motivation 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
• Wiley Open Access Collection

Motivation to learn: an overview of contemporary theories

David a cook.

1 Mayo Clinic Online Learning, Mayo Clinic College of Medicine, Rochester, Minnesota, USA

2 Multidisciplinary Simulation Center, Mayo Clinic College of Medicine, Rochester, Minnesota, USA

3 Division of General Internal Medicine, Mayo Clinic, Rochester, Minnesota, USA

Anthony R Artino, Jr

4 Division of Health Professions Education, Department of Medicine, Uniformed Services University of the Health Sciences, Bethesda, Maryland, USA

Associated Data

Table S2. A research agenda for motivation in education.

To succinctly summarise five contemporary theories about motivation to learn, articulate key intersections and distinctions among these theories, and identify important considerations for future research.

Motivation has been defined as the process whereby goal‐directed activities are initiated and sustained. In expectancy‐value theory, motivation is a function of the expectation of success and perceived value. Attribution theory focuses on the causal attributions learners create to explain the results of an activity, and classifies these in terms of their locus, stability and controllability. Social‐ cognitive theory emphasises self‐efficacy as the primary driver of motivated action, and also identifies cues that influence future self‐efficacy and support self‐regulated learning. Goal orientation theory suggests that learners tend to engage in tasks with concerns about mastering the content (mastery goal, arising from a ‘growth’ mindset regarding intelligence and learning) or about doing better than others or avoiding failure (performance goals, arising from a ‘fixed’ mindset). Finally, self‐determination theory proposes that optimal performance results from actions motivated by intrinsic interests or by extrinsic values that have become integrated and internalised. Satisfying basic psychosocial needs of autonomy, competence and relatedness promotes such motivation. Looking across all five theories, we note recurrent themes of competence, value, attributions, and interactions between individuals and the learning context.

Conclusions

To avoid conceptual confusion, and perhaps more importantly to maximise the theory‐building potential of their work, researchers must be careful (and precise) in how they define, operationalise and measure different motivational constructs. We suggest that motivation research continue to build theory and extend it to health professions domains, identify key outcomes and outcome measures, and test practical educational applications of the principles thus derived.

Short abstract

Discuss ideas arising from the article at www.mededuc.com discuss.

Introduction

The concept of motivation pervades our professional and personal lives. We colloquially speak of motivation to get out of bed, write a paper, do household chores, answer the phone, and of course, to learn. We sense that motivation to learn exists (as opposed to being a euphemism, intellectual invention or epiphenomenon) and is important as both a dependent variable (higher or lower levels of motivation resulting from specific educational activities) 1 and an independent variable 2 (motivational manipulations to enhance learning) 3 , 4 , 5 . But what do we really mean by motivation to learn, and how can a better understanding of motivation influence what we do as educators?

Countless theories have been proposed to explain human motivation. 6 Although each sheds light on specific aspects of motivation, each of necessity neglects others. The diversity of theories creates confusion because most have areas of conceptual overlap and disagreement, and many employ an idiosyncratic vocabulary using different words for the same concept and the same word for different concepts. 7 Although this can be disconcerting, each contemporary theory nonetheless contributes a unique perspective with potentially novel insights and distinct implications for practice and future research.

Previous reviews of motivation in health professions education have focused on practical implications or broad overviews without extended theoretical elaborations, 2 , 3 or focused on only one theory. 4 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 A review that explains and contrasts multiple theories will encourage a more nuanced understanding of motivational principles, and will facilitate additional research to advance the science in this field.

The purpose of this cross‐cutting edge article is to succinctly summarise five contemporary theories about motivation to learn, clearly articulate key intersections and distinctions among theories, and identify important considerations for future research. We selected these theories based on their presence in recent reviews; 6 , 17 , 18 , 19 we sought but did not find other broadly‐recognised modern theories. Our goal is not to present a comprehensive examination of recent evidence, but to make the theoretical foundations of motivation accessible to medical educators. We acknowledge that for each theory we can scarcely scratch the surface, and thus suggest further reading for those who wish to study in greater depth (see Table  1 ).

Summary of contemporary motivation theories

AT = attribution theory; EVT = expectancy‐value theory; GOT = goal orientation theory; SCT = social‐cognitive theory; SDT = self‐determination theory.

For this review we define motivation as ‘the process whereby goal‐directed activities are instigated and sustained’, 6 (pg 5) Although others exist, this definition highlights four key concepts: motivation is a process; it is focused on a goal; and it deals with both the initiation and the continuation of activity directed at achieving that goal.

