radio-Frequency Single-Electron Transistor (RF-SET): A Fast and Ultrasensitive Electrometer"

This is a summary of your article. Generally between 50-100 words, it should state the goals, results, and the main conclusions of your study. You should list the parameters of your study (when and where was it conducted, if applicable; your sample size; the specific species, proteins, genes, etc., studied). Think of the process of writing the abstract as taking one or two sentences from each of your sections (an introductory sentence, a sentence stating the specific question addressed, a sentence listing your main techniques or procedures, two or three sentences describing your results, and one sentence describing your main conclusion).

Example One

Hypertension, diabetes and hyperlipidemia are risk factors for life-threatening complications such as end-stage renal disease, coronary artery disease and stroke. Why some patients develop complications is unclear, but only susceptibility genes may be involved. To test this notion, we studied crosses involving the fawn-hooded rat, an animal model of hypertension that develops chronic renal failure. Here, we report the localization of two genes, Rf-1 and Rf-2 , responsible for about half of the genetic variation in key indices of renal impairment. In addition, we localize a gene, Bpfh-1 , responsible for about 26% of the genetic variation in blood pressure. Rf-1 strongly affects the risk of renal impairment, but has no significant effect on blood pressure. Our results show that susceptibility to a complication of hypertension is under at least partially independent genetic control from susceptibility to hypertension itself.

Brown, Donna M, A.P. Provoost, M.J. Daly, E.S. Lander, & H.J. Jacob. 1996. "Renal disease susceptibility and hypertension are under indpendent genetic control in the faun-hooded rat." Nature Genetics , 12(1):44-51.

Example Two

We studied survival of 220 calves of radiocollared moose ( Alces alces ) from parturition to the end of July in southcentral Alaska from 1994 to 1997. Prior studies established that predation by brown bears ( Ursus arctos ) was the primary cause of mortality of moose calves in the region. Our objectives were to characterize vulnerability of moose calves to predation as influenced by age, date, snow depths, and previous reproductive success of the mother. We also tested the hypothesis that survival of twin moose calves was independent and identical to that of single calves. Survival of moose calves from parturition through July was 0.27 ± 0.03 SE, and their daily rate of mortality declined at a near constant rate with age in that period. Mean annual survival was 0.22 ± 0.03 SE. Previous winter's snow depths or survival of the mother's previous calf was not related to neonatal survival. Selection for early parturition was evidenced in the 4 years of study by a 6.3% increase in the hazard of death with each daily increase in parturition date. Although there was no significant difference in survival of twin and single moose calves, most twins that died disappeared together during the first 15 days after birth and independently thereafter, suggesting that predators usually killed both when encountered up to that age.

Key words: Alaska, Alces alces , calf survival, moose, Nelchina, parturition synchrony, predation

Testa, J.W., E.F. Becker, & G.R. Lee. 2000. "Temporal patterns in the survival of twin and single moose ( alces alces ) calves in southcentral Alaska." Journal of Mammalogy , 81(1):162-168.

Example Three

We monitored breeding phenology and population levels of Rana yavapaiensis by use of repeated egg mass censuses and visual encounter surveys at Agua Caliente Canyon near Tucson, Arizona, from 1994 to 1996. Adult counts fluctuated erratically within each year of the study but annual means remained similar. Juvenile counts peaked during the fall recruitment season and fell to near zero by early spring. Rana yavapaiensis deposited eggs in two distinct annual episodes, one in spring (March-May) and a much smaller one in fall (September-October). Larvae from the spring deposition period completed metamorphosis in earlv summer. Over the two years of study, 96.6% of egg masses successfully produced larvae. Egg masses were deposited during periods of predictable, moderate stream flow, but not during seasonal periods when flash flooding or drought were likely to affect eggs or larvae. Breeding phenology of Rana yavapaiensis is particularly well suited for life in desert streams with natural flow regimes which include frequent flash flooding and drought at predictable times. The exotic predators of R. yavapaiensis are less able to cope with fluctuating conditions. Unaltered stream flow regimes that allow natural fluctuations in stream discharge may provide refugia for this declining ranid frog from exotic predators by excluding those exotic species that are unable to cope with brief flash flooding and habitat drying.

Sartorius, Shawn S., and Philip C. Rosen. 2000. "Breeding phenology of the lowland leopard frog ( Rana yavepaiensis )." Southwestern Naturalist , 45(3): 267-273.

Introduction

The introduction is where you sketch out the background of your study, including why you have investigated the question that you have and how it relates to earlier research that has been done in the field. It may help to think of an introduction as a telescoping focus, where you begin with the broader context and gradually narrow to the specific problem addressed by the report. A typical (and very useful) construction of an introduction proceeds as follows:

"Echimyid rodents of the genus Proechimys (spiny rats) often are the most abundant and widespread lowland forest rodents throughout much of their range in the Neotropics (Eisenberg 1989). Recent studies suggested that these rodents play an important role in forest dynamics through their activities as seed predators and dispersers of seeds (Adler and Kestrell 1998; Asquith et al 1997; Forget 1991; Hoch and Adler 1997)." (Lambert and Adler, p. 70)

"Our laboratory has been involved in the analysis of the HLA class II genes and their association with autoimmune disorders such as insulin-dependent diabetes mellitus. As part of this work, the laboratory handles a large number of blood samples. In an effort to minimize the expense and urgency of transportation of frozen or liquid blood samples, we have designed a protocol that will preserve the integrity of lymphocyte DNA and enable the transport and storage of samples at ambient temperatures." (Torrance, MacLeod & Hache, p. 64)

"Despite the ubiquity and abundance of P. semispinosus , only two previous studies have assessed habitat use, with both showing a generalized habitat use. [brief summary of these studies]." (Lambert and Adler, p. 70)

"Although very good results have been obtained using polymerase chain reaction (PCR) amplification of DNA extracted from dried blood spots on filter paper (1,4,5,8,9), this preservation method yields limited amounts of DNA and is susceptible to contamination." (Torrance, MacLeod & Hache, p. 64)

"No attempt has been made to quantitatively describe microhabitat characteristics with which this species may be associated. Thus, specific structural features of secondary forests that may promote abundance of spiny rats remains unknown. Such information is essential to understand the role of spiny rats in Neotropical forests, particularly with regard to forest regeneration via interactions with seeds." (Lambert and Adler, p. 71)

"As an alternative, we have been investigating the use of lyophilization ("freeze-drying") of whole blood as a method to preserve sufficient amounts of genomic DNA to perform PCR and Southern Blot analysis." (Torrance, MacLeod & Hache, p. 64)

"We present an analysis of microhabitat use by P. semispinosus in tropical moist forests in central Panama." (Lambert and Adler, p. 71)

"In this report, we summarize our analysis of genomic DNA extracted from lyophilized whole blood." (Torrance, MacLeod & Hache, p. 64)

Methods and Materials

In this section you describe how you performed your study. You need to provide enough information here for the reader to duplicate your experiment. However, be reasonable about who the reader is. Assume that he or she is someone familiar with the basic practices of your field.

It's helpful to both writer and reader to organize this section chronologically: that is, describe each procedure in the order it was performed. For example, DNA-extraction, purification, amplification, assay, detection. Or, study area, study population, sampling technique, variables studied, analysis method.

Include in this section:

• study design: procedures should be listed and described, or the reader should be referred to papers that have already described the used procedure
• particular techniques used and why, if relevant
• modifications of any techniques; be sure to describe the modification
• specialized equipment, including brand-names
• temporal, spatial, and historical description of study area and studied population
• assumptions underlying the study
• statistical methods, including software programs

Example description of activity

Chromosomal DNA was denatured for the first cycle by incubating the slides in 70% deionized formamide; 2x standard saline citrate (SSC) at 70ºC for 2 min, followed by 70% ethanol at -20ºC and then 90% and 100% ethanol at room temperature, followed by air drying. (Rouwendal et al ., p. 79)

Example description of assumptions

We considered seeds left in the petri dish to be unharvested and those scattered singly on the surface of a tile to be scattered and also unharvested. We considered seeds in cheek pouches to be harvested but not cached, those stored in the nestbox to be larderhoarded, and those buried in caching sites within the arena to be scatterhoarded. (Krupa and Geluso, p. 99)

Examples of use of specialized equipment

• Oligonucleotide primers were prepared using the Applied Biosystems Model 318A (Foster City, CA) DNA Synthesizer according to the manufacturers' instructions. (Rouwendal et al ., p.78)
• We first visually reviewed the complete song sample of an individual using spectrograms produced on a Princeton Applied Research Real Time Spectrum Analyzer (model 4512). (Peters et al ., p. 937)

Example of use of a certain technique

Frogs were monitored using visual encounter transects (Crump and Scott, 1994). (Sartorius and Rosen, p. 269)

Example description of statistical analysis

We used Wilcox rank-sum tests for all comparisons of pre-experimental scores and for all comparisons of hue, saturation, and brightness scores between various groups of birds ... All P -values are two-tailed unless otherwise noted. (Brawner et al ., p. 955)

This section presents the facts--what was found in the course of this investigation. Detailed data--measurements, counts, percentages, patterns--usually appear in tables, figures, and graphs, and the text of the section draws attention to the key data and relationships among data. Three rules of thumb will help you with this section:

• present results clearly and logically
• avoid excess verbiage
• consider providing a one-sentence summary at the beginning of each paragraph if you think it will help your reader understand your data

Remember to use table and figures effectively. But don't expect these to stand alone.

Some examples of well-organized and easy-to-follow results:

• Size of the aquatic habitat at Agua Caliente Canyon varied dramatically throughout the year. The site contained three rockbound tinajas (bedrock pools) that did not dry during this study. During periods of high stream discharge seven more seasonal pools and intermittent stretches of riffle became available. Perennial and seasonal pool levels remained stable from late February through early May. Between mid-May and mid-July seasonal pools dried until they disappeared. Perennial pools shrank in surface area from a range of 30-60 m² to 3-5- M². (Sartorius and Rosen, Sept. 2000: 269)

Notice how the second sample points out what is important in the accompanying figure. It makes us aware of relationships that we may not have noticed quickly otherwise and that will be important to the discussion.

A similar test result is obtained with a primer derived from the human ß-satellite... This primer (AGTGCAGAGATATGTCACAATG-CCCC: Oligo 435) labels 6 sites in the PRINS reaction: the chromosomes 1, one pair of acrocentrics and, more weakly, the chromosomes 9 (Fig. 2a). After 10 cycles of PCR-IS, the number of sites labeled has doubled (Fig. 2b); after 20 cycles, the number of sites labeled is the same but the signals are stronger (Fig. 2c) (Rouwendal et al ., July 93:80).

Related Information: Use Tables and Figures Effectively

Do not repeat all of the information in the text that appears in a table, but do summarize it. For example, if you present a table of temperature measurements taken at various times, describe the general pattern of temperature change and refer to the table.

"The temperature of the solution increased rapidly at first, going from 50º to 80º in the first three minutes (Table 1)."

You don't want to list every single measurement in the text ("After one minute, the temperature had risen to 55º. After two minutes, it had risen to 58º," etc.). There is no hard and fast rule about when to report all measurements in the text and when to put the measurements in a table and refer to them, but use your common sense. Remember that readers have all that data in the accompanying tables and figures, so your task in this section is to highlight key data, changes, or relationships.

In this section you discuss your results. What aspect you choose to focus on depends on your results and on the main questions addressed by them. For example, if you were testing a new technique, you will want to discuss how useful this technique is: how well did it work, what are the benefits and drawbacks, etc. If you are presenting data that appear to refute or support earlier research, you will want to analyze both your own data and the earlier data--what conditions are different? how much difference is due to a change in the study design, and how much to a new property in the study subject? You may discuss the implication of your research--particularly if it has a direct bearing on a practical issue, such as conservation or public health.

This section centers on speculation . However, this does not free you to present wild and haphazard guesses. Focus your discussion around a particular question or hypothesis. Use subheadings to organize your thoughts, if necessary.

This section depends on a logical organization so readers can see the connection between your study question and your results. One typical approach is to make a list of all the ideas that you will discuss and to work out the logical relationships between them--what idea is most important? or, what point is most clearly made by your data? what ideas are subordinate to the main idea? what are the connections between ideas?

Achieving the Scientific Voice

Eight tips will help you match your style for most scientific publications.

• Develop a precise vocabulary: read the literature to become fluent, or at least familiar with, the sort of language that is standard to describe what you're trying to describe.
• Once you've labeled an activity, a condition, or a period of time, use that label consistently throughout the paper. Consistency is more important than creativity.
• Define your terms and your assumptions.
• Include all the information the reader needs to interpret your data.
• Remember, the key to all scientific discourse is that it be reproducible . Have you presented enough information clearly enough that the reader could reproduce your experiment, your research, or your investigation?
• When describing an activity, break it down into elements that can be described and labeled, and then present them in the order they occurred.
• When you use numbers, use them effectively. Don't present them so that they cause more work for the reader.
• Include details before conclusions, but only include those details you have been able to observe by the methods you have described. Do not include your feelings, attitudes, impressions, or opinions.
• Research your format and citations: do these match what have been used in current relevant journals?
• Run a spellcheck and proofread carefully. Read your paper out loud, and/ or have a friend look over it for misspelled words, missing words, etc.

Applying the Principles, Example 1

The following example needs more precise information. Look at the original and revised paragraphs to see how revising with these guidelines in mind can make the text clearer and more informative:

Before: Each male sang a definite number of songs while singing. They start with a whistle and then go from there. Each new song is always different, but made up an overall repertoire that was completed before starting over again. In 16 cases (84%), no new songs were sung after the first 20, even though we counted about 44 songs for each bird.

After: Each male used a discrete number of song types in his singing. Each song began with an introductory whistle, followed by a distinctive, complex series of fluty warbles (Fig. 1). Successive songs were always different, and five of the 19 males presented their entire song repertoire before repeating any of their song types (i.e., the first IO recorded songs revealed the entire repertoire of 10 song types). Each song type recurred in long sequences of singing, so that we could be confident that we had recorded the entire repertoire of commonly used songs by each male. For 16 of the 19 males, no new song types were encountered after the first 20 songs, even though we analyzed and average of 44 songs/male (range 30-59).

Applying the Principles, Example 2

In this set of examples, even a few changes in wording result in a more precise second version. Look at the original and revised paragraphs to see how revising with these guidelines in mind can make the text clearer and more informative:

Before: The study area was on Mt. Cain and Maquilla Peak in British Columbia, Canada. The study area is about 12,000 ha of coastal montane forest. The area is both managed and unmanaged and ranges from 600-1650m. The most common trees present are mountain hemlock ( Tsuga mertensiana ), western hemlock ( Tsuga heterophylla ), yellow cedar ( Chamaecyparis nootkatensis ), and amabilis fir ( Abies amabilis ).

After: The study took place on Mt. Cain and Maquilla Peak (50'1 3'N, 126'1 8'W), Vancouver Island, British Columbia. The study area encompassed 11,800 ha of coastal montane forest. The landscape consisted of managed and unmanaged stands of coastal montane forest, 600-1650 m in elevation. The dominant tree species included mountain hemlock ( Tsuga mertensiana ), western hemlock ( Tsuga heterophylla ), yellow cedar ( Chamaecyparis nootkatensis ), and amabilis fir ( Abies amabilis ).

Two Tips for Sentence Clarity

Although you will want to consider more detailed stylistic revisions as you become more comfortable with scientific writing, two tips can get you started:

First, the verb should follow the subject as soon as possible.

Really Hard to Read : "The smallest of the URF's (URFA6L), a 207-nucleotide (nt) reading frame overlapping out of phase the NH2- terminal portion of the adenosinetriphosphatase (ATPase) subunit 6 gene has been identified as the animal equivalent of the recently discovered yeast H+-ATPase subunit gene."

Less Hard to Read : "The smallest of the UR-F's is URFA6L, a 207-nucleotide (nt) reading frame overlapping out of phase the NH2-terminal portion of the adenosinetriphosphatase (ATPase) subunit 6 gene; it has been identified as the animal equivalent of the recently discovered yeast H+-ATPase subunit 8 gene."

Second, place familiar information first in a clause, a sentence, or a paragraph, and put the new and unfamiliar information later.

More confusing : The epidermis, the dermis, and the subcutaneous layer are the three layers of the skin. A layer of dead skin cells makes up the epidermis, which forms the body's shield against the world. Blood vessels, carrying nourishment, and nerve endings, which relay information about the outside world, are found in the dermis. Sweat glands and fat cells make up the third layer, the subcutaneous layer.

Less confusing : The skin consists of three layers: the epidermis, the dermis, and the subcutaneous layer. The epidermis is made up of dead skin cells, and forms a protective shield between the body and the world. The dermis contains the blood vessels and nerve endings that nourish the skin and make it receptive to outside stimuli. The subcutaneous layer contains the sweat glands and fat cells which perform other functions of the skin.

Bibliography

• Scientific Writing for Graduate Students . F. P. Woodford. Bethesda, MD: Council of Biology Editors, 1968. [A manual on the teaching of writing to graduate students--very clear and direct.]
• Scientific Style and Format . Council of Biology Editors. Cambridge: Cambridge University Press, 1994.
• "The science of scientific writing." George Gopen and Judith Swann. The American Scientist , Vol. 78, Nov.-Dec. 1990. Pp 550-558.
• "What's right about scientific writing." Alan Gross and Joseph Harmon. The Scientist , Dec. 6 1999. Pp. 20-21.
• "A Quick Fix for Figure Legends and Table Headings." Donald Kroodsma. The Auk , 117 (4): 1081-1083, 2000.

Wortman-Wunder, Emily, & Kate Kiefer. (1998). Writing the Scientific Paper. Writing@CSU . Colorado State University. https://writing.colostate.edu/resources/writing/guides/.