Common themes

We have identified four recurrent themes across the five theories discussed below, and believe that an up‐front overview will help readers recognise commonalities and differences across theories. Table  1 offers a concise summary of each theory and Table  2 attempts to clarify overlapping terminology.

Similar concepts and terminology across several contemporary theories: clarifying confusable terminology

All contemporary theories include a concept related to beliefs about competence . Variously labelled expectancy of success, self‐efficacy, confidence and self‐concept, these beliefs all address, in essence, the question ‘Can I do it?’. However, there are important distinctions both between and within theories, as elaborated below. For example, self‐concept and earlier conceptions of expectancy of success (expectancy‐value theory) viewed these beliefs in general terms (e.g. spanning a broad domain such as ‘athletics’ or ‘clinical medicine’, or generalising across time or situations). By contrast, self‐efficacy (social‐cognitive theory) and later conceptions of expectancy of success viewed these beliefs in much more task‐ and situation‐specific terms (e.g. ‘Can I grade the severity of aortic stenosis?’).

Most theories also include a concept regarding the value or anticipated result of the learning task. These beliefs include specific terms such as task value, outcome expectation and intrinsic versus extrinsic motivation. All address the question, ‘Do I want do to it?’ or ‘What will happen (good or bad) if I do?’. Again, there are important distinctions between theories. For example, task value (expectancy‐value theory) focuses on the perceived importance or usefulness of successful task completion, whereas outcome expectation (social‐cognitive theory) focuses on the probable (expected) result of an action if full effort is invested.

Most theories discuss the importance of attributions in shaping beliefs and future actions. Learners frequently establish conscious or unconscious links between an observed event or outcome and the personal factors that led to this outcome (i.e. the underlying cause). To the degree that learners perceive that the underlying cause is changeable and within their control, they will be more likely to persist in the face of initial failure.

Finally, all contemporary theories of motivation are ‘cognitive’ in the sense that, by contrast with some earlier theories, they presume the involvement of mental processes that are not directly observable. Moreover, recent theories increasingly recognise that motivation cannot be fully explained as an individual phenomenon, but rather that it often involves interactions between an individual and a larger social context. Bandura labelled his theory a ‘social‐cognitive theory’ of learning, but all of the theories discussed below include both social and cognitive elements .

Again, each theory operationalises each concept slightly differently and we encourage readers to pay attention to such distinctions (using Table  2 for support) for the remainder of this text.

Expectancy‐value theories

In a nutshell, expectancy‐value theories 20 , 21 identify two key independent factors that influence behaviour (Fig.  1 ): the degree to which individuals believe they will be successful if they try (expectancy of success), and the degree to which they perceive that there is a personal importance, value or intrinsic interest in doing the task (task value).

Expectancy‐value theory. This is a simplified version of Wigfield and Eccles's theory; it does not contain all of the details of their theory and blurs some subtle but potentially important distinctions. The key constructs of task value and expectancy of success are influenced by motivational beliefs, which are in turn determined by social influences that are perceived and interpreted by learner cognitive processes

Expectancy of success is more than a perception of general competence; it represents a future‐oriented conviction that one can accomplish the anticipated task. If I do not believe I will be successful in accomplishing a task, I am unlikely to begin. Such beliefs can be both general (e.g. global self‐concept) and specific (judgements of ability to learn a specific skill or topic). According to Wigfield and Eccles, 20 expectancy of success is shaped by motivational beliefs that fall into three broad categories: goals, self‐concept and task difficulty. Goals refer to specific short‐ and long‐term learning objectives. Self‐concept refers to general impressions about one's capacity in this task domain (e.g. academic ability, athletic prowess, social skills or good looks). Task difficulty refers to the perceived (not necessarily actual) difficulty of the specific task. Empirical studies show that expectancy beliefs predict both engagement in learning activities and learning achievement (e.g. test scores and grades). In fact, expectancy of success may be a stronger predictor of success than past performance. 20

According to expectancy‐value theorists, however, motivation requires more than just a conviction that I can succeed; I must also expect some immediate or future personal gain or value. Like expectancy of success, task value or valence is perceived (not necessarily actual) and at times idiosyncratic. At least four factors have been conceived as contributing to task value: a given topic might be particularly interesting or enjoyable to the learner (interest or intrinsic value ); learning about a topic or mastering a skill might be perceived as useful for practical reasons, or a necessary step toward a future goal (utility or extrinsic value ); successfully learning a skill might hold personal importance in its own right or as an affirmation of the learner's self‐concept (importance or attainment value ); and focusing time and energy on one task means that other tasks are neglected (opportunity costs ). Other costs and potential negative consequences include anxiety, effort and the possibility of failure. For example, a postgraduate physician might spend extra time learning cardiac auscultation simply because he finds it fascinating, or because he believes it will help him provide better care for patients, or because he perceives this as a fundamental part of his persona as a physician. Alternatively, he might spend less time learning this skill in order to spend more time mastering surgical skills, or because he simply doesn't feel it is worth the effort. Although some evidence suggests that these four factors (interest, utility, importance and cost) are distinguishable from one another in measurement, 20 it is not yet known whether learners make these distinctions in practice. Task value is, in theory, primarily shaped by one motivational belief: affective memories (reactions and emotions associated with prior experiences). Favourable experiences enhance perceived value; unfavourable experiences diminish it.

The motivational beliefs that determine expectancy of success (goals, self‐concept and task difficulty) and task value (affective memories) are in turn shaped by life events, social influences (parents, teacher or peer pressure, professional values, etc.) and the environment. These shaping forces are interpreted through the learner's personal perspectives and perceptions (i.e. cognitive processes). It is perception, and not necessarily reality, that governs motivational beliefs.

Empirical studies (nearly all of them outside of medical education) show that both expectancy of success and value are associated with learning outcomes, including choice of topics to study, degree of involvement in learning (engagement and persistence) and achievement (performance). Task value seems most strongly associated with choice, whereas expectancy of success seems most strongly associated with engagement, depth of processing and learning achievement. 20 In other words, in choosing whether to learn something the task value matters most; once that choice has been made, expectancy of success is most strongly associated with actual success.

Attribution theory

Attribution theory (Fig.  2 ) explains why people react variably to a given experience, suggesting that different responses arise from differences in the perceived cause of the initial outcome. Success or failure in mastering a new skill, for example, might be attributed to personal effort, innate ability, other people (e.g. the teacher) or luck. These attributions are often subconscious, but strongly influence future activities. Failure attributed to lack of ability might discourage future effort, whereas failure attributed to poor teaching or bad luck might suggest the need to try again, especially if the teacher or luck is expected to change. Attributions directly influence expectancy of future success, and indirectly influence perceived value as mediated by the learner's emotional response to success or failure.

Attribution theory. This is a simplified version of Weiner's theory; it does not contain all of the details of his theory and blurs some subtle but potentially important distinctions. The process begins with an event; if the outcome is expected or positive, it will often directly elicit emotions (happiness or frustration) without any further action. However, outcomes that are unexpected, negative or perceived as important will often awaken the inquisitive ‘naïve scientist’ who seeks to identify a causal explanation. The individual will interpret the outcome in light of personal and environmental conditions to ‘hypothesise’ a perceived cause, which can be organised along three dimensions: locus, stability and controllability. Stability influences perceived expectancy of success. Locus, controllability and stability collectively influence emotional responses (which reflect the subjective value) and these in turn drive future behaviours

Attribution theory postulates that humans have a tacit goal of understanding and mastering themselves and their environment, and act as ‘naïve scientists’ to establish cause‐effect relationships for events in their lives. The process of attribution starts with an event, such as receiving a grade or learning a skill. If the result is expected and positive, the learner is content and the naïve scientist is not aroused (i.e. there is nothing to investigate). Conversely, if the result is negative, unexpected or particularly important, the scientist begins to search (often subconsciously) for an explanation, taking into account personal and environmental factors to come up with an hypothesis (i.e. an attribution: ability, effort, luck, health, mood, etc.). However, attributions do not directly motivate behaviour. Rather, they are interpreted or reframed into psychologically meaningful (actionable) responses. Empirical research suggests that such interpretations occur along three distinct conceptual dimensions: locus (internal to the learner or external), stability (likely to change or fixed) and controllability (within or outside the learner's control). For example, poor instructional quality (external locus) might be stable (the only teacher for this topic) or unstable (several other teachers available), and controllable (selected by the learner) or uncontrollable (assigned by others), depending on the learner's perception of the situation. Bad luck is typically interpreted as external, unstable and uncontrollable; personal effort is internal, changeable and controllable; and innate skill is internal, largely fixed and uncontrollable.

Weiner linked attributions with motivation through the constructs of expectancy of success and task value. 22 Expectancy of success is directly influenced by perceived causes, primarily through the stability dimension: ‘If conditions (the presence or absence of causes) are expected to remain the same, then the outcome(s) experienced in the past will be expected to recur. … If the causal conditions are perceived as likely to change, then … there is likely to be uncertainty about subsequent outcomes’. 22 Locus and controllability are not strongly linked with expectancy of success, because past success (regardless of locus orientation or degree of controllability) will predict future success if conditions remain stable.