Writers' Center

Eastern Washington University

Scientific Writing

Thinking and writing like a scientist, writing clearly, citation help.

[ Back to resource home ]

[email protected] 509.359.2779

Cheney Campus   JFK Library Learning Commons

Spokane Campus Catalyst Building C451 and C452

Stay Connected! Instagram  Facebook

Helpful Links

For more information on writing your scientific papers, here are some good resources:

http://writingcenter.unc.edu/handouts/scientific-reports/

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1559667/

The Scientific Method

When writing or reading about science, it is useful to keep the scientific method in mind. The scientific method is used as a model to construct writing that can be shared with others in a logical and informative way. Any piece of scientific writing is informative and persuasive: informative because the author is telling the audience how he or she conducted their research and what new information they learned, and persuasive because science papers demonstrate how that new information was obtained and what conclusions can be drawn from the data collected. The format of most journal articles follows the steps of the scientific method, with Introduction, Methods, and Results sections at a minimum. 

Evidence and Argumentation

Science writing has a persuasive element to it. Researchers need to convince others that they have done their experiments properly and that they have answered their central research questions. Therefore, all science papers, even theoretical ones, make use of evidence to support their points. Remember that statistical measures, while extremely useful, are not the only source of evidence. Observations of even a single event can be useful in the right context. Remember to use logic to link your evidence and claims together!

Tips for writing a persuasive paper:  http://www.facstaff.bucknell.edu/awolaver/term1.htm

Relating evidence and ideas:  http://undsci.berkeley.edu/article/coreofscience_01

Clarity and Reader Expectations

Many people complain that scientific literature is difficult to understand because of the complicated language and the use of jargon. However, scientific literature can be difficult to understand even if one is familiar with the concepts being discussed.

To avoid confusing the reader, writers should focus on writing clearly and keeping the reader’s expectations in mind. In a scientific paper, it is important that most readers will agree with the information being presented. Writing in a clear and concise way helps the writer accomplish this. Use the paper as a story-telling medium. Concentrate on showing the reader that your experiments show definite conclusion, and how this contribution changes the state of knowledge in the field.

Gopen and Swan (1990) offer the following seven easy ways to make your writing more clear and to say what you want the reader to hear.

• Follow a grammatical subject as soon as possible with its verb.
• Place in the stress position the "new information" you want the reader to emphasize. 
• Place the person or thing whose "story" a sentence is telling at the beginning of the sentence, in the topic position. 
• Place appropriate "old information" (material already stated in the discourse) in the topic position for linkage backward and contextualization forward. 
• Articulate the action of every clause or sentence in its verb. 
• In general, provide context for your reader before asking that reader to consider anything new. 
• In general, try to ensure that the relative emphases of the substance coincide with the relative expectations for emphasis raised by the structure.

[Gopen, G. and Swan, J. 1990. The Science of Scientific Writing. American Scientist. Available here: https://www.americanscientist.org/blog/the-long-view/the-science-of-scientific-writing  ]

One major problem many students have when they start writing papers is using so called “running jumps.” This is the placement of unnecessary words at the beginning of a sentence. For example:

RUNNING JUMP: According to the researchers, the control group showed more change in chlorophyll production (Smith et. al., 2014).

NO RUNNING JUMP: The control group showed more change in chlorophyll production (Smith et. al., 2014).

We’ve already cited a study, so it’s clear that we are referring to researchers and their findings. So the first part of the sentence is unnecessary. 

Try to limit the number of ideas expressed in a single sentence. If a sentence seems like it is trying to say more than two things at once, split it into two sentences.  If a sentence is long and tangled and just doesn't make sense, don't try to perform "surgery" to fix the sentence. Instead, "kill" the sentence and start over, using short, direct sentences to express what you mean.

If you can make a noun phrase into a verb, do it! For example, made note of = noted, provided a similar opinion = agreed with, conducted an experiment = experimented, etc.

Avoid the Passive Voice Where You Can

If a sentence is written in the passive voice, the subject of the sentence (person/thing doing the action) does not come first; rather, the object of the sentence (person/thing not doing the action) is the first noun in the sentence. 

PASSIVE: Radiation was the mechanism by which the samples were sterilized.

MORE ACTIVE: The samples were sterilized using radiation.

ACTIVE: We sterilized the samples using radiation.

Professors (and scientific journals) have differing opinions on the use of passive voice. Some consider it unacceptable, but many are more lenient. And in fact, often it will make more sense to write in the passive voice in certain sections (i.e. Methods) and when you can't use first-person pronouns like "I." In any case, reducing overuse of passive voice in your writing makes it more concise and easier to understand.

Here's a useful link for clear scientific writing style: http://www.nature.com/scitable/topicpage/effective-writing-13815989

Posters are often used as an accompaniment to a talk or presentation, or as a substitute. You’ve probably seen posters hanging up around campus, showcasing students' research. The idea of a poster is to simplify a study and present it in a visual way, so it can be understood by a wide audience. The most important thing to remember when designing a poster (or completing any kind of published work) is to follow the guidelines given. If your instructor, or the conference you’re presenting at, wants a certain format, adhere to that format. These three rules are especially important to follow:

• Shorter is better: make sure that your poster does not contain too much text! Packing text onto the poster makes it difficult to read and understand.
• Bigger is better. No, this is not a contradiction of rule 1! Make sure your text is large enough to read, and readable against the background of the poster.
• Use images. The key aspect of a poster is that it is a visual medium. Include graphs, photos, and illustrations of your work.

Here are some excellent tips and templates for research posters: 

1.  http://colinpurrington.com/tips/academic/posterdesign

2. http://www.waspacegrant.org

3.  http://www.personal.psu.edu/drs18/postershow/ : Poster tips from Penn State

Most scientific citation styles are based on APA format. It’s totally okay to use a resource to look up how to format a paper in APA style! As you become more familiar with the format, you will become less reliant on these resources, but for now, here are some sites that may be useful .

Our APA guide

APA Style Official Website

• Last Updated: Jul 21, 2021 3:01 PM
• URL: https://research.ewu.edu/writers_center_sci_writing

• Utility Menu

de5f0c5840276572324fc6e2ece1a882

• How to Use This Site
• Core Competencies
• Writing Tips

Resources and guidelines to help you become a better writer.

Resources for writing.

Teaching Effective Written Communication Skills with AI

Writing for busy readers, preferred terms for select population groups and communities, science writes, social media guide, harvardwrites, using english for academic purposes, duke graduate school scientific writing resource, "american medical association manual of style: a guide for authors and editors", citation guidance from the nih, the science of scientific writing, the elements of style, “the write rules,” by katherine hartmann.

• Publishing Your Manuscript
• Plain Language

Write Like a Scientist

A Guide to Scientific Communication

What is scientific writing ?

Scientific writing is a technical form of writing that is designed to communicate scientific information to other scientists. Depending on the specific scientific genre—a journal article, a scientific poster, or a research proposal, for example—some aspects of the writing may change, such as its  purpose , audience , or organization . Many aspects of scientific writing, however, vary little across these writing genres. Important hallmarks of all scientific writing are summarized below. Genre-specific information is located  here  and under the “By Genre” tab at the top of the page.

What are some important hallmarks of professional scientific writing?

1. Its primary audience is other scientists. Because of its intended audience, student-oriented or general-audience details, definitions, and explanations — which are often necessary in lab manuals or reports — are not terribly useful. Explaining general-knowledge concepts or how routine procedures were performed actually tends to obstruct clarity, make the writing wordy, and detract from its professional tone.

2. It is concise and precise . A goal of scientific writing is to communicate scientific information clearly and concisely. Flowery, ambiguous, wordy, and redundant language run counter to the purpose of the writing.

3. It must be set within the context of other published work. Because science builds on and corrects itself over time, scientific writing must be situated in and  reference the findings of previous work . This context serves variously as motivation for new work being proposed or the paper being written, as points of departure or congruence for new findings and interpretations, and as evidence of the authors’ knowledge and expertise in the field.

All of the information under “The Essentials” tab is intended to help you to build your knowledge and skills as a scientific writer regardless of the scientific discipline you are studying or the specific assignment you might be working on. In addition to discussions of audience and purpose , professional conventions like conciseness and specificity, and how to find and use literature references appropriately, we also provide guidelines for how to organize your writing and how to avoid some common mechanical errors .

If you’re new to this site or to professional scientific writing, we recommend navigating the sub-sections under “The Essentials” tab in the order they’re provided. Once you’ve covered these essentials, you might find information on  genre-  or discipline-specific writing useful.

• Search This Site All UCSD Sites Faculty/Staff Search Term
• Contact & Directions
• Climate Statement
• Cognitive Behavioral Neuroscience
• Cognitive Psychology
• Developmental Psychology
• Social Psychology
• Adjunct Faculty
• Non-Senate Instructors
• Researchers
• Psychology Grads
• Affiliated Grads
• New and Prospective Students
• Honors Program
• Experiential Learning
• Programs & Events
• Psi Chi / Psychology Club
• Prospective PhD Students
• Current PhD Students
• Area Brown Bags
• Colloquium Series
• Anderson Distinguished Lecture Series
• Speaker Videos
• Undergraduate Program
• Academic and Writing Resources

Writing Research Papers

• Improving Scientific Writing

Writing is an art – an expression of skill, creativity, and in many cases, imagination.  Writing research papers is no exception.  Here we provide several examples of common ways to improve scientific writing. 

Please note that these examples refer to specific sections of research papers, but generally apply to any section of a research paper.

Topic Sentences and Connective Words and Phrases 

Most research papers present data or information that the reader may not be immediately familiar with.  For this reason, the importance of clarity and avoiding confusion cannot be overstated.  For instance, a major source of reader difficulty is the presentation of text that is not accompanied by any information that orients the reader as to its organization or focus.  That is illustrated in the following example (taken from a rough draft of a research paper):

Original Paragraph:  “ Mickelson (2013) found that sleep deprivation reduced procedural memory performance independently of procedural memory task ability.  According to Maxwell (2015), sleep deprivation reduces cognitive performance; however, its effects on prospective memory are unknown. In that study, they found that sleep deprivation can reduce participant’s performance during a finger-tapping task, which is caused by difficulties in staying aroused.”

To improve clarity, the use of topic sentences at the start of paragraphs can be especially helpful.  The topic sentence alerts the reader as to the organization and focus of the text that follows.  It also helps to ensure that each sentence follows the next in a logical, easy-to-read fashion.  That can be aided by connective words and phrases (for example, the words and phrases furthermore, moreover, additionally , in addition , for example, etc.).  Returning to the prior example, it can be modified as follows: 

Improved Paragraph  (key changes underlined) : “ Recent studies on the effects of sleep deprivation provide evidence that sleep can impact procedural learning and skills .  For example , Mickelson (2013) found that sleep deprivation reduced procedural memory performance independently of procedural memory task ability.  In addition , Maxwell (2015) found that sleep deprivation can reduce participant’s performance during a finger-tapping task, a finding that was attributed to difficulties in staying aroused.  Together, these and other studies suggest that inadequate sleep has a deleterious effect on a wide range of tasks involving motor skills .”

The improved paragraph begins with a topic sentence (“Recent studies…”) and the subsequent sentences include connective phrases (“For example,…”, “In addition,…”).  A concluding sentence (“Together, these…”) also summarizes the information that was presented in that paragraph.

Transitions Between Paragraphs and Ideas

As with topic sentences and connective words and phrases, the use of  transition sentences can help improve readers’ ability to advance from one paragraph to the next or one idea to the next.  This is especially important when two adjacent paragraphs discuss disparate topics.  In the absence of those transitions, as illustrated in the example paragraphs below, readers may miss the main points of the text.  They may also find the text unclear or even jarring to read in some cases.

Original Paragraphs:  “Bargh, Chen, and Burrows (1996) demonstrated perhaps the most famous of “social priming” effects.  In their study, participants completed a scrambled-sentence task and then left the laboratory.  When the task included words that reflected old age stereotypes, participants were recorded exiting the experimenter room more slowly.  Other researchers subsequently demonstrated social priming effects for other types of stereotypes and tasks.  For instance, Dijksterhuis and van Knippenberg (1998) found that activation of soccer hooliganism stereotypes reduced performance on general knowledge tests. Shanks et al. (2013) attempted to replicate several social priming effects across nine experiments.  A Bayesian analysis found evidence in favor of the null hypothesis in all cases.  Pashler, Coburn, and Harris (2012) attempted to replicate “spatial distance priming” effects (Williams & Bargh, 2008), in which plotting a pair of points affected participants’ reports of closeness with family members and food calorie estimates.  Across two experiments, the original findings did not replicate.”

In the above example, the lack of a transition sentence can lead readers to be surprised by the content of the second paragraph (or even miss the contrast with the first).  However, that problem can be easily remedied with such a sentence, for instance as follows:

Improved Paragraphs  (key changes underlined) : “…Other researchers subsequently demonstrated social priming effects for other types of stereotypes and tasks.  For instance, Dijksterhuis and van Knippenberg (1998) found that activation of soccer hooliganism stereotypes led to reduced performance on general knowledge tests. Although the “social priming” effects demonstrated by Bargh et al. (1996) and others seemed compelling to many observers, more recent work suggests that such effects do not reliably occur .  Specifically , Shanks et al. (2013) attempted to replicate several social priming effects across nine experiments.  A Bayesian analysis found evidence in favor of the null hypothesis in all cases.  Similarly , Pashler, Coburn, and Harris (2012) attempted to…”

As illustrated in the improved paragraphs, the addition of a transition sentence at the start of the second paragraph (“Although those…”) alerts the reader to the contrast between the two paragraphs.  Connective words and phrases also improve comprehension and flow.  As such, the reader immediately knows that the studies discussed in the first paragraph are disputed by those in the second.

Statistics vs. Prose  

When reporting statistical results, it is often less desirable to describe those results in a technically-dense, matter-of-fact manner (for example, describing the analyses in the exact order that they were performed and without focusing on the most meaningful results, as shown below 1 ). 

Original Paragraph:  “A two-way, 2x2 between-subjects ANOVA was performed on ratings of the vividness of childhood memories in which the independent variables were participant sex (male or female) and induced mood (happy, sad). There was no main effect for sex (F, p), but there was a main effect of mood, (F, p), and a mood by sex interaction (F, p).  Happy people had more vivid memories than sad people, overall. This effect was stronger for women than it was for men. As can be seen in the results from Tukey’s studentized range test reported in Table 1, the vividness of happy and sad female participants’ memories differed significantly, but the vividness of happy and sad male participants’ memories did not.”

An improved (that is, completely rewritten) version of the prior paragraph organizes the statistical analyses in a more easily understood fashion, highlights the most important results, clearly relates the findings to the study hypothesis, and prefaces the entire paragraph with an introductory sentence that orients the reader (with major improvements underlined as shown below 1 ).

Improved Paragraph  (key changes underlined) : “ Table 1 provides the vividness ratings for men and women who experienced happy or sad moods . The childhood memories of men and women did not differ in vividness, (F, p). The most striking finding, however , was that the usual tendency for happy people to report more vivid memories than people in sad moods (F, p) was stronger for women than men, as indicated by a significant sex by mood interaction, (F,p). This finding is consistent with hypothesis that mood has a more pronounced effect on the quality of childhood memories among women than men and was confirmed with the Tukey’s studentized range test reported in Table 1.”

A further type of improvement in that paragraph is the use of phrases which directly connects results statements to their statistical evidence (“as indicated by”, “was confirmed with”), and makes it clear to the reader how the authors’ conclusions are supported.  The improved paragraph also takes advantage of the fact that data are presented in a table by referring readers to that table at the outset.

Workshops and Downloadable Resources

• For in-person discussion of the process of writing research papers, please consider attending this department’s “Writing Research Papers” workshop (for dates and times, please check the undergraduate workshops calendar).

Downloadable Resources

• How to Write APA Style Research Papers (a comprehensive guide) [ PDF ]
• Tips for Writing APA Style Research Papers (a brief summary) [ PDF ]

Further Resources  

How-To Videos     

• Writing Research Paper Videos

Further Reading

• Strunk, W. (2007). The elements of style . Penguin [Book] Continuously published since at least 1920, this is one of the most influential guides to writing style in American English.  Recommended by Dr. Stephen Link.

External Resources

• How to Write Better Scientific Papers (Elsevier Publishing)
• Top Ten Writing Tips for Scientists
• Twenty-One Suggestions for Writing Good Scientific Papers
• Tutorial on Scientific Writing from Duke University [Tutorial]

1 Carver, L. (2014).  Writing the research paper [Workshop]. 

Prepared by s. c. pan for ucsd psychology.

Back to top

• Research Paper Structure
• Formatting Research Papers
• Using Databases and Finding References
• What Types of References Are Appropriate?
• Evaluating References and Taking Notes
• Citing References
• Writing a Literature Review
• Writing Process and Revising
• Academic Integrity and Avoiding Plagiarism
• Writing Research Papers Videos

ownership =~ tr/po/op/

Scientific Writing: Beyond Tips and Tricks

• communication
• best-worst-practices
• matching handout

My own experience in grad school: I had great mentors who taught me the practice of doing science, but when it came time to write nobody had any advice. So you just do it as best you know how.

Although my data remained the same, my results did not. The process of writing changed what I thought I had. And in fact, what I had, and what I published, and what I’m known for.

Because the process of writing causes you to learn and discover new things, it is actually a key part of science.

Think about the person who runs your research group. How many of them write about science once a week or more. Guess what, you just signed up for the job of writer. You are going to live and die, not only by the quality of the research you produce, but more by your ability to get it out fast enough and to get it out in the right place, in the hands of the right readers.

Science writing isn’t only hard because you’re writing about numbers/data or images (which economists and art historians do), but because you’re writing multi-author documents. Scientists are de-facto teachers of writing.

I’ve worked with the EPA where they routinely write 2200 page reports with 60+ co-authors.

You are responsible not only for some results that end up in the publication, but also writing up those results, and additionally for giving feedback to all the other people who are writing with you.

Unless your co-authors give you what you want and what you need without any feedback from you, you are in the same situation as a new writing instructor facing their first stack of papers.

Let me tell you a secret that I tell all my students. Your faculty are expert writers. They’re not expert teachers of writing, and they often don’t even have a language for capturing their expertise. They learned it the hard-earned way. They’ve been copy-edited to death, which is the only way we knew how to teach writing until about 30 years ago.

Because they don’t have a language about writing, they don’t have a way to pass on that knowledge. So we have to develop a systematic way to learn and teach these skills.

The research article is shooting a tiny target from far away, it’s a process of erasure.

Early in your graduate career you’re spending all your time figuring out what is needed to convince yourself your data is real, before you can then figure out what is needed to convince other experts your data is real.

Worry about writing from the reader’s perspective.

I actually would define graduate school […] you’re learning to become a professional reader of science.

Reading takes energy

Readers need to figure out “what does this thing mean?”

Writers produce text. Each reader is trying to interpret the text. They are looking at things they know about the writer (their purpose, setting, language, and field) and what the text means in that context. Because of this, there’s no way to control the way that all readers will interpret your text.

I call this the thermodynamics moment […]. Thermo gets recited as: you can’t win, you can only break even; you can’t break even, except at absolute zero; and you can’t get to absolute zero.

We have to think richly about readers.

Readers read for structure. After they’ve interpreted structure, then they go for substance. And reader energy is finite, so the energy spent on structure and substance is zero-sum, so energy spent to figure out the structure is not available to understand the substance.

Bad scientific writing is writing that requires a disproportionate amount of the reader’s energy simply to figure out what it’s all about

[16:48] Reader expectations

If you know something about what readers expect in given locations, you can make predictions about what they do with information in those locations. Research articles are the most structured document type within any academic field: if you want to see the experimental details, you have to look in the Method section.

[18:09] Emphasis

How can we ensure readers emphasize some information as more important than some other information?

Readers emphasize information in the main clause/independent clause over the subordinate clauses. Or information that comes at a point of closure: end of sentence, end of paragraph, end of section, end of paper. Or, something that the paper spends more length/time on. Or, repetition. Semantics/words can also tell you what’s most important, the writer can spell out “this is the most important result”

Traditional writing instruction says that writing is 80% word choice (finding le mot juste ), and 20% structure; but the assertion is that readers take 80% of their instructions from structure.

[24:30] First example: the facts do not speak for themselves

4 sentences present the same 2 facts in different sentence structures: Fred is a nice guy (Compliment), Fred beats his dog (Concerning). Our question about these is not what do you think about Fred, but what do you think the author of this sentence wants you to think about Fred?

What does the audience think was the author’s sentiment about Fred?

• A) Although Compliment, Concerning: >90% Negative
• B) Although Concerning, Compliment: >90% Positive
• C) Compliment but Concerning: 60-70% Negative
• D) Concerning but Compliment: 60-70% Positive

Congratulations, Carnegie Mellon, you are a normal audience. I’ve done this about 100 times, most readers respond as you did.

If you were in the minority on any of these positions, don’t worry, you are part of the normal minority. This is precisely because we’re talking about not the reader , there is no such thing as the reader , it’s readers .

Notice that you changed your minds, but the data did not change between these sentences.

We have two clauses, both the same length. And here we see the difference main vs subclause and end vs not end.

• “Although Compliment, Concerning” gives a negative at the end in the main clause, so it makes sense that it was viewed as negative
• “Although Concerning, Compliment” gives the positive at the end in the main clause, so it makes sense that the sentiment was perceived as positive

The other two put two signals of emphasis into conflict.

• What happens here is not that individuals seesaw between a negative and positive sentiment, but the crowd partitions into negative and positive
• These results might be explained by end-placement outweighing main clause.
• Or maybe when structures conflict readers are more prone to weigh the facts according to their own preferences
• It’s a good thing we controlled for this with a 4th example
• So it does seem like structure is causing this, and end position imparts more emphasis than clausal dominance.

If you were ever in the minority here, not only should you feel okay about it, but what you’ve just discovered is that most people don’t read like you. You might need to deliberately, consciously force these cues to align for the reader even when they already feel okay to you.

[34:45] You can use this to induce ambivalence

Sometimes you want to do this.

• it comes off as a pro-Fred apology: by the time we’re done with the whole sentence, lots of people have forgotten this
• Length does seem to have made a difference
• E communicates that the reader is pro-Fred but is worried about his dog problem.

Length seems to intensify whatever the existing structural elements have communicated.

Suppose you’re a congressperson, you just voted on the controversial MRX plan. You got your arm twisted to vote for it (by your party or whatever), but your constituents are ambivalent about it. You want to avoid alienating constituents on either side so you want to communicate ambivalence.

Or, imagine you voted for it, your district has no problems with it, but it’s 2 years before your re-election, and the plan has risks that might be realized when your election campaign is in full swing. If it blows up, you want to be on the record with reservations. So you wish to name the risks but de-emphasize them.

Suppose the MRX plan is the single issue your constituency sent you to Washington to pass. What do you do? Emphasize pro-MRX in as many ways possible: clausally, positionally, and in length, and bury the caveat.

How many of you got the writing advice to omit needless words? Substantially, needless words are useless; structurally they are ballast that keep the ship afloat.

I’m not saying that structure is 100% and words are zero. Okay? If you have a strong enough word, it can carry the day. For instance, try this sentence: “although Fred’s a nice guy, he commits genocide.”

What happens in this situation is not that the structure is erased in the reader’s mind. The initial pro-Fred statement is so overshadowed by the reader’s negative judgment about Fred that the reader concludes that not only is Fred a bad guy but the writer is a horrible apologist.

Subtly different, “although Fred commits genocide, he’s a nice guy” starts with the huge negative, so the reader who gets to the end will be wondering whether the sentence meant sarcastically/satirically.

Readers read linearly through time and are constructing meaning on the fly based on the substance and the structure.

[46:30] NIH handout

At NIH, they constantly have to tell the people upstairs (who make funding decisions) and principal investigators (who wrote the grant) what the judgment was on the grant.

Structure is invisible rhetoric. We don’t notice it, and so it does things to us and we’re not conscious of it.

As a PI, how would you gauge your likelihood and expected funding based on a given NIH summary?

• this is pretty good
• The main clause is positive, the end is positive
• the bad news is buried between the subject and verb, a place where readers don’t pay a lot of attention to
• the words are the same but the structure spells out death
• this is such a puzzle
• There are two pieces of good news and 1 piece of bad news. The bad news is muted (“somewhat”), and the good news is structured for emphasis
• But the piece of good news at the end is about the investigator, not the application.
• And as we saw before, end position carries slightly more emphasis than main clause
• By wasting the strongest structural cue for emphasis, we are damning with faint praise
• It’s not good, but it can be salvaged by the coming sentences.
• With the bad news emphasized, but counterbalanced by some good news, it’s still ambivalent but a bit worse
• This is pretty good assessment to receive
• As a PI, you know you just need to clean up the flaw and then it’s green lights all the way.
• “truly” adds length, and serves to minimize the flaw

[54:36] Exercise: rewrite statements to force positive or negative sentiment

I don’t want you to invent new words. You’ll have to change the words as you move things around, so this is not you can’t change the words at all. But, context controls meaning, so you can always change how we interpret this by telling us all these other things, and I want you to do that.

I want you to discover what structure does for you. Because one of the things that will come out of this experience is that we tend to recognize structural problems in our gut, and then try to solve them by throwing words at them. And words cannot undo a problem of structure.

[1:00:40] Discussion of the exercise

Think about how we write. Which clause are you likely to write first? The main clause, right? And the positive, or the clause you want to emphasize most. “This is really good, although there is this problem.”, “This is really horrible, although I guess we could consider it this way.” You hear how we’re gonna spontaneously probably produce the ambivalent structure when we think we’re communicating a strong message one way or the other.

I have one rule for talking about writing: No rules. We’re doing epistemology, we’re making truth claims, we’re trying to tell the stories of things that cannot speak for themselves. And that I think is more important than making your English teacher, whoever it was, happy with your sentence. So, I don’t want to turn it into rules. The rules will fail. They will fail. If we had more time I could give you a set of expectations that contradict one another and you’d say “what do I do to get it right?”. It’s right if it’s delivering the message you want to send to the majority of your readers.

[1:19:30] Context controls meaning

We’ve looked at sentences in isolation, which is good for workshopping sentence structure, but a very small part of the picture. The last example was “This is a very short application with little experimental detail by two new investigators who are very well trained in mouse genetics.”, and we worked on ways to rephrase it to come across much more positive or much more negative. But knowing the context, the existing sentence is already crystal clear. The subject of the grant application was mouse genetics, so it seems more positive. But the application was on human genetics.

It’s sort of flat, yeah. It’s only if you’re the PI and you know that the project’s on human genetics and the emphasis is on mouse genetics. At that moment, you know what happened. And I think there’s a covert message in here, that says “Hey guys. Guys, we know who you are. You come from really good labs. We expect good things of you, and we know you’re smart. But you’ve done the classic newbie’s mistake. You have written a proposal out-of-field, without sufficient experimental detail to justify it.” And rather than excoriating them, because they think “You know, you’re really smart, you’re gonna get it, as soon as you show this to anybody (and clearly you didn’t show it to anyone until you sent it to us).” But rather than say “Oh you idiots”, they say “Hey, here’s this thing, here’s the score, and now you know what the problem is.” What’s the implicit instruction? “Elaborate. You’ve got room. Resubmit.” And this is implicit, not a guarantee, but almost a promise here that says, when you resubmit what will study section write? “This application is dramatically improved over the first submission and these very talented investigators […]”

You might feel uncomfortable here that I’m asking you to make up stories about your data. You’re tied to your data, and I’m going to trust that you’re faithful to your data. But your data don’t speak. They cannot speak.

You have to go figure out, what do you believe you are allowed to claim on the basis of these data. That’s number one. You must choose. There is no neutral sentence. Every sentence sends instructions, and if the instructions are “I have no instructions” then the message is “and I’m incompetent, don’t listen to me.”

You must make up your mind what you think. You must communicate that to yourself and others clearly. And then, […] you have to listen to what your readers say. There will be arguments, science is contentious, it’s difficult. We see the same thing and we interpret it differently. I want those arguments to be about the substance, and we can’t argue about the substance if the structure fails to transmit that substance clearly to us. That’s your job. You need to decide what your facts are, and then speak for them. That’s whether you’re talking to your other scientists, but especially when you are talking to everyone else.

Listen to readers, they are always right. The reasons they give you for what they think, don’t listen to those. [But permute your structure to emphasize your goals, and resubmit, and see if they now understand/agree more with your point]

Solar Eclipse 2024 Reading Comprehension Passages, Writing & Research Activities

What educators are saying

Description.

Get your students excited and informed about the Solar Eclipse 2024 with ready-to-use reading comprehension passages with questions , linked articles/videos, creative writing prompts, and other engaging activities! 

Students love learning about the world around them! This solar eclipse mini unit creates an opportunity for students to learn about a unique science experience. You can complete the solar eclipse elements in one day or spread them out longer. 

Give students an opportunity to get excited about the solar eclipse by using it as a teachable experience. Students can get immersed in the solar eclipse through reading, writing and drawing. Students will love learning about the solar eclipse! 

Here's what you'll get:

• 3 Reading passages and questions -Answer Keys
• Directed Drawing with 3 writing prompts -Printable -Slides to display
• Selfie Writing 
• Video of the Day -3 Turn and Talk prompts -3 Writing prompts -Slides to display
• Mini Research Google Slides -1 Article with questions & Answer key -1 Video with questions & Answer Key

Your students will enjoy learning all about the solar eclipse. This is a unique opportunity that can get students excited and engaged about science with little work for you. Students can work in small groups, partners or even independently. The students will be engaged while using real-life applications.

Prep is quick and easy... Just assign the research project to your students, print or display!

___________________________________

You may also like…

→ Weather Research Project

→ Hero Research Project

→ Research Bundle

Also, Click HERE   to follow me, to be notified when new products are uploaded and on sale! New products are always discounted the first 24 hours they are uploaded! 

Copyright © Teach Create Motivate.

Permission to copy for single classroom use only.

Please purchase additional licenses if you intend to share this product.

Questions & Answers

Teach create motivate.

• We're hiring
• Help & FAQ
• Privacy policy
• Student privacy
• Terms of service
• Tell us what you think

Scientists Say "Grandma Brain" Isn't Just in Your Head

MRIs reveal that grandmothers may feel more emotional empathy toward their grandkids than they did with their own kids.

My son-in-law recently introduced me to his friend the following way: “This is Kate. She’s an interesting person and a good conversationalist … unless Anne is in the room. In that case, don’t even try.”

Anne is my toddler-aged granddaughter, and I am guilty as charged. Something happens to me when I’m with her. I’m enchanted. Bewitched. Besotted.

It turns out that this isn’t just love, though there’s plenty of that. I’m experiencing “Grandma Brain.” That’s not a technical term, but there is a scientific explanation for what’s going on. Neurological research has shown that something unique happens to our brains when we see our grandchildren — and it may explain our intense devotion and delight.

“What really jumps out in the data is the activation in areas of the brain associated with emotional empathy,” Rilling says. “That suggests that grandmothers are geared towards feeling what their grandchildren are feeling when they interact with them. If their grandchild is smiling, they’re feeling the child’s joy. And if their grandchild is crying, they’re feeling the child’s pain and distress.”

Retired teacher Janet Meisel, who has three grandsons, wrote an essay on Medium with a title that’s a near-perfect illustration of the heightened emotional empathy that Rilling’s study revealed. Titled “Are Grandparents Supposed To Feel This Much Love?” her subtitle reads, “Sometimes it feels like my heart will burst with joy and sometimes with pain.”

Reflecting on the difference between being a mom and a grandma, Meisel writes, “I had given my children so much of myself, but this feeling was different. It was love on steroids.”

What makes the new research particularly interesting is that the grandmothers’ brains did not light up in the same areas when women looked at pictures of their own children. When photos of their offspring were shown to the same women, a different area was activated, one associated with cognitive empathy. With cognitive empathy, a person can understand what another is feeling and why. But with emotional empathy, a person experiences what someone else is feeling.

That’s no surprise to Nancy Claus, a Connecticut grandmother of two. “It’s like a melting,” she says, sighing. “Isabella’s just at this magical stage where you hold her, she looks into your eyes, locks on you, and just bursts into this smile. She’s just wiggling with happiness all over and then I feel so happy, just flooded with this warm, delightful feeling.”

Grandparents tend to joke, “If I’d known how great it would be to be a grandparent, I’d have done it before I had kids.” This sentiment has less to do with brain activity than it does with the relationship. (“I can enjoy the kid and then hand them back.”) Grandparents, of course, have less responsibility than parents. Combine that with heightened emotional empathy swirling around in their brains and a “don’t sweat the small stuff” sensibility of many older folks, and it makes for a joyful ride.

“I could have a terrible day or a wonderful day, but as soon as I walk into the room with my grandson or my granddaughter, anything that happened before just dissipates,” says Melanie Schaffran, a New York grandmother. “I’m in complete rapture. Not to say I’m not aware of the mess being created, but it doesn’t matter.”

For mothers and fathers, mess does matter. Parents are often up to their ears in work stress, financial stress, kid stress, eldercare stress and everything else pulling on them at the stage of life when kids are young.

“I don’t even remember that life,” says Marti Gallardo, a Texas grandmother of three, with two more on the way. “We were just so busy.” Gallardo, now retired from her career in advertising sales, chatted while covered in “green apple peel spit.” She was watching her 14-month-old granddaughter, Clare, who only likes the inside of apples. As a grandma — and a woman who was no longer has to rush to work looking professional nor worry about if the baby is eating the right foods — Gallardo was relaxed.

Rilling’s original research was conducted in 2021. His study included placing 50 grandmothers in MRI machines and observing their reactions to photos of grandchildren who ranged from ages three to 12. The women were also shown photos of their own biological children as adults, as well as pictures of unknown people of all ages.

The data is still being analyzed. Currently, Rilling is studying saliva samples to see whether grandmothers have more oxytocin — nicknamed “the love hormone” for the good feeling it produces — compared to women of the same age who are not grandmothers. His team is also exploring if being a grandmother affects the rate at which the brain ages, with the hypothesis that it slows the process. He also hopes to conduct longitudinal studies by taking brain images and studying hormones before and after women become grandmothers. Rilling also plans to study the brains of grandfathers.

The biology of grandparents is “unexplored territory,” Rilling says. “A lot of research on older people’s brains is in the context of pathology and degeneration. Here we’re looking at healthy older brains and what they may have been designed for.” He believes his is the first study to examine grandmothers’ brain activity.

That said, other non-neurological studies have made the connection between being an active grandparent and better health. AARP conducted a comprehensive survey of grandparents and in its summary refers to grandkids as “the elixir of life.” Benefits include a more physically active lifestyle, more sociability and improved mental well-being.

The Berlin Aging Study , which tracked health outcomes of more than 500 people ages 70 and older, found that grandparents who helped care for their grandchildren had lower mortality rates over a 20-year period than those who did not. Plenty of research has also documented the benefits to grandchildren of having grandparents in their lives.

Why have grandmother’s brains evolved as they have? As an anthropologist as well as a psychologist, Rilling speculates that close grandmaternal bonds make it easier for parents to reproduce and continue the species. Or to put it in everyday terms, if parents have some help with the first baby , they’re more likely to go for number two. Or three.