By contrast, the link between attributions and ‘goal incentives’ (i.e. task value) is less direct, being mediated instead by the learner's emotions or ‘affective response’. Weiner distinguishes the objective value of achieving a goal (e.g. earning a dollar or learning a skill) from the subjective or affective value of that achievement (e.g. happiness or pride), and argues that there is ‘no blatant reason to believe that objective value is influenced by perceived causality … but [causal ascriptions] do determine or guide emotional reactions, or the subjective consequences of goal attainment’. 22 Other emotional reactions include gratitude, serenity, surprise, anger, guilt, hopelessness, pity and shame. Cognitive processes influence the interplay between an event, the perceived cause and the attributed emotional reaction, with complex and often idiosyncratic results (i.e. how we think influences how we feel). ‘For example, a dollar attained because of good luck could elicit surprise; a dollar earned by hard work might produce pride; and a dollar received from a friend when in need is likely to beget gratitude’, 22 although it might also beget shame or guilt. Weiner distinguishes outcome‐dependent and attribution‐dependent emotions. Outcome‐dependent emotions are the direct result of success (e.g. happiness) or failure (e.g. sadness and frustration). Attribution‐dependent emotions are, as the name implies, determined by the inferred causal dimension: pride and self‐esteem (‘internal’ emotions) are linked with locus; anger, gratitude, guilt, pity and shame (‘social’ emotions) are connected with controllability; and hopelessness and the intensity of many other emotions are associated with stability (i.e. one might feel greater gratitude or greater shame because of a stable cause).

Attribution theory proposes several ‘antecedent conditions’ that influence the attributional process. Environmental antecedents include social norms and information received from self and others (e.g. feedback). Personal antecedents include differences in causal rules, attributional biases and prior knowledge. Attributional biases or errors include: the ‘fundamental attribution error’, in which situation or context‐specific factors are ignored, such that a single event is extrapolated into a universal trait of the individual; self‐serving bias, in which success is ascribed to internal causes and failure is ascribed to external causes; and actor‐observer bias, in which the learner's actions are situation specific and the actions of others are a general trait.

Social‐cognitive theory

Social‐cognitive theory is most generally a theory of learning. It contends that people learn through reciprocal interactions with their environment and by observing others, rather than simply through direct reinforcement of behaviours (as proposed by behaviourist theories of learning). 23 As regards motivation, the theory emphasises that humans are not thoughtless actors responding involuntarily to rewards and punishments, but that cognition governs how individuals interpret their environment and self‐regulate their thoughts, feelings and actions.

Bandura 23 theorised that human performance results from reciprocal interactions between three factors (‘triadic reciprocal determinism’): personal factors (e.g. beliefs, expectations, attitudes and biology), behavioral factors, and environmental factors (both the social and physical environment). Humans are thus proactive and self‐regulating rather than reactive organisms shaped only by the environment; they are ‘both products and producers of their own environments and of their own social systems’. 24 Consider, for example, a medical student in a surgery clerkship that is full of highly competitive peers and is run by a physician with little tolerance of mistakes. Such an environment will interact with the student's personal characteristics (e.g. his confidence, emotions and prior knowledge) to shape how he behaves and whether or not he learns. At the same time, how he behaves will influence the environment and may change some of his personal factors (e.g. his thoughts and feelings). Thus, the extent to which this student is motivated to learn and perform is determined by the reciprocal interactions of his own thoughts and feelings, the nature of the learning environment and his actions.

The active process of regulating one's behaviour and manipulating the environment in pursuit of personal goals is fundamental to functioning as a motivated individual. Whether or not people choose to pursue their goals depends, in no small measure, on beliefs about their own capabilities, values and interests. 24 Chief among these self‐beliefs is self‐efficacy, defined as ‘People's beliefs about their capabilities to produce designated levels of performance that exercise influence over events that affect their lives’. 25 Self‐efficacy is a belief about what a person can do rather than a personal judgement about one's physical or psychological attributes. 26 In Bandura's words, ‘Unless people believe they can produce desired effects by their actions, they have little incentive to act’. 27 Thus, self‐efficacy forms the foundation for motivated action.

Unlike broader notions of self‐concept or self‐esteem, self‐efficacy is domain, task and context‐specific. For instance, a medical student might report fairly high self‐efficacy for simple suturing but may have much lower self‐efficacy for other surgical procedures, or might have lower self‐efficacy in a competitive environment than in a cooperative one.