That may explain another facet of intense grandmother love. The recognition of past, present and future in a grandchild’s face. “I see little bits of my mother in Olivia already,” Claus says of her 2-year-old granddaughter. “I don’t want to burst into singing ‘Circle of Life’ here, but it’s seeing your genes passing on and moving to the future. Seeing the next generation, the continuity, just brings a contentment.”

@media(max-width: 64rem){.css-o9j0dn:before{margin-bottom:0.5rem;margin-right:0.625rem;color:#ffffff;width:1.25rem;bottom:-0.2rem;height:1.25rem;content:'_';display:inline-block;position:relative;line-height:1;background-repeat:no-repeat;}.loaded .css-o9j0dn:before{background-image:url(/_assets/design-tokens/goodhousekeeping/static/images/Clover.5c7a1a0.svg);}}@media(min-width: 48rem){.loaded .css-o9j0dn:before{background-image:url(/_assets/design-tokens/goodhousekeeping/static/images/Clover.5c7a1a0.svg);}} Parenting Tips & Advice

Heartwarming Quotes for Kids

The Easter Bunny's Magic Is Real

Here's What You Really Need When You Give Birth

110 Belíssimo Italian Baby Names to Consider

Why Is Summer Child Care So Hard to Find?

The 10 Most Popular Baby Names from the 1980s

105 Beautiful Irish Girl Names for Your New Baby

125 Unique Irish Boy Names for Your Little Guy

The Best Ways for Parents to Teach Teens to Drive

"Bird Nesting" for Divorced Parents Is on the Rise

70 Creative Baby Shower Ideas

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 26 March 2024

Predicting and improving complex beer flavor through machine learning

• Michiel Schreurs   ORCID: orcid.org/0000-0002-9449-5619 1 , 2 , 3   na1 ,
• Supinya Piampongsant 1 , 2 , 3   na1 ,
• Miguel Roncoroni   ORCID: orcid.org/0000-0001-7461-1427 1 , 2 , 3   na1 ,
• Lloyd Cool   ORCID: orcid.org/0000-0001-9936-3124 1 , 2 , 3 , 4 ,
• Beatriz Herrera-Malaver   ORCID: orcid.org/0000-0002-5096-9974 1 , 2 , 3 ,
• Christophe Vanderaa   ORCID: orcid.org/0000-0001-7443-5427 4 ,
• Florian A. Theßeling 1 , 2 , 3 ,
• Łukasz Kreft   ORCID: orcid.org/0000-0001-7620-4657 5 ,
• Alexander Botzki   ORCID: orcid.org/0000-0001-6691-4233 5 ,
• Philippe Malcorps 6 ,
• Luk Daenen 6 ,
• Tom Wenseleers   ORCID: orcid.org/0000-0002-1434-861X 4 &
• Kevin J. Verstrepen   ORCID: orcid.org/0000-0002-3077-6219 1 , 2 , 3  

Nature Communications volume  15 , Article number:  2368 ( 2024 ) Cite this article

41k Accesses

778 Altmetric

Metrics details

• Chemical engineering
• Gas chromatography
• Machine learning
• Metabolomics
• Taste receptors

The perception and appreciation of food flavor depends on many interacting chemical compounds and external factors, and therefore proves challenging to understand and predict. Here, we combine extensive chemical and sensory analyses of 250 different beers to train machine learning models that allow predicting flavor and consumer appreciation. For each beer, we measure over 200 chemical properties, perform quantitative descriptive sensory analysis with a trained tasting panel and map data from over 180,000 consumer reviews to train 10 different machine learning models. The best-performing algorithm, Gradient Boosting, yields models that significantly outperform predictions based on conventional statistics and accurately predict complex food features and consumer appreciation from chemical profiles. Model dissection allows identifying specific and unexpected compounds as drivers of beer flavor and appreciation. Adding these compounds results in variants of commercial alcoholic and non-alcoholic beers with improved consumer appreciation. Together, our study reveals how big data and machine learning uncover complex links between food chemistry, flavor and consumer perception, and lays the foundation to develop novel, tailored foods with superior flavors.

Similar content being viewed by others

BitterSweet: Building machine learning models for predicting the bitter and sweet taste of small molecules

Rudraksh Tuwani, Somin Wadhwa & Ganesh Bagler

Sensory lexicon and aroma volatiles analysis of brewing malt

Xiaoxia Su, Miao Yu, … Tianyi Du

Predicting odor from molecular structure: a multi-label classification approach

Kushagra Saini & Venkatnarayan Ramanathan

Introduction

Predicting and understanding food perception and appreciation is one of the major challenges in food science. Accurate modeling of food flavor and appreciation could yield important opportunities for both producers and consumers, including quality control, product fingerprinting, counterfeit detection, spoilage detection, and the development of new products and product combinations (food pairing) 1 , 2 , 3 , 4 , 5 , 6 . Accurate models for flavor and consumer appreciation would contribute greatly to our scientific understanding of how humans perceive and appreciate flavor. Moreover, accurate predictive models would also facilitate and standardize existing food assessment methods and could supplement or replace assessments by trained and consumer tasting panels, which are variable, expensive and time-consuming 7 , 8 , 9 . Lastly, apart from providing objective, quantitative, accurate and contextual information that can help producers, models can also guide consumers in understanding their personal preferences 10 .

Despite the myriad of applications, predicting food flavor and appreciation from its chemical properties remains a largely elusive goal in sensory science, especially for complex food and beverages 11 , 12 . A key obstacle is the immense number of flavor-active chemicals underlying food flavor. Flavor compounds can vary widely in chemical structure and concentration, making them technically challenging and labor-intensive to quantify, even in the face of innovations in metabolomics, such as non-targeted metabolic fingerprinting 13 , 14 . Moreover, sensory analysis is perhaps even more complicated. Flavor perception is highly complex, resulting from hundreds of different molecules interacting at the physiochemical and sensorial level. Sensory perception is often non-linear, characterized by complex and concentration-dependent synergistic and antagonistic effects 15 , 16 , 17 , 18 , 19 , 20 , 21 that are further convoluted by the genetics, environment, culture and psychology of consumers 22 , 23 , 24 . Perceived flavor is therefore difficult to measure, with problems of sensitivity, accuracy, and reproducibility that can only be resolved by gathering sufficiently large datasets 25 . Trained tasting panels are considered the prime source of quality sensory data, but require meticulous training, are low throughput and high cost. Public databases containing consumer reviews of food products could provide a valuable alternative, especially for studying appreciation scores, which do not require formal training 25 . Public databases offer the advantage of amassing large amounts of data, increasing the statistical power to identify potential drivers of appreciation. However, public datasets suffer from biases, including a bias in the volunteers that contribute to the database, as well as confounding factors such as price, cult status and psychological conformity towards previous ratings of the product.

Classical multivariate statistics and machine learning methods have been used to predict flavor of specific compounds by, for example, linking structural properties of a compound to its potential biological activities or linking concentrations of specific compounds to sensory profiles 1 , 26 . Importantly, most previous studies focused on predicting organoleptic properties of single compounds (often based on their chemical structure) 27 , 28 , 29 , 30 , 31 , 32 , 33 , thus ignoring the fact that these compounds are present in a complex matrix in food or beverages and excluding complex interactions between compounds. Moreover, the classical statistics commonly used in sensory science 34 , 35 , 36 , 37 , 38 , 39 require a large sample size and sufficient variance amongst predictors to create accurate models. They are not fit for studying an extensive set of hundreds of interacting flavor compounds, since they are sensitive to outliers, have a high tendency to overfit and are less suited for non-linear and discontinuous relationships 40 .

In this study, we combine extensive chemical analyses and sensory data of a set of different commercial beers with machine learning approaches to develop models that predict taste, smell, mouthfeel and appreciation from compound concentrations. Beer is particularly suited to model the relationship between chemistry, flavor and appreciation. First, beer is a complex product, consisting of thousands of flavor compounds that partake in complex sensory interactions 41 , 42 , 43 . This chemical diversity arises from the raw materials (malt, yeast, hops, water and spices) and biochemical conversions during the brewing process (kilning, mashing, boiling, fermentation, maturation and aging) 44 , 45 . Second, the advent of the internet saw beer consumers embrace online review platforms, such as RateBeer (ZX Ventures, Anheuser-Busch InBev SA/NV) and BeerAdvocate (Next Glass, inc.). In this way, the beer community provides massive data sets of beer flavor and appreciation scores, creating extraordinarily large sensory databases to complement the analyses of our professional sensory panel. Specifically, we characterize over 200 chemical properties of 250 commercial beers, spread across 22 beer styles, and link these to the descriptive sensory profiling data of a 16-person in-house trained tasting panel and data acquired from over 180,000 public consumer reviews. These unique and extensive datasets enable us to train a suite of machine learning models to predict flavor and appreciation from a beer’s chemical profile. Dissection of the best-performing models allows us to pinpoint specific compounds as potential drivers of beer flavor and appreciation. Follow-up experiments confirm the importance of these compounds and ultimately allow us to significantly improve the flavor and appreciation of selected commercial beers. Together, our study represents a significant step towards understanding complex flavors and reinforces the value of machine learning to develop and refine complex foods. In this way, it represents a stepping stone for further computer-aided food engineering applications 46 .

To generate a comprehensive dataset on beer flavor, we selected 250 commercial Belgian beers across 22 different beer styles (Supplementary Fig.  S1 ). Beers with ≤ 4.2% alcohol by volume (ABV) were classified as non-alcoholic and low-alcoholic. Blonds and Tripels constitute a significant portion of the dataset (12.4% and 11.2%, respectively) reflecting their presence on the Belgian beer market and the heterogeneity of beers within these styles. By contrast, lager beers are less diverse and dominated by a handful of brands. Rare styles such as Brut or Faro make up only a small fraction of the dataset (2% and 1%, respectively) because fewer of these beers are produced and because they are dominated by distinct characteristics in terms of flavor and chemical composition.

Extensive analysis identifies relationships between chemical compounds in beer

For each beer, we measured 226 different chemical properties, including common brewing parameters such as alcohol content, iso-alpha acids, pH, sugar concentration 47 , and over 200 flavor compounds (Methods, Supplementary Table  S1 ). A large portion (37.2%) are terpenoids arising from hopping, responsible for herbal and fruity flavors 16 , 48 . A second major category are yeast metabolites, such as esters and alcohols, that result in fruity and solvent notes 48 , 49 , 50 . Other measured compounds are primarily derived from malt, or other microbes such as non- Saccharomyces yeasts and bacteria (‘wild flora’). Compounds that arise from spices or staling are labeled under ‘Others’. Five attributes (caloric value, total acids and total ester, hop aroma and sulfur compounds) are calculated from multiple individually measured compounds.

As a first step in identifying relationships between chemical properties, we determined correlations between the concentrations of the compounds (Fig.  1 , upper panel, Supplementary Data  1 and 2 , and Supplementary Fig.  S2 . For the sake of clarity, only a subset of the measured compounds is shown in Fig.  1 ). Compounds of the same origin typically show a positive correlation, while absence of correlation hints at parameters varying independently. For example, the hop aroma compounds citronellol, and alpha-terpineol show moderate correlations with each other (Spearman’s rho=0.39 and 0.57), but not with the bittering hop component iso-alpha acids (Spearman’s rho=0.16 and −0.07). This illustrates how brewers can independently modify hop aroma and bitterness by selecting hop varieties and dosage time. If hops are added early in the boiling phase, chemical conversions increase bitterness while aromas evaporate, conversely, late addition of hops preserves aroma but limits bitterness 51 . Similarly, hop-derived iso-alpha acids show a strong anti-correlation with lactic acid and acetic acid, likely reflecting growth inhibition of lactic acid and acetic acid bacteria, or the consequent use of fewer hops in sour beer styles, such as West Flanders ales and Fruit beers, that rely on these bacteria for their distinct flavors 52 . Finally, yeast-derived esters (ethyl acetate, ethyl decanoate, ethyl hexanoate, ethyl octanoate) and alcohols (ethanol, isoamyl alcohol, isobutanol, and glycerol), correlate with Spearman coefficients above 0.5, suggesting that these secondary metabolites are correlated with the yeast genetic background and/or fermentation parameters and may be difficult to influence individually, although the choice of yeast strain may offer some control 53 .

Spearman rank correlations are shown. Descriptors are grouped according to their origin (malt (blue), hops (green), yeast (red), wild flora (yellow), Others (black)), and sensory aspect (aroma, taste, palate, and overall appreciation). Please note that for the chemical compounds, for the sake of clarity, only a subset of the total number of measured compounds is shown, with an emphasis on the key compounds for each source. For more details, see the main text and Methods section. Chemical data can be found in Supplementary Data  1 , correlations between all chemical compounds are depicted in Supplementary Fig.  S2 and correlation values can be found in Supplementary Data  2 . See Supplementary Data  4 for sensory panel assessments and Supplementary Data  5 for correlation values between all sensory descriptors.

Interestingly, different beer styles show distinct patterns for some flavor compounds (Supplementary Fig.  S3 ). These observations agree with expectations for key beer styles, and serve as a control for our measurements. For instance, Stouts generally show high values for color (darker), while hoppy beers contain elevated levels of iso-alpha acids, compounds associated with bitter hop taste. Acetic and lactic acid are not prevalent in most beers, with notable exceptions such as Kriek, Lambic, Faro, West Flanders ales and Flanders Old Brown, which use acid-producing bacteria ( Lactobacillus and Pediococcus ) or unconventional yeast ( Brettanomyces ) 54 , 55 . Glycerol, ethanol and esters show similar distributions across all beer styles, reflecting their common origin as products of yeast metabolism during fermentation 45 , 53 . Finally, low/no-alcohol beers contain low concentrations of glycerol and esters. This is in line with the production process for most of the low/no-alcohol beers in our dataset, which are produced through limiting fermentation or by stripping away alcohol via evaporation or dialysis, with both methods having the unintended side-effect of reducing the amount of flavor compounds in the final beer 56 , 57 .

Besides expected associations, our data also reveals less trivial associations between beer styles and specific parameters. For example, geraniol and citronellol, two monoterpenoids responsible for citrus, floral and rose flavors and characteristic of Citra hops, are found in relatively high amounts in Christmas, Saison, and Brett/co-fermented beers, where they may originate from terpenoid-rich spices such as coriander seeds instead of hops 58 .

Tasting panel assessments reveal sensorial relationships in beer

To assess the sensory profile of each beer, a trained tasting panel evaluated each of the 250 beers for 50 sensory attributes, including different hop, malt and yeast flavors, off-flavors and spices. Panelists used a tasting sheet (Supplementary Data  3 ) to score the different attributes. Panel consistency was evaluated by repeating 12 samples across different sessions and performing ANOVA. In 95% of cases no significant difference was found across sessions ( p  > 0.05), indicating good panel consistency (Supplementary Table  S2 ).

Aroma and taste perception reported by the trained panel are often linked (Fig.  1 , bottom left panel and Supplementary Data  4 and 5 ), with high correlations between hops aroma and taste (Spearman’s rho=0.83). Bitter taste was found to correlate with hop aroma and taste in general (Spearman’s rho=0.80 and 0.69), and particularly with “grassy” noble hops (Spearman’s rho=0.75). Barnyard flavor, most often associated with sour beers, is identified together with stale hops (Spearman’s rho=0.97) that are used in these beers. Lactic and acetic acid, which often co-occur, are correlated (Spearman’s rho=0.66). Interestingly, sweetness and bitterness are anti-correlated (Spearman’s rho = −0.48), confirming the hypothesis that they mask each other 59 , 60 . Beer body is highly correlated with alcohol (Spearman’s rho = 0.79), and overall appreciation is found to correlate with multiple aspects that describe beer mouthfeel (alcohol, carbonation; Spearman’s rho= 0.32, 0.39), as well as with hop and ester aroma intensity (Spearman’s rho=0.39 and 0.35).

Similar to the chemical analyses, sensorial analyses confirmed typical features of specific beer styles (Supplementary Fig.  S4 ). For example, sour beers (Faro, Flanders Old Brown, Fruit beer, Kriek, Lambic, West Flanders ale) were rated acidic, with flavors of both acetic and lactic acid. Hoppy beers were found to be bitter and showed hop-associated aromas like citrus and tropical fruit. Malt taste is most detected among scotch, stout/porters, and strong ales, while low/no-alcohol beers, which often have a reputation for being ‘worty’ (reminiscent of unfermented, sweet malt extract) appear in the middle. Unsurprisingly, hop aromas are most strongly detected among hoppy beers. Like its chemical counterpart (Supplementary Fig.  S3 ), acidity shows a right-skewed distribution, with the most acidic beers being Krieks, Lambics, and West Flanders ales.

Tasting panel assessments of specific flavors correlate with chemical composition

We find that the concentrations of several chemical compounds strongly correlate with specific aroma or taste, as evaluated by the tasting panel (Fig.  2 , Supplementary Fig.  S5 , Supplementary Data  6 ). In some cases, these correlations confirm expectations and serve as a useful control for data quality. For example, iso-alpha acids, the bittering compounds in hops, strongly correlate with bitterness (Spearman’s rho=0.68), while ethanol and glycerol correlate with tasters’ perceptions of alcohol and body, the mouthfeel sensation of fullness (Spearman’s rho=0.82/0.62 and 0.72/0.57 respectively) and darker color from roasted malts is a good indication of malt perception (Spearman’s rho=0.54).

Heatmap colors indicate Spearman’s Rho. Axes are organized according to sensory categories (aroma, taste, mouthfeel, overall), chemical categories and chemical sources in beer (malt (blue), hops (green), yeast (red), wild flora (yellow), Others (black)). See Supplementary Data  6 for all correlation values.