Self‐efficacy should not be confused with outcome expectation – the belief that certain outcomes will result from given actions 18 (i.e. the anticipated value to the individual). Because self‐efficacy beliefs help to determine the outcomes one expects, the two constructs are typically positively correlated, yet sometimes self‐efficacy and outcome expectations diverge. For example, a high‐performing, highly efficacious college student may choose not to apply to the most elite medical school because she expects a rejection. In this case, academic self‐efficacy is high but outcome expectations are low. Research indicates that self‐efficacy beliefs are usually better predictors of behaviour than are outcome expectations. 26 , 27 Ultimately, however, both self‐efficacy and favourable outcome expectations are required for optimal motivation. 18

Bandura, Zimmerman and Schunk have identified the key role of self‐efficacy in activating core learning processes, including cognition, motivation, affect and selection. 6 , 25 , 28 , 29 Learners come to any learning task with past experiences, aptitudes and social supports that collectively determine their pre‐task self‐efficacy. Several factors influence self‐efficacy during the task (Fig.  3 ), and during and after the task learners interpret cues that further shape self‐efficacy. 27 Among these sources of self‐efficacy, the most powerful is how learners interpret previous experiences (so‐called enactive mastery experiences ). Generally speaking, successes reinforce one's self‐efficacy, whereas failures weaken it. In addition, learners interpret the outcomes of others’ actions ( modelling ). Learners may adjust their own efficacy beliefs based on such vicarious experiences, particularly if they perceive the model as similar to themselves (e.g. a near‐peer). The influence of verbal persuasion (‘You can do it!’) appears to be limited at best. Furthermore, persuasion that proves unrealistic (e.g. persuasion to attempt a task that results in failure) can damage self‐efficacy and lowers the persuader's credibility. Finally, physiological and emotional information shapes self‐efficacy beliefs: enthusiasm and positive emotions typically enhance self‐efficacy whereas negative emotions diminish it. 24 , 27

Social‐cognitive model of motivated learning. This is adapted from Schunk's model of motivated learning; it incorporates additional concepts from Bandura and other authors. Learners begin a learning task with pre‐existing self‐efficacy determined by past experiences, aptitudes and social supports. Learners can perform the task themselves or watch others (e.g. instructor or peer models) perform the task. During the task, self‐efficacy, together with other personal and situational factors, influences cognitive engagement, motivation to learn, emotional response and task selection. During and after the task, learners perceive and interpret cues that influence self‐efficacy for future tasks. Zimmerman defined a three‐phase self‐regulation cycle that mirrors this model, comprised of forethought (pre‐task), performance and volitional control (during task) and self‐reflection (after task)

One way in which social‐cognitive theory has been operationalised for practical application involves the concept of self‐regulation, which addresses how students manage their motivation and learning. Zimmerman proposed a model of self‐regulation 30 comprising three cyclical stages: forethought (before the task, e.g. appraising self‐efficacy, and establishing goals and strategies), performance (during the task, e.g. self‐monitoring) and self‐reflection (after the task). Self‐regulation is an area of active investigation in medical education. 14 , 15

Goal orientation (achievement goal) theories

The meaning of ‘goals’ in goal orientation theories 31 , 32 , 33 , 34 (also called achievement goal theory) is different from that in most other motivation theories. Rather than referring to learning objectives (‘My goal is to learn about cardiology’), the goals in this cluster of theories refer to broad orientations or purposes in learning that are commonly subconscious. With performance goals the primary concern is to do better than others and avoid looking dumb: ‘I want to get a good grade’. Mastery goals , by contrast, focus on the intrinsic value of learning (i.e. gaining new knowledge or skills): ‘I want to understand the material’. These broad orientations lead in turn to different learning behaviours or approaches. Dweck's theory of ‘implicit theories of intelligence’ takes these two orientations further, suggesting that they reflect learners’ underlying attributions (‘mindsets’, or dispositional attitudes and beliefs) regarding their ability to learn (Fig.  4 ).