Interestingly, for some relationships between chemical compounds and perceived flavor, correlations are weaker than expected. For example, the rose-smelling phenethyl acetate only weakly correlates with floral aroma. This hints at more complex relationships and interactions between compounds and suggests a need for a more complex model than simple correlations. Lastly, we uncovered unexpected correlations. For instance, the esters ethyl decanoate and ethyl octanoate appear to correlate slightly with hop perception and bitterness, possibly due to their fruity flavor. Iron is anti-correlated with hop aromas and bitterness, most likely because it is also anti-correlated with iso-alpha acids. This could be a sign of metal chelation of hop acids 61 , given that our analyses measure unbound hop acids and total iron content, or could result from the higher iron content in dark and Fruit beers, which typically have less hoppy and bitter flavors 62 .

Public consumer reviews complement expert panel data

To complement and expand the sensory data of our trained tasting panel, we collected 180,000 reviews of our 250 beers from the online consumer review platform RateBeer. This provided numerical scores for beer appearance, aroma, taste, palate, overall quality as well as the average overall score.

Public datasets are known to suffer from biases, such as price, cult status and psychological conformity towards previous ratings of a product. For example, prices correlate with appreciation scores for these online consumer reviews (rho=0.49, Supplementary Fig.  S6 ), but not for our trained tasting panel (rho=0.19). This suggests that prices affect consumer appreciation, which has been reported in wine 63 , while blind tastings are unaffected. Moreover, we observe that some beer styles, like lagers and non-alcoholic beers, generally receive lower scores, reflecting that online reviewers are mostly beer aficionados with a preference for specialty beers over lager beers. In general, we find a modest correlation between our trained panel’s overall appreciation score and the online consumer appreciation scores (Fig.  3 , rho=0.29). Apart from the aforementioned biases in the online datasets, serving temperature, sample freshness and surroundings, which are all tightly controlled during the tasting panel sessions, can vary tremendously across online consumers and can further contribute to (among others, appreciation) differences between the two categories of tasters. Importantly, in contrast to the overall appreciation scores, for many sensory aspects the results from the professional panel correlated well with results obtained from RateBeer reviews. Correlations were highest for features that are relatively easy to recognize even for untrained tasters, like bitterness, sweetness, alcohol and malt aroma (Fig.  3 and below).

RateBeer text mining results can be found in Supplementary Data  7 . Rho values shown are Spearman correlation values, with asterisks indicating significant correlations ( p  < 0.05, two-sided). All p values were smaller than 0.001, except for Esters aroma (0.0553), Esters taste (0.3275), Esters aroma—banana (0.0019), Coriander (0.0508) and Diacetyl (0.0134).

Besides collecting consumer appreciation from these online reviews, we developed automated text analysis tools to gather additional data from review texts (Supplementary Data  7 ). Processing review texts on the RateBeer database yielded comparable results to the scores given by the trained panel for many common sensory aspects, including acidity, bitterness, sweetness, alcohol, malt, and hop tastes (Fig.  3 ). This is in line with what would be expected, since these attributes require less training for accurate assessment and are less influenced by environmental factors such as temperature, serving glass and odors in the environment. Consumer reviews also correlate well with our trained panel for 4-vinyl guaiacol, a compound associated with a very characteristic aroma. By contrast, correlations for more specific aromas like ester, coriander or diacetyl are underrepresented in the online reviews, underscoring the importance of using a trained tasting panel and standardized tasting sheets with explicit factors to be scored for evaluating specific aspects of a beer. Taken together, our results suggest that public reviews are trustworthy for some, but not all, flavor features and can complement or substitute taste panel data for these sensory aspects.

Models can predict beer sensory profiles from chemical data

The rich datasets of chemical analyses, tasting panel assessments and public reviews gathered in the first part of this study provided us with a unique opportunity to develop predictive models that link chemical data to sensorial features. Given the complexity of beer flavor, basic statistical tools such as correlations or linear regression may not always be the most suitable for making accurate predictions. Instead, we applied different machine learning models that can model both simple linear and complex interactive relationships. Specifically, we constructed a set of regression models to predict (a) trained panel scores for beer flavor and quality and (b) public reviews’ appreciation scores from beer chemical profiles. We trained and tested 10 different models (Methods), 3 linear regression-based models (simple linear regression with first-order interactions (LR), lasso regression with first-order interactions (Lasso), partial least squares regressor (PLSR)), 5 decision tree models (AdaBoost regressor (ABR), extra trees (ET), gradient boosting regressor (GBR), random forest (RF) and XGBoost regressor (XGBR)), 1 support vector regression (SVR), and 1 artificial neural network (ANN) model.

To compare the performance of our machine learning models, the dataset was randomly split into a training and test set, stratified by beer style. After a model was trained on data in the training set, its performance was evaluated on its ability to predict the test dataset obtained from multi-output models (based on the coefficient of determination, see Methods). Additionally, individual-attribute models were ranked per descriptor and the average rank was calculated, as proposed by Korneva et al. 64 . Importantly, both ways of evaluating the models’ performance agreed in general. Performance of the different models varied (Table  1 ). It should be noted that all models perform better at predicting RateBeer results than results from our trained tasting panel. One reason could be that sensory data is inherently variable, and this variability is averaged out with the large number of public reviews from RateBeer. Additionally, all tree-based models perform better at predicting taste than aroma. Linear models (LR) performed particularly poorly, with negative R 2 values, due to severe overfitting (training set R 2  = 1). Overfitting is a common issue in linear models with many parameters and limited samples, especially with interaction terms further amplifying the number of parameters. L1 regularization (Lasso) successfully overcomes this overfitting, out-competing multiple tree-based models on the RateBeer dataset. Similarly, the dimensionality reduction of PLSR avoids overfitting and improves performance, to some extent. Still, tree-based models (ABR, ET, GBR, RF and XGBR) show the best performance, out-competing the linear models (LR, Lasso, PLSR) commonly used in sensory science 65 .

GBR models showed the best overall performance in predicting sensory responses from chemical information, with R 2 values up to 0.75 depending on the predicted sensory feature (Supplementary Table  S4 ). The GBR models predict consumer appreciation (RateBeer) better than our trained panel’s appreciation (R 2 value of 0.67 compared to R 2 value of 0.09) (Supplementary Table  S3 and Supplementary Table  S4 ). ANN models showed intermediate performance, likely because neural networks typically perform best with larger datasets 66 . The SVR shows intermediate performance, mostly due to the weak predictions of specific attributes that lower the overall performance (Supplementary Table  S4 ).

Model dissection identifies specific, unexpected compounds as drivers of consumer appreciation

Next, we leveraged our models to infer important contributors to sensory perception and consumer appreciation. Consumer preference is a crucial sensory aspects, because a product that shows low consumer appreciation scores often does not succeed commercially 25 . Additionally, the requirement for a large number of representative evaluators makes consumer trials one of the more costly and time-consuming aspects of product development. Hence, a model for predicting chemical drivers of overall appreciation would be a welcome addition to the available toolbox for food development and optimization.

Since GBR models on our RateBeer dataset showed the best overall performance, we focused on these models. Specifically, we used two approaches to identify important contributors. First, rankings of the most important predictors for each sensorial trait in the GBR models were obtained based on impurity-based feature importance (mean decrease in impurity). High-ranked parameters were hypothesized to be either the true causal chemical properties underlying the trait, to correlate with the actual causal properties, or to take part in sensory interactions affecting the trait 67 (Fig.  4A ). In a second approach, we used SHAP 68 to determine which parameters contributed most to the model for making predictions of consumer appreciation (Fig.  4B ). SHAP calculates parameter contributions to model predictions on a per-sample basis, which can be aggregated into an importance score.

A The impurity-based feature importance (mean deviance in impurity, MDI) calculated from the Gradient Boosting Regression (GBR) model predicting RateBeer appreciation scores. The top 15 highest ranked chemical properties are shown. B SHAP summary plot for the top 15 parameters contributing to our GBR model. Each point on the graph represents a sample from our dataset. The color represents the concentration of that parameter, with bluer colors representing low values and redder colors representing higher values. Greater absolute values on the horizontal axis indicate a higher impact of the parameter on the prediction of the model. C Spearman correlations between the 15 most important chemical properties and consumer overall appreciation. Numbers indicate the Spearman Rho correlation coefficient, and the rank of this correlation compared to all other correlations. The top 15 important compounds were determined using SHAP (panel B).

Both approaches identified ethyl acetate as the most predictive parameter for beer appreciation (Fig.  4 ). Ethyl acetate is the most abundant ester in beer with a typical ‘fruity’, ‘solvent’ and ‘alcoholic’ flavor, but is often considered less important than other esters like isoamyl acetate. The second most important parameter identified by SHAP is ethanol, the most abundant beer compound after water. Apart from directly contributing to beer flavor and mouthfeel, ethanol drastically influences the physical properties of beer, dictating how easily volatile compounds escape the beer matrix to contribute to beer aroma 69 . Importantly, it should also be noted that the importance of ethanol for appreciation is likely inflated by the very low appreciation scores of non-alcoholic beers (Supplementary Fig.  S4 ). Despite not often being considered a driver of beer appreciation, protein level also ranks highly in both approaches, possibly due to its effect on mouthfeel and body 70 . Lactic acid, which contributes to the tart taste of sour beers, is the fourth most important parameter identified by SHAP, possibly due to the generally high appreciation of sour beers in our dataset.

Interestingly, some of the most important predictive parameters for our model are not well-established as beer flavors or are even commonly regarded as being negative for beer quality. For example, our models identify methanethiol and ethyl phenyl acetate, an ester commonly linked to beer staling 71 , as a key factor contributing to beer appreciation. Although there is no doubt that high concentrations of these compounds are considered unpleasant, the positive effects of modest concentrations are not yet known 72 , 73 .

To compare our approach to conventional statistics, we evaluated how well the 15 most important SHAP-derived parameters correlate with consumer appreciation (Fig.  4C ). Interestingly, only 6 of the properties derived by SHAP rank amongst the top 15 most correlated parameters. For some chemical compounds, the correlations are so low that they would have likely been considered unimportant. For example, lactic acid, the fourth most important parameter, shows a bimodal distribution for appreciation, with sour beers forming a separate cluster, that is missed entirely by the Spearman correlation. Additionally, the correlation plots reveal outliers, emphasizing the need for robust analysis tools. Together, this highlights the need for alternative models, like the Gradient Boosting model, that better grasp the complexity of (beer) flavor.

Finally, to observe the relationships between these chemical properties and their predicted targets, partial dependence plots were constructed for the six most important predictors of consumer appreciation 74 , 75 , 76 (Supplementary Fig.  S7 ). One-way partial dependence plots show how a change in concentration affects the predicted appreciation. These plots reveal an important limitation of our models: appreciation predictions remain constant at ever-increasing concentrations. This implies that once a threshold concentration is reached, further increasing the concentration does not affect appreciation. This is false, as it is well-documented that certain compounds become unpleasant at high concentrations, including ethyl acetate (‘nail polish’) 77 and methanethiol (‘sulfury’ and ‘rotten cabbage’) 78 . The inability of our models to grasp that flavor compounds have optimal levels, above which they become negative, is a consequence of working with commercial beer brands where (off-)flavors are rarely too high to negatively impact the product. The two-way partial dependence plots show how changing the concentration of two compounds influences predicted appreciation, visualizing their interactions (Supplementary Fig.  S7 ). In our case, the top 5 parameters are dominated by additive or synergistic interactions, with high concentrations for both compounds resulting in the highest predicted appreciation.

To assess the robustness of our best-performing models and model predictions, we performed 100 iterations of the GBR, RF and ET models. In general, all iterations of the models yielded similar performance (Supplementary Fig.  S8 ). Moreover, the main predictors (including the top predictors ethanol and ethyl acetate) remained virtually the same, especially for GBR and RF. For the iterations of the ET model, we did observe more variation in the top predictors, which is likely a consequence of the model’s inherent random architecture in combination with co-correlations between certain predictors. However, even in this case, several of the top predictors (ethanol and ethyl acetate) remain unchanged, although their rank in importance changes (Supplementary Fig.  S8 ).

Next, we investigated if a combination of RateBeer and trained panel data into one consolidated dataset would lead to stronger models, under the hypothesis that such a model would suffer less from bias in the datasets. A GBR model was trained to predict appreciation on the combined dataset. This model underperformed compared to the RateBeer model, both in the native case and when including a dataset identifier (R 2  = 0.67, 0.26 and 0.42 respectively). For the latter, the dataset identifier is the most important feature (Supplementary Fig.  S9 ), while most of the feature importance remains unchanged, with ethyl acetate and ethanol ranking highest, like in the original model trained only on RateBeer data. It seems that the large variation in the panel dataset introduces noise, weakening the models’ performances and reliability. In addition, it seems reasonable to assume that both datasets are fundamentally different, with the panel dataset obtained by blind tastings by a trained professional panel.

Lastly, we evaluated whether beer style identifiers would further enhance the model’s performance. A GBR model was trained with parameters that explicitly encoded the styles of the samples. This did not improve model performance (R2 = 0.66 with style information vs R2 = 0.67). The most important chemical features are consistent with the model trained without style information (eg. ethanol and ethyl acetate), and with the exception of the most preferred (strong ale) and least preferred (low/no-alcohol) styles, none of the styles were among the most important features (Supplementary Fig.  S9 , Supplementary Table  S5 and S6 ). This is likely due to a combination of style-specific chemical signatures, such as iso-alpha acids and lactic acid, that implicitly convey style information to the original models, as well as the low number of samples belonging to some styles, making it difficult for the model to learn style-specific patterns. Moreover, beer styles are not rigorously defined, with some styles overlapping in features and some beers being misattributed to a specific style, all of which leads to more noise in models that use style parameters.

Model validation

To test if our predictive models give insight into beer appreciation, we set up experiments aimed at improving existing commercial beers. We specifically selected overall appreciation as the trait to be examined because of its complexity and commercial relevance. Beer flavor comprises a complex bouquet rather than single aromas and tastes 53 . Hence, adding a single compound to the extent that a difference is noticeable may lead to an unbalanced, artificial flavor. Therefore, we evaluated the effect of combinations of compounds. Because Blond beers represent the most extensive style in our dataset, we selected a beer from this style as the starting material for these experiments (Beer 64 in Supplementary Data  1 ).

In the first set of experiments, we adjusted the concentrations of compounds that made up the most important predictors of overall appreciation (ethyl acetate, ethanol, lactic acid, ethyl phenyl acetate) together with correlated compounds (ethyl hexanoate, isoamyl acetate, glycerol), bringing them up to 95 th percentile ethanol-normalized concentrations (Methods) within the Blond group (‘Spiked’ concentration in Fig.  5A ). Compared to controls, the spiked beers were found to have significantly improved overall appreciation among trained panelists, with panelist noting increased intensity of ester flavors, sweetness, alcohol, and body fullness (Fig.  5B ). To disentangle the contribution of ethanol to these results, a second experiment was performed without the addition of ethanol. This resulted in a similar outcome, including increased perception of alcohol and overall appreciation.

Adding the top chemical compounds, identified as best predictors of appreciation by our model, into poorly appreciated beers results in increased appreciation from our trained panel. Results of sensory tests between base beers and those spiked with compounds identified as the best predictors by the model. A Blond and Non/Low-alcohol (0.0% ABV) base beers were brought up to 95th-percentile ethanol-normalized concentrations within each style. B For each sensory attribute, tasters indicated the more intense sample and selected the sample they preferred. The numbers above the bars correspond to the p values that indicate significant changes in perceived flavor (two-sided binomial test: alpha 0.05, n  = 20 or 13).

In a last experiment, we tested whether using the model’s predictions can boost the appreciation of a non-alcoholic beer (beer 223 in Supplementary Data  1 ). Again, the addition of a mixture of predicted compounds (omitting ethanol, in this case) resulted in a significant increase in appreciation, body, ester flavor and sweetness.

Predicting flavor and consumer appreciation from chemical composition is one of the ultimate goals of sensory science. A reliable, systematic and unbiased way to link chemical profiles to flavor and food appreciation would be a significant asset to the food and beverage industry. Such tools would substantially aid in quality control and recipe development, offer an efficient and cost-effective alternative to pilot studies and consumer trials and would ultimately allow food manufacturers to produce superior, tailor-made products that better meet the demands of specific consumer groups more efficiently.

A limited set of studies have previously tried, to varying degrees of success, to predict beer flavor and beer popularity based on (a limited set of) chemical compounds and flavors 79 , 80 . Current sensitive, high-throughput technologies allow measuring an unprecedented number of chemical compounds and properties in a large set of samples, yielding a dataset that can train models that help close the gaps between chemistry and flavor, even for a complex natural product like beer. To our knowledge, no previous research gathered data at this scale (250 samples, 226 chemical parameters, 50 sensory attributes and 5 consumer scores) to disentangle and validate the chemical aspects driving beer preference using various machine-learning techniques. We find that modern machine learning models outperform conventional statistical tools, such as correlations and linear models, and can successfully predict flavor appreciation from chemical composition. This could be attributed to the natural incorporation of interactions and non-linear or discontinuous effects in machine learning models, which are not easily grasped by the linear model architecture. While linear models and partial least squares regression represent the most widespread statistical approaches in sensory science, in part because they allow interpretation 65 , 81 , 82 , modern machine learning methods allow for building better predictive models while preserving the possibility to dissect and exploit the underlying patterns. Of the 10 different models we trained, tree-based models, such as our best performing GBR, showed the best overall performance in predicting sensory responses from chemical information, outcompeting artificial neural networks. This agrees with previous reports for models trained on tabular data 83 . Our results are in line with the findings of Colantonio et al. who also identified the gradient boosting architecture as performing best at predicting appreciation and flavor (of tomatoes and blueberries, in their specific study) 26 . Importantly, besides our larger experimental scale, we were able to directly confirm our models’ predictions in vivo.

Our study confirms that flavor compound concentration does not always correlate with perception, suggesting complex interactions that are often missed by more conventional statistics and simple models. Specifically, we find that tree-based algorithms may perform best in developing models that link complex food chemistry with aroma. Furthermore, we show that massive datasets of untrained consumer reviews provide a valuable source of data, that can complement or even replace trained tasting panels, especially for appreciation and basic flavors, such as sweetness and bitterness. This holds despite biases that are known to occur in such datasets, such as price or conformity bias. Moreover, GBR models predict taste better than aroma. This is likely because taste (e.g. bitterness) often directly relates to the corresponding chemical measurements (e.g., iso-alpha acids), whereas such a link is less clear for aromas, which often result from the interplay between multiple volatile compounds. We also find that our models are best at predicting acidity and alcohol, likely because there is a direct relation between the measured chemical compounds (acids and ethanol) and the corresponding perceived sensorial attribute (acidity and alcohol), and because even untrained consumers are generally able to recognize these flavors and aromas.

The predictions of our final models, trained on review data, hold even for blind tastings with small groups of trained tasters, as demonstrated by our ability to validate specific compounds as drivers of beer flavor and appreciation. Since adding a single compound to the extent of a noticeable difference may result in an unbalanced flavor profile, we specifically tested our identified key drivers as a combination of compounds. While this approach does not allow us to validate if a particular single compound would affect flavor and/or appreciation, our experiments do show that this combination of compounds increases consumer appreciation.

It is important to stress that, while it represents an important step forward, our approach still has several major limitations. A key weakness of the GBR model architecture is that amongst co-correlating variables, the largest main effect is consistently preferred for model building. As a result, co-correlating variables often have artificially low importance scores, both for impurity and SHAP-based methods, like we observed in the comparison to the more randomized Extra Trees models. This implies that chemicals identified as key drivers of a specific sensory feature by GBR might not be the true causative compounds, but rather co-correlate with the actual causative chemical. For example, the high importance of ethyl acetate could be (partially) attributed to the total ester content, ethanol or ethyl hexanoate (rho=0.77, rho=0.72 and rho=0.68), while ethyl phenylacetate could hide the importance of prenyl isobutyrate and ethyl benzoate (rho=0.77 and rho=0.76). Expanding our GBR model to include beer style as a parameter did not yield additional power or insight. This is likely due to style-specific chemical signatures, such as iso-alpha acids and lactic acid, that implicitly convey style information to the original model, as well as the smaller sample size per style, limiting the power to uncover style-specific patterns. This can be partly attributed to the curse of dimensionality, where the high number of parameters results in the models mainly incorporating single parameter effects, rather than complex interactions such as style-dependent effects 67 . A larger number of samples may overcome some of these limitations and offer more insight into style-specific effects. On the other hand, beer style is not a rigid scientific classification, and beers within one style often differ a lot, which further complicates the analysis of style as a model factor.

Our study is limited to beers from Belgian breweries. Although these beers cover a large portion of the beer styles available globally, some beer styles and consumer patterns may be missing, while other features might be overrepresented. For example, many Belgian ales exhibit yeast-driven flavor profiles, which is reflected in the chemical drivers of appreciation discovered by this study. In future work, expanding the scope to include diverse markets and beer styles could lead to the identification of even more drivers of appreciation and better models for special niche products that were not present in our beer set.

In addition to inherent limitations of GBR models, there are also some limitations associated with studying food aroma. Even if our chemical analyses measured most of the known aroma compounds, the total number of flavor compounds in complex foods like beer is still larger than the subset we were able to measure in this study. For example, hop-derived thiols, that influence flavor at very low concentrations, are notoriously difficult to measure in a high-throughput experiment. Moreover, consumer perception remains subjective and prone to biases that are difficult to avoid. It is also important to stress that the models are still immature and that more extensive datasets will be crucial for developing more complete models in the future. Besides more samples and parameters, our dataset does not include any demographic information about the tasters. Including such data could lead to better models that grasp external factors like age and culture. Another limitation is that our set of beers consists of high-quality end-products and lacks beers that are unfit for sale, which limits the current model in accurately predicting products that are appreciated very badly. Finally, while models could be readily applied in quality control, their use in sensory science and product development is restrained by their inability to discern causal relationships. Given that the models cannot distinguish compounds that genuinely drive consumer perception from those that merely correlate, validation experiments are essential to identify true causative compounds.

Despite the inherent limitations, dissection of our models enabled us to pinpoint specific molecules as potential drivers of beer aroma and consumer appreciation, including compounds that were unexpected and would not have been identified using standard approaches. Important drivers of beer appreciation uncovered by our models include protein levels, ethyl acetate, ethyl phenyl acetate and lactic acid. Currently, many brewers already use lactic acid to acidify their brewing water and ensure optimal pH for enzymatic activity during the mashing process. Our results suggest that adding lactic acid can also improve beer appreciation, although its individual effect remains to be tested. Interestingly, ethanol appears to be unnecessary to improve beer appreciation, both for blond beer and alcohol-free beer. Given the growing consumer interest in alcohol-free beer, with a predicted annual market growth of >7% 84 , it is relevant for brewers to know what compounds can further increase consumer appreciation of these beers. Hence, our model may readily provide avenues to further improve the flavor and consumer appreciation of both alcoholic and non-alcoholic beers, which is generally considered one of the key challenges for future beer production.

Whereas we see a direct implementation of our results for the development of superior alcohol-free beverages and other food products, our study can also serve as a stepping stone for the development of novel alcohol-containing beverages. We want to echo the growing body of scientific evidence for the negative effects of alcohol consumption, both on the individual level by the mutagenic, teratogenic and carcinogenic effects of ethanol 85 , 86 , as well as the burden on society caused by alcohol abuse and addiction. We encourage the use of our results for the production of healthier, tastier products, including novel and improved beverages with lower alcohol contents. Furthermore, we strongly discourage the use of these technologies to improve the appreciation or addictive properties of harmful substances.

The present work demonstrates that despite some important remaining hurdles, combining the latest developments in chemical analyses, sensory analysis and modern machine learning methods offers exciting avenues for food chemistry and engineering. Soon, these tools may provide solutions in quality control and recipe development, as well as new approaches to sensory science and flavor research.

Beer selection

250 commercial Belgian beers were selected to cover the broad diversity of beer styles and corresponding diversity in chemical composition and aroma. See Supplementary Fig.  S1 .

Chemical dataset

Sample preparation.

Beers within their expiration date were purchased from commercial retailers. Samples were prepared in biological duplicates at room temperature, unless explicitly stated otherwise. Bottle pressure was measured with a manual pressure device (Steinfurth Mess-Systeme GmbH) and used to calculate CO 2 concentration. The beer was poured through two filter papers (Macherey-Nagel, 500713032 MN 713 ¼) to remove carbon dioxide and prevent spontaneous foaming. Samples were then prepared for measurements by targeted Headspace-Gas Chromatography-Flame Ionization Detector/Flame Photometric Detector (HS-GC-FID/FPD), Headspace-Solid Phase Microextraction-Gas Chromatography-Mass Spectrometry (HS-SPME-GC-MS), colorimetric analysis, enzymatic analysis, Near-Infrared (NIR) analysis, as described in the sections below. The mean values of biological duplicates are reported for each compound.

HS-GC-FID/FPD

HS-GC-FID/FPD (Shimadzu GC 2010 Plus) was used to measure higher alcohols, acetaldehyde, esters, 4-vinyl guaicol, and sulfur compounds. Each measurement comprised 5 ml of sample pipetted into a 20 ml glass vial containing 1.75 g NaCl (VWR, 27810.295). 100 µl of 2-heptanol (Sigma-Aldrich, H3003) (internal standard) solution in ethanol (Fisher Chemical, E/0650DF/C17) was added for a final concentration of 2.44 mg/L. Samples were flushed with nitrogen for 10 s, sealed with a silicone septum, stored at −80 °C and analyzed in batches of 20.

The GC was equipped with a DB-WAXetr column (length, 30 m; internal diameter, 0.32 mm; layer thickness, 0.50 µm; Agilent Technologies, Santa Clara, CA, USA) to the FID and an HP-5 column (length, 30 m; internal diameter, 0.25 mm; layer thickness, 0.25 µm; Agilent Technologies, Santa Clara, CA, USA) to the FPD. N 2 was used as the carrier gas. Samples were incubated for 20 min at 70 °C in the headspace autosampler (Flow rate, 35 cm/s; Injection volume, 1000 µL; Injection mode, split; Combi PAL autosampler, CTC analytics, Switzerland). The injector, FID and FPD temperatures were kept at 250 °C. The GC oven temperature was first held at 50 °C for 5 min and then allowed to rise to 80 °C at a rate of 5 °C/min, followed by a second ramp of 4 °C/min until 200 °C kept for 3 min and a final ramp of (4 °C/min) until 230 °C for 1 min. Results were analyzed with the GCSolution software version 2.4 (Shimadzu, Kyoto, Japan). The GC was calibrated with a 5% EtOH solution (VWR International) containing the volatiles under study (Supplementary Table  S7 ).

HS-SPME-GC-MS

HS-SPME-GC-MS (Shimadzu GCMS-QP-2010 Ultra) was used to measure additional volatile compounds, mainly comprising terpenoids and esters. Samples were analyzed by HS-SPME using a triphase DVB/Carboxen/PDMS 50/30 μm SPME fiber (Supelco Co., Bellefonte, PA, USA) followed by gas chromatography (Thermo Fisher Scientific Trace 1300 series, USA) coupled to a mass spectrometer (Thermo Fisher Scientific ISQ series MS) equipped with a TriPlus RSH autosampler. 5 ml of degassed beer sample was placed in 20 ml vials containing 1.75 g NaCl (VWR, 27810.295). 5 µl internal standard mix was added, containing 2-heptanol (1 g/L) (Sigma-Aldrich, H3003), 4-fluorobenzaldehyde (1 g/L) (Sigma-Aldrich, 128376), 2,3-hexanedione (1 g/L) (Sigma-Aldrich, 144169) and guaiacol (1 g/L) (Sigma-Aldrich, W253200) in ethanol (Fisher Chemical, E/0650DF/C17). Each sample was incubated at 60 °C in the autosampler oven with constant agitation. After 5 min equilibration, the SPME fiber was exposed to the sample headspace for 30 min. The compounds trapped on the fiber were thermally desorbed in the injection port of the chromatograph by heating the fiber for 15 min at 270 °C.

The GC-MS was equipped with a low polarity RXi-5Sil MS column (length, 20 m; internal diameter, 0.18 mm; layer thickness, 0.18 µm; Restek, Bellefonte, PA, USA). Injection was performed in splitless mode at 320 °C, a split flow of 9 ml/min, a purge flow of 5 ml/min and an open valve time of 3 min. To obtain a pulsed injection, a programmed gas flow was used whereby the helium gas flow was set at 2.7 mL/min for 0.1 min, followed by a decrease in flow of 20 ml/min to the normal 0.9 mL/min. The temperature was first held at 30 °C for 3 min and then allowed to rise to 80 °C at a rate of 7 °C/min, followed by a second ramp of 2 °C/min till 125 °C and a final ramp of 8 °C/min with a final temperature of 270 °C.

Mass acquisition range was 33 to 550 amu at a scan rate of 5 scans/s. Electron impact ionization energy was 70 eV. The interface and ion source were kept at 275 °C and 250 °C, respectively. A mix of linear n-alkanes (from C7 to C40, Supelco Co.) was injected into the GC-MS under identical conditions to serve as external retention index markers. Identification and quantification of the compounds were performed using an in-house developed R script as described in Goelen et al. and Reher et al. 87 , 88 (for package information, see Supplementary Table  S8 ). Briefly, chromatograms were analyzed using AMDIS (v2.71) 89 to separate overlapping peaks and obtain pure compound spectra. The NIST MS Search software (v2.0 g) in combination with the NIST2017, FFNSC3 and Adams4 libraries were used to manually identify the empirical spectra, taking into account the expected retention time. After background subtraction and correcting for retention time shifts between samples run on different days based on alkane ladders, compound elution profiles were extracted and integrated using a file with 284 target compounds of interest, which were either recovered in our identified AMDIS list of spectra or were known to occur in beer. Compound elution profiles were estimated for every peak in every chromatogram over a time-restricted window using weighted non-negative least square analysis after which peak areas were integrated 87 , 88 . Batch effect correction was performed by normalizing against the most stable internal standard compound, 4-fluorobenzaldehyde. Out of all 284 target compounds that were analyzed, 167 were visually judged to have reliable elution profiles and were used for final analysis.

Discrete photometric and enzymatic analysis

Discrete photometric and enzymatic analysis (Thermo Scientific TM Gallery TM Plus Beermaster Discrete Analyzer) was used to measure acetic acid, ammonia, beta-glucan, iso-alpha acids, color, sugars, glycerol, iron, pH, protein, and sulfite. 2 ml of sample volume was used for the analyses. Information regarding the reagents and standard solutions used for analyses and calibrations is included in Supplementary Table  S7 and Supplementary Table  S9 .

NIR analyses

NIR analysis (Anton Paar Alcolyzer Beer ME System) was used to measure ethanol. Measurements comprised 50 ml of sample, and a 10% EtOH solution was used for calibration.

Correlation calculations

Pairwise Spearman Rank correlations were calculated between all chemical properties.

Sensory dataset

Trained panel.

Our trained tasting panel consisted of volunteers who gave prior verbal informed consent. All compounds used for the validation experiment were of food-grade quality. The tasting sessions were approved by the Social and Societal Ethics Committee of the KU Leuven (G-2022-5677-R2(MAR)). All online reviewers agreed to the Terms and Conditions of the RateBeer website.

Sensory analysis was performed according to the American Society of Brewing Chemists (ASBC) Sensory Analysis Methods 90 . 30 volunteers were screened through a series of triangle tests. The sixteen most sensitive and consistent tasters were retained as taste panel members. The resulting panel was diverse in age [22–42, mean: 29], sex [56% male] and nationality [7 different countries]. The panel developed a consensus vocabulary to describe beer aroma, taste and mouthfeel. Panelists were trained to identify and score 50 different attributes, using a 7-point scale to rate attributes’ intensity. The scoring sheet is included as Supplementary Data  3 . Sensory assessments took place between 10–12 a.m. The beers were served in black-colored glasses. Per session, between 5 and 12 beers of the same style were tasted at 12 °C to 16 °C. Two reference beers were added to each set and indicated as ‘Reference 1 & 2’, allowing panel members to calibrate their ratings. Not all panelists were present at every tasting. Scores were scaled by standard deviation and mean-centered per taster. Values are represented as z-scores and clustered by Euclidean distance. Pairwise Spearman correlations were calculated between taste and aroma sensory attributes. Panel consistency was evaluated by repeating samples on different sessions and performing ANOVA to identify differences, using the ‘stats’ package (v4.2.2) in R (for package information, see Supplementary Table  S8 ).

Online reviews from a public database

The ‘scrapy’ package in Python (v3.6) (for package information, see Supplementary Table  S8 ). was used to collect 232,288 online reviews (mean=922, min=6, max=5343) from RateBeer, an online beer review database. Each review entry comprised 5 numerical scores (appearance, aroma, taste, palate and overall quality) and an optional review text. The total number of reviews per reviewer was collected separately. Numerical scores were scaled and centered per rater, and mean scores were calculated per beer.

For the review texts, the language was estimated using the packages ‘langdetect’ and ‘langid’ in Python. Reviews that were classified as English by both packages were kept. Reviewers with fewer than 100 entries overall were discarded. 181,025 reviews from >6000 reviewers from >40 countries remained. Text processing was done using the ‘nltk’ package in Python. Texts were corrected for slang and misspellings; proper nouns and rare words that are relevant to the beer context were specified and kept as-is (‘Chimay’,’Lambic’, etc.). A dictionary of semantically similar sensorial terms, for example ‘floral’ and ‘flower’, was created and collapsed together into one term. Words were stemmed and lemmatized to avoid identifying words such as ‘acid’ and ‘acidity’ as separate terms. Numbers and punctuation were removed.

Sentences from up to 50 randomly chosen reviews per beer were manually categorized according to the aspect of beer they describe (appearance, aroma, taste, palate, overall quality—not to be confused with the 5 numerical scores described above) or flagged as irrelevant if they contained no useful information. If a beer contained fewer than 50 reviews, all reviews were manually classified. This labeled data set was used to train a model that classified the rest of the sentences for all beers 91 . Sentences describing taste and aroma were extracted, and term frequency–inverse document frequency (TFIDF) was implemented to calculate enrichment scores for sensorial words per beer.

The sex of the tasting subject was not considered when building our sensory database. Instead, results from different panelists were averaged, both for our trained panel (56% male, 44% female) and the RateBeer reviews (70% male, 30% female for RateBeer as a whole).

Beer price collection and processing

Beer prices were collected from the following stores: Colruyt, Delhaize, Total Wine, BeerHawk, The Belgian Beer Shop, The Belgian Shop, and Beer of Belgium. Where applicable, prices were converted to Euros and normalized per liter. Spearman correlations were calculated between these prices and mean overall appreciation scores from RateBeer and the taste panel, respectively.

Pairwise Spearman Rank correlations were calculated between all sensory properties.

Machine learning models

Predictive modeling of sensory profiles from chemical data.