Goal orientation theory and implicit theories of intelligence. This is a simplified illustration of Dweck's theory; it does not contain all of the details of her theory and blurs some subtle but potentially important distinctions. Learners tend toward one of two implicit self‐theories or mindsets regarding their ability. Those with an entity mindset view ability as fixed, and because low performance or difficult learning would threaten their self‐concept they unconsciously pursue ‘performance’ goals that help them to look smart and avoid failure. By contrast, those with an incremental mindset view ability as something to be enhanced with practice, and thus pursue goals that cause them to stretch and grow (‘mastery’ goals). Evidence and further theoretical refinements also support the distinction of performance‐approach goals (‘look smart’; typically associated with high performance) and performance‐avoidance goals (‘avoid failure’; invariably associated with poor performance)

Learners with performance goals have a (subconscious) self‐theory that intelligence or ability is a stable fixed trait (an ‘entity’ mindset). People are either smart (or good at basketball or art) or they're not. Because this stable trait cannot be changed, learners are concerned about looking and feeling like they have ‘enough’, which requires that they perform well. Easy, low‐effort successes make them feel smarter and encourage continued study; challenging, effortful tasks and poor performance are interpreted as indicating low ability and lead learners to progressively disengage and eventually give up. Learners with this entity mindset magnify their failures and forget their successes, give up quickly in the face of challenge, and adopt defensive or self‐sabotaging behaviours. A strong belief in their ability may lead them to persevere after failure. However, low confidence will cause them to disengage into a ‘helpless’ state because it is psychologically safer to blame failure on lack of effort (‘I wasn't really trying’) than on lack of intelligence. Dweck noted, ‘It is ironic that those students who are most concerned with looking smart may be at a disadvantage for this very reason’. 32

Learners with a mastery goal orientation, by contrast, have a self‐theory that intelligence and ability can increase or improve through learning (an ‘incremental’ mindset). People get smarter (or better at basketball or art) by studying and practising. This mindset leads people to seek learning opportunities because these will make them smarter. They thrive on challenge and even initial failure because they have an implicit ‘No pain, no gain’ belief. In fact, even learners with low confidence in their current ability will choose challenging tasks if they have an incremental mindset. Learners with an incremental mindset feel smart when they fully engage in learning and stretch their ability (the mastery goal orientation); easy tasks hold little or no value and failure is viewed as simply a cue to look for a better strategy and exert renewed effort.

Mindsets are related to the controllability and stability dimensions of attribution theory: entity mindsets lead to attributions of fixed and uncontrollable causes (e.g. ability), whereas incremental mindsets lead to attributions of controllable and changeable causes (e.g. effort). 31 , 35 Mindsets are typically a matter of degree, not black‐and‐white, and appear to be domain and situation specific: a learner might have predominantly entity beliefs about procedural tasks but incremental beliefs about communication skills. Mindsets change with age: young children typically have incremental mindsets, whereas most people have shifted toward entity mindsets by age 12. 32

Researchers building on the work of Dweck and others 33 , 36 , 37 have separated performance goals into those that make the learner look good (performance ‘approach’ goals such as trying to outperform others) and those in which the learner tries to avoid looking bad (performance ‘avoidance’ goals such as avoiding challenging or uncertain tasks). 38 , 39 Empirical results from real‐world settings differ for different outcomes: performance‐approach goals are consistently more associated with higher achievement (e.g. better grades) than are mastery goals, whereas mastery goals are associated with greater interest and deep learning strategies. These empirical observations require further explanation but could reflect shortcomings in mastery‐oriented study strategies (i.e. learners focus on areas of interest rather than studying broadly) or grading systems that favour superficial learning. 40 Performance‐avoidance goals, by contrast, are consistently associated with low achievement and other negative outcomes.

One of the most compelling findings of Dweck's theory is that the incremental mindset is teachable. Randomised trials demonstrate that teaching students that the brain is malleable and has limitless learning capacity leads them to seek more, and more difficult, learning opportunities and to persevere in the face of challenge. 32 The duration of this effect and its transfer to future tasks remain incompletely elucidated.

Unfortunately, the entity mindset also appears to be teachable, or at least unintentionally reinforced by individuals and learning climates that encourage competition, frame abilities as static or praise quick and easy success. Feedback intended to boost a learner's confidence (‘You did really well on that test; you must be really smart!’) may inadvertently encourage an entity mindset. Rather than emphasising innate ability, teachers should instill confidence that anyone can learn if they work at it.

Other motivation theories attempt to explain other aspects of goals, such as goal setting and goal content. 6 , 41 Goal orientation theories focus on the why and how of approach and engagement. Goal setting theories focus on the standard of performance, exploring issues such as goal properties (proximity, specificity and difficulty) and the factors that influence goal choice, the targeted level of performance and commitment. 42 Goal content theories focus on what is trying to be achieved (i.e. the expected consequences). Ford and Nichols 41 developed a content taxonomy of 24 basic goals that they categorised as within‐person goals (e.g. entertainment, happiness and intellectual creativity) and goals dealing with interactions between the person and environment (e.g. superiority, belongingness, equity and safety).