Regression models were constructed to predict (a) trained panel scores for beer flavors and quality from beer chemical profiles and (b) public reviews’ appreciation scores from beer chemical profiles. Z-scores were used to represent sensory attributes in both data sets. Chemical properties with log-normal distributions (Shapiro-Wilk test, p  <  0.05 ) were log-transformed. Missing chemical measurements (0.1% of all data) were replaced with mean values per attribute. Observations from 250 beers were randomly separated into a training set (70%, 175 beers) and a test set (30%, 75 beers), stratified per beer style. Chemical measurements (p = 231) were normalized based on the training set average and standard deviation. In total, three linear regression-based models: linear regression with first-order interaction terms (LR), lasso regression with first-order interaction terms (Lasso) and partial least squares regression (PLSR); five decision tree models, Adaboost regressor (ABR), Extra Trees (ET), Gradient Boosting regressor (GBR), Random Forest (RF) and XGBoost regressor (XGBR); one support vector machine model (SVR) and one artificial neural network model (ANN) were trained. The models were implemented using the ‘scikit-learn’ package (v1.2.2) and ‘xgboost’ package (v1.7.3) in Python (v3.9.16). Models were trained, and hyperparameters optimized, using five-fold cross-validated grid search with the coefficient of determination (R 2 ) as the evaluation metric. The ANN (scikit-learn’s MLPRegressor) was optimized using Bayesian Tree-Structured Parzen Estimator optimization with the ‘Optuna’ Python package (v3.2.0). Individual models were trained per attribute, and a multi-output model was trained on all attributes simultaneously.

Model dissection

GBR was found to outperform other methods, resulting in models with the highest average R 2 values in both trained panel and public review data sets. Impurity-based rankings of the most important predictors for each predicted sensorial trait were obtained using the ‘scikit-learn’ package. To observe the relationships between these chemical properties and their predicted targets, partial dependence plots (PDP) were constructed for the six most important predictors of consumer appreciation 74 , 75 .

The ‘SHAP’ package in Python (v0.41.0) was implemented to provide an alternative ranking of predictor importance and to visualize the predictors’ effects as a function of their concentration 68 .

Validation of causal chemical properties

To validate the effects of the most important model features on predicted sensory attributes, beers were spiked with the chemical compounds identified by the models and descriptive sensory analyses were carried out according to the American Society of Brewing Chemists (ASBC) protocol 90 .

Compound spiking was done 30 min before tasting. Compounds were spiked into fresh beer bottles, that were immediately resealed and inverted three times. Fresh bottles of beer were opened for the same duration, resealed, and inverted thrice, to serve as controls. Pairs of spiked samples and controls were served simultaneously, chilled and in dark glasses as outlined in the Trained panel section above. Tasters were instructed to select the glass with the higher flavor intensity for each attribute (directional difference test 92 ) and to select the glass they prefer.

The final concentration after spiking was equal to the within-style average, after normalizing by ethanol concentration. This was done to ensure balanced flavor profiles in the final spiked beer. The same methods were applied to improve a non-alcoholic beer. Compounds were the following: ethyl acetate (Merck KGaA, W241415), ethyl hexanoate (Merck KGaA, W243906), isoamyl acetate (Merck KGaA, W205508), phenethyl acetate (Merck KGaA, W285706), ethanol (96%, Colruyt), glycerol (Merck KGaA, W252506), lactic acid (Merck KGaA, 261106).

Significant differences in preference or perceived intensity were determined by performing the two-sided binomial test on each attribute.

Reporting summary

Further information on research design is available in the  Nature Portfolio Reporting Summary linked to this article.

Data availability

The data that support the findings of this work are available in the Supplementary Data files and have been deposited to Zenodo under accession code 10653704 93 . The RateBeer scores data are under restricted access, they are not publicly available as they are property of RateBeer (ZX Ventures, USA). Access can be obtained from the authors upon reasonable request and with permission of RateBeer (ZX Ventures, USA).  Source data are provided with this paper.

Code availability

The code for training the machine learning models, analyzing the models, and generating the figures has been deposited to Zenodo under accession code 10653704 93 .

Tieman, D. et al. A chemical genetic roadmap to improved tomato flavor. Science 355 , 391–394 (2017).

Article   ADS   CAS   PubMed   Google Scholar  

Plutowska, B. & Wardencki, W. Application of gas chromatography–olfactometry (GC–O) in analysis and quality assessment of alcoholic beverages – A review. Food Chem. 107 , 449–463 (2008).

Article   CAS   Google Scholar  

Legin, A., Rudnitskaya, A., Seleznev, B. & Vlasov, Y. Electronic tongue for quality assessment of ethanol, vodka and eau-de-vie. Anal. Chim. Acta 534 , 129–135 (2005).

Loutfi, A., Coradeschi, S., Mani, G. K., Shankar, P. & Rayappan, J. B. B. Electronic noses for food quality: A review. J. Food Eng. 144 , 103–111 (2015).

Ahn, Y.-Y., Ahnert, S. E., Bagrow, J. P. & Barabási, A.-L. Flavor network and the principles of food pairing. Sci. Rep. 1 , 196 (2011).

Article   CAS   PubMed   PubMed Central   Google Scholar  

Bartoshuk, L. M. & Klee, H. J. Better fruits and vegetables through sensory analysis. Curr. Biol. 23 , R374–R378 (2013).

Article   CAS   PubMed   Google Scholar  

Piggott, J. R. Design questions in sensory and consumer science. Food Qual. Prefer. 3293 , 217–220 (1995).

Article   Google Scholar  

Kermit, M. & Lengard, V. Assessing the performance of a sensory panel-panellist monitoring and tracking. J. Chemom. 19 , 154–161 (2005).

Cook, D. J., Hollowood, T. A., Linforth, R. S. T. & Taylor, A. J. Correlating instrumental measurements of texture and flavour release with human perception. Int. J. Food Sci. Technol. 40 , 631–641 (2005).

Chinchanachokchai, S., Thontirawong, P. & Chinchanachokchai, P. A tale of two recommender systems: The moderating role of consumer expertise on artificial intelligence based product recommendations. J. Retail. Consum. Serv. 61 , 1–12 (2021).

Ross, C. F. Sensory science at the human-machine interface. Trends Food Sci. Technol. 20 , 63–72 (2009).

Chambers, E. IV & Koppel, K. Associations of volatile compounds with sensory aroma and flavor: The complex nature of flavor. Molecules 18 , 4887–4905 (2013).

Pinu, F. R. Metabolomics—The new frontier in food safety and quality research. Food Res. Int. 72 , 80–81 (2015).

Danezis, G. P., Tsagkaris, A. S., Brusic, V. & Georgiou, C. A. Food authentication: state of the art and prospects. Curr. Opin. Food Sci. 10 , 22–31 (2016).

Shepherd, G. M. Smell images and the flavour system in the human brain. Nature 444 , 316–321 (2006).

Meilgaard, M. C. Prediction of flavor differences between beers from their chemical composition. J. Agric. Food Chem. 30 , 1009–1017 (1982).

Xu, L. et al. Widespread receptor-driven modulation in peripheral olfactory coding. Science 368 , eaaz5390 (2020).

Kupferschmidt, K. Following the flavor. Science 340 , 808–809 (2013).

Billesbølle, C. B. et al. Structural basis of odorant recognition by a human odorant receptor. Nature 615 , 742–749 (2023).

Article   ADS   PubMed   PubMed Central   Google Scholar  

Smith, B. Perspective: Complexities of flavour. Nature 486 , S6–S6 (2012).

Pfister, P. et al. Odorant receptor inhibition is fundamental to odor encoding. Curr. Biol. 30 , 2574–2587 (2020).

Moskowitz, H. W., Kumaraiah, V., Sharma, K. N., Jacobs, H. L. & Sharma, S. D. Cross-cultural differences in simple taste preferences. Science 190 , 1217–1218 (1975).

Eriksson, N. et al. A genetic variant near olfactory receptor genes influences cilantro preference. Flavour 1 , 22 (2012).

Ferdenzi, C. et al. Variability of affective responses to odors: Culture, gender, and olfactory knowledge. Chem. Senses 38 , 175–186 (2013).

Article   PubMed   Google Scholar  

Lawless, H. T. & Heymann, H. Sensory evaluation of food: Principles and practices. (Springer, New York, NY). https://doi.org/10.1007/978-1-4419-6488-5 (2010).

Colantonio, V. et al. Metabolomic selection for enhanced fruit flavor. Proc. Natl. Acad. Sci. 119 , e2115865119 (2022).

Fritz, F., Preissner, R. & Banerjee, P. VirtualTaste: a web server for the prediction of organoleptic properties of chemical compounds. Nucleic Acids Res 49 , W679–W684 (2021).

Tuwani, R., Wadhwa, S. & Bagler, G. BitterSweet: Building machine learning models for predicting the bitter and sweet taste of small molecules. Sci. Rep. 9 , 1–13 (2019).

Dagan-Wiener, A. et al. Bitter or not? BitterPredict, a tool for predicting taste from chemical structure. Sci. Rep. 7 , 1–13 (2017).

Pallante, L. et al. Toward a general and interpretable umami taste predictor using a multi-objective machine learning approach. Sci. Rep. 12 , 1–11 (2022).

Malavolta, M. et al. A survey on computational taste predictors. Eur. Food Res. Technol. 248 , 2215–2235 (2022).

Lee, B. K. et al. A principal odor map unifies diverse tasks in olfactory perception. Science 381 , 999–1006 (2023).

Mayhew, E. J. et al. Transport features predict if a molecule is odorous. Proc. Natl. Acad. Sci. 119 , e2116576119 (2022).

Niu, Y. et al. Sensory evaluation of the synergism among ester odorants in light aroma-type liquor by odor threshold, aroma intensity and flash GC electronic nose. Food Res. Int. 113 , 102–114 (2018).

Yu, P., Low, M. Y. & Zhou, W. Design of experiments and regression modelling in food flavour and sensory analysis: A review. Trends Food Sci. Technol. 71 , 202–215 (2018).

Oladokun, O. et al. The impact of hop bitter acid and polyphenol profiles on the perceived bitterness of beer. Food Chem. 205 , 212–220 (2016).

Linforth, R., Cabannes, M., Hewson, L., Yang, N. & Taylor, A. Effect of fat content on flavor delivery during consumption: An in vivo model. J. Agric. Food Chem. 58 , 6905–6911 (2010).

Guo, S., Na Jom, K. & Ge, Y. Influence of roasting condition on flavor profile of sunflower seeds: A flavoromics approach. Sci. Rep. 9 , 11295 (2019).

Ren, Q. et al. The changes of microbial community and flavor compound in the fermentation process of Chinese rice wine using Fagopyrum tataricum grain as feedstock. Sci. Rep. 9 , 3365 (2019).

Hastie, T., Friedman, J. & Tibshirani, R. The Elements of Statistical Learning. (Springer, New York, NY). https://doi.org/10.1007/978-0-387-21606-5 (2001).

Dietz, C., Cook, D., Huismann, M., Wilson, C. & Ford, R. The multisensory perception of hop essential oil: a review. J. Inst. Brew. 126 , 320–342 (2020).

CAS   Google Scholar  

Roncoroni, Miguel & Verstrepen, Kevin Joan. Belgian Beer: Tested and Tasted. (Lannoo, 2018).

Meilgaard, M. Flavor chemistry of beer: Part II: Flavor and threshold of 239 aroma volatiles. in (1975).

Bokulich, N. A. & Bamforth, C. W. The microbiology of malting and brewing. Microbiol. Mol. Biol. Rev. MMBR 77 , 157–172 (2013).

Dzialo, M. C., Park, R., Steensels, J., Lievens, B. & Verstrepen, K. J. Physiology, ecology and industrial applications of aroma formation in yeast. FEMS Microbiol. Rev. 41 , S95–S128 (2017).

Article   PubMed   PubMed Central   Google Scholar  

Datta, A. et al. Computer-aided food engineering. Nat. Food 3 , 894–904 (2022).

American Society of Brewing Chemists. Beer Methods. (American Society of Brewing Chemists, St. Paul, MN, U.S.A.).

Olaniran, A. O., Hiralal, L., Mokoena, M. P. & Pillay, B. Flavour-active volatile compounds in beer: production, regulation and control. J. Inst. Brew. 123 , 13–23 (2017).

Verstrepen, K. J. et al. Flavor-active esters: Adding fruitiness to beer. J. Biosci. Bioeng. 96 , 110–118 (2003).

Meilgaard, M. C. Flavour chemistry of beer. part I: flavour interaction between principal volatiles. Master Brew. Assoc. Am. Tech. Q 12 , 107–117 (1975).

Briggs, D. E., Boulton, C. A., Brookes, P. A. & Stevens, R. Brewing 227–254. (Woodhead Publishing). https://doi.org/10.1533/9781855739062.227 (2004).

Bossaert, S., Crauwels, S., De Rouck, G. & Lievens, B. The power of sour - A review: Old traditions, new opportunities. BrewingScience 72 , 78–88 (2019).

Google Scholar  

Verstrepen, K. J. et al. Flavor active esters: Adding fruitiness to beer. J. Biosci. Bioeng. 96 , 110–118 (2003).

Snauwaert, I. et al. Microbial diversity and metabolite composition of Belgian red-brown acidic ales. Int. J. Food Microbiol. 221 , 1–11 (2016).

Spitaels, F. et al. The microbial diversity of traditional spontaneously fermented lambic beer. PLoS ONE 9 , e95384 (2014).

Blanco, C. A., Andrés-Iglesias, C. & Montero, O. Low-alcohol Beers: Flavor Compounds, Defects, and Improvement Strategies. Crit. Rev. Food Sci. Nutr. 56 , 1379–1388 (2016).

Jackowski, M. & Trusek, A. Non-Alcohol. beer Prod. – Overv. 20 , 32–38 (2018).

Takoi, K. et al. The contribution of geraniol metabolism to the citrus flavour of beer: Synergy of geraniol and β-citronellol under coexistence with excess linalool. J. Inst. Brew. 116 , 251–260 (2010).

Kroeze, J. H. & Bartoshuk, L. M. Bitterness suppression as revealed by split-tongue taste stimulation in humans. Physiol. Behav. 35 , 779–783 (1985).

Mennella, J. A. et al. A spoonful of sugar helps the medicine go down”: Bitter masking bysucrose among children and adults. Chem. Senses 40 , 17–25 (2015).

Wietstock, P., Kunz, T., Perreira, F. & Methner, F.-J. Metal chelation behavior of hop acids in buffered model systems. BrewingScience 69 , 56–63 (2016).

Sancho, D., Blanco, C. A., Caballero, I. & Pascual, A. Free iron in pale, dark and alcohol-free commercial lager beers. J. Sci. Food Agric. 91 , 1142–1147 (2011).

Rodrigues, H. & Parr, W. V. Contribution of cross-cultural studies to understanding wine appreciation: A review. Food Res. Int. 115 , 251–258 (2019).

Korneva, E. & Blockeel, H. Towards better evaluation of multi-target regression models. in ECML PKDD 2020 Workshops (eds. Koprinska, I. et al.) 353–362 (Springer International Publishing, Cham, 2020). https://doi.org/10.1007/978-3-030-65965-3_23 .

Gastón Ares. Mathematical and Statistical Methods in Food Science and Technology. (Wiley, 2013).

Grinsztajn, L., Oyallon, E. & Varoquaux, G. Why do tree-based models still outperform deep learning on tabular data? Preprint at http://arxiv.org/abs/2207.08815 (2022).

Gries, S. T. Statistics for Linguistics with R: A Practical Introduction. in Statistics for Linguistics with R (De Gruyter Mouton, 2021). https://doi.org/10.1515/9783110718256 .

Lundberg, S. M. et al. From local explanations to global understanding with explainable AI for trees. Nat. Mach. Intell. 2 , 56–67 (2020).

Ickes, C. M. & Cadwallader, K. R. Effects of ethanol on flavor perception in alcoholic beverages. Chemosens. Percept. 10 , 119–134 (2017).

Kato, M. et al. Influence of high molecular weight polypeptides on the mouthfeel of commercial beer. J. Inst. Brew. 127 , 27–40 (2021).

Wauters, R. et al. Novel Saccharomyces cerevisiae variants slow down the accumulation of staling aldehydes and improve beer shelf-life. Food Chem. 398 , 1–11 (2023).

Li, H., Jia, S. & Zhang, W. Rapid determination of low-level sulfur compounds in beer by headspace gas chromatography with a pulsed flame photometric detector. J. Am. Soc. Brew. Chem. 66 , 188–191 (2008).

Dercksen, A., Laurens, J., Torline, P., Axcell, B. C. & Rohwer, E. Quantitative analysis of volatile sulfur compounds in beer using a membrane extraction interface. J. Am. Soc. Brew. Chem. 54 , 228–233 (1996).

Molnar, C. Interpretable Machine Learning: A Guide for Making Black-Box Models Interpretable. (2020).

Zhao, Q. & Hastie, T. Causal interpretations of black-box models. J. Bus. Econ. Stat. Publ. Am. Stat. Assoc. 39 , 272–281 (2019).

Article   MathSciNet   Google Scholar  

Hastie, T., Tibshirani, R. & Friedman, J. The Elements of Statistical Learning. (Springer, 2019).

Labrado, D. et al. Identification by NMR of key compounds present in beer distillates and residual phases after dealcoholization by vacuum distillation. J. Sci. Food Agric. 100 , 3971–3978 (2020).

Lusk, L. T., Kay, S. B., Porubcan, A. & Ryder, D. S. Key olfactory cues for beer oxidation. J. Am. Soc. Brew. Chem. 70 , 257–261 (2012).

Gonzalez Viejo, C., Torrico, D. D., Dunshea, F. R. & Fuentes, S. Development of artificial neural network models to assess beer acceptability based on sensory properties using a robotic pourer: A comparative model approach to achieve an artificial intelligence system. Beverages 5 , 33 (2019).

Gonzalez Viejo, C., Fuentes, S., Torrico, D. D., Godbole, A. & Dunshea, F. R. Chemical characterization of aromas in beer and their effect on consumers liking. Food Chem. 293 , 479–485 (2019).

Gilbert, J. L. et al. Identifying breeding priorities for blueberry flavor using biochemical, sensory, and genotype by environment analyses. PLOS ONE 10 , 1–21 (2015).

Goulet, C. et al. Role of an esterase in flavor volatile variation within the tomato clade. Proc. Natl. Acad. Sci. 109 , 19009–19014 (2012).

Article   ADS   CAS   PubMed   PubMed Central   Google Scholar  

Borisov, V. et al. Deep Neural Networks and Tabular Data: A Survey. IEEE Trans. Neural Netw. Learn. Syst. 1–21 https://doi.org/10.1109/TNNLS.2022.3229161 (2022).

Statista. Statista Consumer Market Outlook: Beer - Worldwide.

Seitz, H. K. & Stickel, F. Molecular mechanisms of alcoholmediated carcinogenesis. Nat. Rev. Cancer 7 , 599–612 (2007).

Voordeckers, K. et al. Ethanol exposure increases mutation rate through error-prone polymerases. Nat. Commun. 11 , 3664 (2020).

Goelen, T. et al. Bacterial phylogeny predicts volatile organic compound composition and olfactory response of an aphid parasitoid. Oikos 129 , 1415–1428 (2020).

Article   ADS   Google Scholar  

Reher, T. et al. Evaluation of hop (Humulus lupulus) as a repellent for the management of Drosophila suzukii. Crop Prot. 124 , 104839 (2019).

Stein, S. E. An integrated method for spectrum extraction and compound identification from gas chromatography/mass spectrometry data. J. Am. Soc. Mass Spectrom. 10 , 770–781 (1999).

American Society of Brewing Chemists. Sensory Analysis Methods. (American Society of Brewing Chemists, St. Paul, MN, U.S.A., 1992).

McAuley, J., Leskovec, J. & Jurafsky, D. Learning Attitudes and Attributes from Multi-Aspect Reviews. Preprint at https://doi.org/10.48550/arXiv.1210.3926 (2012).

Meilgaard, M. C., Carr, B. T. & Carr, B. T. Sensory Evaluation Techniques. (CRC Press, Boca Raton). https://doi.org/10.1201/b16452 (2014).

Schreurs, M. et al. Data from: Predicting and improving complex beer flavor through machine learning. Zenodo https://doi.org/10.5281/zenodo.10653704 (2024).

Download references

Acknowledgements

We thank all lab members for their discussions and thank all tasting panel members for their contributions. Special thanks go out to Dr. Karin Voordeckers for her tremendous help in proofreading and improving the manuscript. M.S. was supported by a Baillet-Latour fellowship, L.C. acknowledges financial support from KU Leuven (C16/17/006), F.A.T. was supported by a PhD fellowship from FWO (1S08821N). Research in the lab of K.J.V. is supported by KU Leuven, FWO, VIB, VLAIO and the Brewing Science Serves Health Fund. Research in the lab of T.W. is supported by FWO (G.0A51.15) and KU Leuven (C16/17/006).

Author information

These authors contributed equally: Michiel Schreurs, Supinya Piampongsant, Miguel Roncoroni.

Authors and Affiliations

VIB—KU Leuven Center for Microbiology, Gaston Geenslaan 1, B-3001, Leuven, Belgium

Michiel Schreurs, Supinya Piampongsant, Miguel Roncoroni, Lloyd Cool, Beatriz Herrera-Malaver, Florian A. Theßeling & Kevin J. Verstrepen

CMPG Laboratory of Genetics and Genomics, KU Leuven, Gaston Geenslaan 1, B-3001, Leuven, Belgium

Leuven Institute for Beer Research (LIBR), Gaston Geenslaan 1, B-3001, Leuven, Belgium

Laboratory of Socioecology and Social Evolution, KU Leuven, Naamsestraat 59, B-3000, Leuven, Belgium

Lloyd Cool, Christophe Vanderaa & Tom Wenseleers

VIB Bioinformatics Core, VIB, Rijvisschestraat 120, B-9052, Ghent, Belgium

Łukasz Kreft & Alexander Botzki

AB InBev SA/NV, Brouwerijplein 1, B-3000, Leuven, Belgium

Philippe Malcorps & Luk Daenen

You can also search for this author in PubMed   Google Scholar

Contributions

S.P., M.S. and K.J.V. conceived the experiments. S.P., M.S. and K.J.V. designed the experiments. S.P., M.S., M.R., B.H. and F.A.T. performed the experiments. S.P., M.S., L.C., C.V., L.K., A.B., P.M., L.D., T.W. and K.J.V. contributed analysis ideas. S.P., M.S., L.C., C.V., T.W. and K.J.V. analyzed the data. All authors contributed to writing the manuscript.

Corresponding author

Correspondence to Kevin J. Verstrepen .

Ethics declarations

Competing interests.

K.J.V. is affiliated with bar.on. The other authors declare no competing interests.

Peer review

Peer review information.

Nature Communications thanks Florian Bauer, Andrew John Macintosh and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary information, peer review file, description of additional supplementary files, supplementary data 1, supplementary data 2, supplementary data 3, supplementary data 4, supplementary data 5, supplementary data 6, supplementary data 7, reporting summary, source data, source data, rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Schreurs, M., Piampongsant, S., Roncoroni, M. et al. Predicting and improving complex beer flavor through machine learning. Nat Commun 15 , 2368 (2024). https://doi.org/10.1038/s41467-024-46346-0

Download citation

Received : 30 October 2023

Accepted : 21 February 2024

Published : 26 March 2024

DOI : https://doi.org/10.1038/s41467-024-46346-0

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing: Translational Research newsletter — top stories in biotechnology, drug discovery and pharma.

• Skip to main content
• Keyboard shortcuts for audio player

• Your Health
• Treatments & Tests
• Health Inc.
• Public Health

How to Thrive as You Age

Women who do strength training live longer. how much is enough.

Allison Aubrey

Strength training is good for everyone, but women who train regularly get a significantly higher boost in longevity than men. Gary Yeowell/Getty Images hide caption

Strength training is good for everyone, but women who train regularly get a significantly higher boost in longevity than men.

Resistance training does more than help us build strong muscles.

A new study finds women who do strength training exercises two to three days a week are more likely to live longer and have a lower risk of death from heart disease, compared to women who do none.

"We were incredibly impressed by the finding," says study author Martha Gulati , who is also the director of preventive cardiology at Cedars Sinai in Los Angeles.

Of the 400,000 people included in the study, only 1 in 5 women did regular weight training. But those who did, saw tremendous benefits.

"What surprised us the most was the fact that women who do muscle strengthening had a reduction in their cardiovascular mortality by 30%," Gulati says. "We don't have many things that reduce mortality in that way."

Strength training is also good for bones, joints, mood and metabolic health. And at a time when many women focus on aerobic activity and hesitate to do weight training, the findings add to the evidence that a combination of both types of exercise is powerful medicine. "Both should be prescribed," Gulati says.

Shots - Health News

Millions of women are 'under-muscled.' these foods help build strength.

The findings are part of a larger study, published in The Journal of the American College of Cardiology, which evaluated the differences in the effects of exercise between men and women.

While the study finds that even small doses of exercise are beneficial for everyone, the data show that women need less exercise than men to get the same gains in longevity.

Women who did moderate intensity exercise, such as brisk walking, five times a week, reduced their risk of premature death by 24%, compared to 18% for men.

"The take home message is – let's start moving," says Eric Shiroma , a prevention-focused researcher at the National Heart, Lung, and Blood Institute, part of the National Institutes of Health, which provided grant support for the research..

It's not exactly clear what drives the variance between sexes, but there are physiological differences between men and women, and differences in heart disease risks , too.

People born female have less muscle and lower aerobic capacity in general. Also, women have more capillaries feeding part of their muscles, Shiroma says. The findings show women need to do less exercise to change their baseline of aerobic and muscular strength. "It might be that this relative increase in strength [in women compared to men] is what's driving this difference in benefit," he says.

The results show a little can go a long way. "The benefits start as soon as you start moving," Shiroma says.

It's increasingly common to see female weight lifters and body builders on social media, and many gyms and work-out studios now incorporate weight training into many of their classes and offerings. But, given that about 80% of women in the study said they don't participate in regular weight training, there's still a lot of hesitancy.

Ann Martin says her mood improves with resistance training. "It gets your blood flowing," she says. "It feels good." Deb Cutler/Ann Martin hide caption

"I was always the awkward one in gym class back in school days," says Ann Martin, 69, of Wilmington, Del. She shied away from gyms and weight-training machines. Martin has always been a walker, but she realized she needed to build more strength, so last year she started working out with a trainer to learn how to use the equipment. "It's fun now," she says. "I can feel my muscles getting stronger."

Strength training can be intimidating, Shiroma says. "But it's not all bodybuilders trying to lift super amounts of weight." He says there are many ways to incorporate resistance training into your life.

All activities that require your muscles to work against a weight or force count as strength training. This includes the use of weight machines, resistance bands or tubes, as well as all the many ways we can use our own body weight, as we do with push-ups and squats.

The findings of this new study fit with the Physical Activity Guidelines for Americans , which recommend that adults get a minimum of 2.5 hours of moderate-intensity exercise a week, that's about 30 minutes, most days of the week. The guidelines also call for doing strength-based activities at least two days a week.

This 22-Minute Workout Has Everything You Need

The increase in lifespan can likely be explained in part, by the well-being that comes from the other hidden benefits. Here are 5 ways building strength can boost good health.

1. Strength training helps protect joints.

Physical therapists often recommend resistance training for patients with knee and hip pain. "Strength training protects joints, resulting in less stress through the body," says Todd Wheeler, a physical therapist at MedStar Health Physical Therapy in Washington, D.C . "If joints could talk, they would say 'It's not my fault I'm irritated," Wheeler says. They'd blame it on weak muscles. He says strong muscles support the joints, which can help decrease joint pain. Wheeler recommends starting small and simply. For instance, try a few squats and table pushups. "Listen to your body and gradually increase intensity over time," he says.

2. Building muscle burns more calories

Aerobic exercise – such as running and cycling – typically burns more calories in real time compared to strength training. But people who weight train can get a boost in calorie burning over the long-term.

"When you're doing resistance training, you're building muscle. That muscle requires energy," says Bryant Johnson , a trainer who wrote The RBJ Workout . So, adding muscle mass can help people burn more calories.

Dr. Gulati also points to research that shows weight lifting and resistance training can help people lose more fat and improve body composition.

3. Resistance training protects against injuries and falls

As we've reported, millions of Americans, especially women, are under-muscled, and muscle mass is a predictor of longevity .

Since muscle mass peaks in our 30s and then starts a long, slow decline, we need to take steps to slow this down. If we don't do strength training exercise, we're more likely to become weak, increasing the risk of falls, which is the top cause of death among older adults in the U.S.

And since muscle loss - also known as sarcopenia - affects more than 45% of older adults in the U.S., "it's important to know about it and take steps to prevent it," says Richard Joseph , a wellness focused physician. He says strength training improves bone density which also protects against injuries and falls.

Joseph says people can get the biggest bang for their buck when they're starting out by focusing on lower body exercises that work big muscle groups in the legs.

4. Strength training helps control blood sugar

About 1 in 3 adults in the U.S. has prediabetes. Strength training can help control blood sugar by clearing glucose out of the bloodstream.

When we use our muscles during exercise, whether it's pushing, pulling, lifting or moving, they require more glucose for energy. This explains why exercise after meals can help control blood sugar.

And a recent study found strength training can be even more effective than aerobic activity in controlling blood sugar in people with diabetes.

5. Muscle building may help boost mood

A meta-analysis published in the medical journal JAMA Psychiatry in 2018, which included the results of more than 30 clinical trials, found a reduction in symptoms of depression among people who did weight training two times a week or more.

Strength training has also been shown to improve depressive symptoms in people at risk of metabolic disease. And, research shows strength training can tamp down anxiety, too.

Ann Martin says it makes sense that our moods improve when we move. "It gets your blood flowing," she says. "It feels good."

Scientists can tell how fast you're aging. Now, the trick is to slow it down

This story was edited by Jane Greenhalgh

• resistance training
• strength training
• weight training
• heart disease

More than one alcoholic drink a day raises heart disease risk for women

Young to middle-aged women who drink more than one alcoholic beverage a day, on average, were more likely to develop coronary heart disease than people who drink less, according to new research by Kaiser Permanente Northern California.

Women in the study who reported drinking eight or more alcoholic beverages per week were 33 to 51 percent more likely to develop coronary heart disease. And women who binge drink — three alcoholic beverages per day — were 68 percent more likely to develop coronary heart disease than those who drink in moderation, the research showed.

“There has been an increasing prevalence of alcohol use among young and middle-aged women as women may feel they’re protected against heart disease until they’re older, but this study shows that even in that age group, women who drink more than the recommended amount of one drink per day or tend to binge drink, are at risk for coronary heart disease,” Jamal Rana , a cardiologist with the Permanente Medical Group and the study’s lead author, wrote in an email.

The study will be presented at the American College of Cardiology’s Annual Scientific Session in early April. It was funded by the National Institutes of Health (NIH) and the National Institute on Alcohol Abuse and Alcoholism.

Risk is highest for binge drinking

The study used data from 432,265 adults, ages 18 to 65, who received care in the Kaiser Permanente Northern California integrated health organization. The group was composed of about 243,000 men and 189,000 women who filled out routine assessments between 2014 and 2015 in which they reported their alcohol intake. Researchers then looked at the coronary heart disease diagnoses among participants over the four years that followed.

Get Well+Being tips straight to your inbox

Participants were divided into three groups, according to their alcohol intake: low (one to two drinks per week), moderate (three to 14 drinks per week for men and three to seven drinks per week for women), or high (15 or more drinks per week for men and eight or more drinks per week for women).

Participants were also categorized as either engaging in binge drinking or not, with binge drinking being defined for men as having more than four drinks in a single day and for women as having more than three drinks a day, in the prior three months. Those who reported no alcohol use were not included.

During the four-year follow-up period, 3,108 participants were diagnosed with coronary heart disease. Higher levels of alcohol consumption were associated with a higher incidence of coronary heart disease. Both men and women who reported heavy episodic drinking, or binge drinking, had the highest risk.

The link between alcohol and coronary heart disease proved to be especially strong among women, the data showed.

Coronary heart disease, also known as coronary artery disease, is the most common type of heart disease in the United States, according to the Centers for Disease Control and Prevention . The disease develops when the arteries of the heart are unable to deliver enough oxygen-rich blood to the heart because of plaque buildup.

Heart disease is the No. 1 killer of American women , according to the CDC. Symptoms differ, and often there are none until people suffer from a heart attack or other problem, a NIH report said.

“There has long been this idea that alcohol is good for the heart, but more and more evidence is challenging that notion,” Rana wrote.

Alcohol is a risk factor for many health issues

Alcohol is actually a toxin to the heart, said Nieca Goldberg , a clinical associate professor of medicine at NYU Grossman School of Medicine and medical director of Atria New York City. Alcohol raises blood pressure , increases the risk for heart rhythm problems, especially during times of binge drinking, is associated with an enlarged heart and is a toxin to the heart muscle , she said.

“I think this is an important study to do because for a while, people thought that alcohol was protective against the heart because of earlier studies that were done in the past. But in fact, we don’t prescribe alcohol to fight heart disease,” she said.

Alcohol use is rising among women

The link between alcohol and heart disease for women is cause for concern, given that alcohol use among women is on the rise. While men used to drink more, studies over the past several years show that gap is closing.

Approximately 13 percent of adult women report binge drinking , with 25 percent of those women saying they do so at least weekly, on average, and 25 percent saying they consume at least six drinks during a binge drinking occasion, according to the CDC. A study in July in JAMA Network Open showed the number of alcohol-related deaths among women was rising at a faster rate than those among men, particularly for people 65 and older.

“I think this raises an important issue, because oftentimes, we think of heavy drinkers as men only. But we have to have heightened awareness that women may be heavy alcohol drinkers,” Goldberg said.

Occasional binge drinking can affect heart health

But it wasn’t just heavy drinkers who were affected, said Mary Ann McLaughlin , cardiologist at the Mount Sinai Fuster Heart Hospital. The study is interesting because it showed that even occasional drinking, if it reaches the level of binge drinking, can affect heart health, she said.

There are those who thought just drinking on the weekends was not a big deal, because they weren’t drinking every day, she said. “But the fact is, if they have more than four drinks as a woman or more than five drinks as a man on one day, in the past three months, they were at increased risk,” McLaughlin said.

Women are more adversely affected by alcohol

It is not a surprise that alcohol poses a higher risk for women than men when it comes to heart health, said C. Noel Bairey Merz , director of the Barbra Streisand Women’s Heart Center in the Smidt Heart Institute at Cedars-Sinai.

Women are more adversely affected than men by a lot of things such as cigarettes and pharmaceuticals and a bottle of beer or a glass of wine, where the dosage for men and women is the same and yet women are smaller, she said, referring to women being physically smaller, on average. Women also metabolize differently, their blood pressure is different, their liver function is different, they even deposit fat differently, Merz said.

“Women and men are built differently,” she said, noting she wasn’t even referring to the obvious reproductive differences. “It could be that in addition to body surface area … there are just pure biological differences in how the alcohol is metabolized.”

It’s possible to mitigate some of the ill effects of alcohol, the experts said. For instance, when people reduce or stop drinking, their blood pressure can improve and some lose weight as alcohol is a sugar that is no longer being consumed.

But issues such as enlargement of the heart happen with long-term heavy drinking, and even if the person stops, that may not improve, they said.

“If one stops drinking, some of the risk could reverse,” McLaughlin said. “The degree of improvement would depend on the age of the person and number of years of drinking.”

Read more from Well+Being

Well+Being shares news and advice for living well every day. Sign up for our newsletter to get tips directly in your inbox.

Vitamin B12 for fatigue has no proven benefit unless you have a deficiency that causes anemia.

Flashes, shimmers and blind spots: Here’s what migraine aura looks like.

Sparkling water is a better choice for your teeth than most popular beverages.

You can help your brain form healthy habits and break the bad ones.

Our 7 best tips to build an exercise habit .