Self‐determination theory

Self‐determination theory (Fig.  5 ) posits that motivation varies not only in quantity (magnitude) but also in quality (type and orientation). Humans innately desire to be autonomous – to use their will (the capacity to choose how to satisfy needs) as they interact with their environment – and tend to pursue activities they find inherently enjoyable. Our highest, healthiest and most creative and productive achievements typically occur when we are motivated by an intrinsic interest in the task. Unfortunately, although young children tend to act from intrinsic motivation, by the teenage years and into adulthood we progressively face external (extrinsic) influences to do activities that are not inherently interesting. These influences, coming in the form of career goals, societal values, promised rewards, deadlines and penalties, are not necessarily bad but ultimately subvert intrinsic motivation. Strong evidence indicates that rewards diminish intrinsic motivation. 43 Deci and Ryan developed self‐determination theory to explain how to promote intrinsic motivation and also how to enhance motivation when external pressures are operative.

Self‐determination theory. This is adapted from Ryan and Deci's theory. Self‐determination theory hypothesises three main motivation types: amotivation (lack of motivation), extrinsic motivation and intrinsic motivation, and six ‘regulatory styles’ (dark‐background boxes). Intrinsic motivation (intrinsic regulation) is entirely internal, emerging from pure personal interest, curiosity or enjoyment of the task. At the other extreme, amotivation (non‐regulation) results in inaction or action without real intent. In the middle is extrinsic motivation, with four regulatory styles that vary from external regulation (actions motivated purely by anticipated favourable or unfavourable consequences) to integrated regulation (in which external values and goals have become fully integrated into one's self‐image). The transition from external to integrated regulation requires that values and goals become internalised (personally important) and integrated (fully assimilated into one's sense of self). Internalisation and integration are promoted (or inhibited) by fulfillment (or non‐fulfillment) of three basic psychosocial needs: relatedness, competence and autonomy

Intrinsic motivation is not caused because it is an innate human propensity, but it is alternatively stifled or encouraged by unfavourable or favourable conditions. Cognitive evaluation theory , a sub‐theory of self‐determination theory, proposes that fulfillment of three basic psychosocial needs will foster intrinsic motivation: autonomy (the opportunity to control one's actions), competence (self‐efficacy) and relatedness (a sense of affiliation with or belonging to others to whom one feels [or would like to feel] connected). Autonomy is promoted by providing opportunities for choice, acknowledging feelings, avoiding judgement and encouraging personal responsibility for actions. Rewards, punishments, deadlines, judgemental assessments and other controlling actions all undermine autonomy. Competence is supported by optimal challenge, and by feedback that promotes self‐efficacy (as outlined above) and avoids negativity. Relatedness is promoted through environments exhibiting genuine caring, mutual respect and safety.

In activities motivated by external influences, both the nature of the motivation and the resultant performance vary greatly. The motivation of a medical student who does his homework for fear of punishment is very different from motivation to learn prompted by a sincere desire to provide patients with optimal care. Deci and Ryan proposed that these qualitative differences arise because of differences in the degree to which external forces have been internalised and integrated (assimilated into the individual's sense of self). A second sub‐theory, organismic integration theory, explains these differences.

Organismic integration theory identifies three regulatory styles: intrinsic motivation at one extreme (highly productive and spontaneous), amotivation at the other extreme (complete lack of volition, failure to act or only going through the motions) and extrinsic motivation in between (actions prompted by an external force or regulation). Extrinsic motivation is divided, in turn, into four levels that vary in the degree to which the external regulation has been internalised (taking in a value or regulation) and integrated (further transformation of that regulation into their own self). 44 , 45 The lowest level is external regulation: acting only to earn rewards or avoid punishment. Next is introjected regulation: acting to avoid guilt or anxiety, or to enhance pride or self‐esteem. The regulation has been partially internalised but not accepted as a personal goal. Identified regulation suggests that the external pressure has become a personally important self‐desired goal, but the goal is valued because it is useful rather than because it is inherently desirable. Finally, with integrated regulation the external influences are integrated with internal (intrinsic) interests, becoming part of one's personal identity and aspirations. Regulatory forces with identified and integrated regulation reflect an internal locus of causality (control) and behaviours are perceived as largely autonomous or self‐determined, whereas both external and introjected regulation reflect an external locus of causality. ‘Thus, it is through internalisation and integration that individuals can be extrinsically motivated and still be committed and authentic.’ 45 Research suggests that the same three psychosocial needs described above promote the internalisation and integration of extrinsic motivations, with relatedness and competence being particularly important for internalisation, and autonomy being critical for integration.

Because optimal motivation and well‐being require meeting all three needs, ‘Social contexts that engender conflicts between basic needs set up the conditions for alienation and psychopathology’. 45 The importance of these needs has been confirmed not only in education, but also in workplace performance, patient compliance and overall health and well‐being. 46

Integration across theories

Over the past 25 years, contemporary motivation theories have increasingly shared and borrowed key concepts. 17 For example, all five theories discussed herein acknowledge human cognition as influencing perceptions and exerting powerful motivational controls. All also highlight reciprocal interactions between individuals and their socio‐environmental context. Definitions of expectancy have evolved to reflect substantial overlap with self‐efficacy. Attribution theory emerged from earlier expectancy‐value theories in an effort to explain the origins and antecedents (the ‘Why?’) of expectancies and values, ultimately emphasising the temporal sequence of events and the importance of emotions. Goal orientation theory merged early goal theories with the concept of implicit attributions. Self‐determination theory emphasises both autonomy (locus and control in attribution theory) and competence (very similar to self‐efficacy). With this conceptual overlap, it is easy to get confused with the terms as operationally defined within each theory. Table  2 attempts to clarify these areas of potential confusion.

Through this effort we have identified four recurrent themes among contemporary theories: competence beliefs, value beliefs, attribution and social‐cognitive interactions. We do not suggest that these theories can be reduced to these four concepts, but that these foundational principles underpin a more nuanced understanding of individual theories. Research conducted using one theoretical framework might also yield insights relevant to another.

Given the progressive blurring of boundaries and increasing conceptual overlap, can – or should – we ever achieve a grand unified theory of motivation? We note that each theory shines light on a different region of a larger picture, and thus contributes a unique perspective on a complex phenomenon involving individual learners and varying social contexts, topics and outcomes. Moreover, despite our and others’ efforts 7 , 47 to clarify terminology, conceptual differences among theories run much deeper than dictionary definitions can resolve. Even within a given theoretical domain, different investigators have operationally defined concepts and outcome measures with subtle but important distinctions that lead to vastly different conclusions. 31 , 37 , 39 The degree to which these differences can be both theoretically and empirically reconciled remains to be seen. 17 For now, we encourage maintaining theoretical distinctions while thoughtfully capitalising on overlapping concepts and explicit theoretical integrations for the enrichments they afford.

Implications and conclusions

Other authors have identified practical applications of motivation theory, most often instructional changes that could enhance motivation. 3 , 4 , 6 , 16 , 32 In Table S1 (available online) we provide a short summary of these suggestions, nearly all of which warrant investigation in health professions education. Educators and researchers will need to determine whether to apply these and other interventions to all learners (i.e. to improve the overall learning environment and instructional quality) or only to those with specific motivational characteristics (e.g. low self‐efficacy, entity mindsets, maladaptive attributions or external motivations). 17 , 48 , 49

We will limit our further discussion to considerations for future research. Pintrich 50 identified seven broad questions for motivation research and suggested general research principles for investigating these questions; we summarise these in Table S2 (available online). By way of elaboration or emphasis, we conclude with four broad considerations that cut across theoretical and methodological boundaries.

First, motivation is far from a unitary construct. This may seem obvious, yet both lay educators and researchers commonly speak of ‘motivation’ without clarity regarding a specific theory or conceptual framework. Although different theories rarely contradict one another outright, each theory emphasises different aspects of motivation, different stages of learning, different learning tasks and different outcomes. 17 , 19 , 51 To avoid conceptual confusion and to optimise the theory‐building potential of their work, we encourage researchers to explicitly identify their theoretical lens, to be precise in defining and operationalising different motivational constructs, and to conduct a careful review of theory‐specific literature early in their study planning.

Second, measuring the outcomes of motivation studies is challenging for at least two reasons: the selection of which outcomes (psychological constructs) to measure and the choice of specific instruments to measure the selected outcomes. The choice of outcomes and instruments, and the timing of outcome assessment, can significantly influence study results. For instance, results (and thus conclusions) for mastery and performance‐approach goal orientations vary for different outcomes. 39 Schunk identified four general motivation outcomes (choice of tasks, effort, persistence and achievement) and suggested tools for measuring each of these. 6 Learners can also rate how motivating they perceive a course to be. 52 The outcome(s) most relevant to a given study will depend on the theory and the research question. In turn, for each outcome there are typically multiple measurement approaches and specific instruments, each with strengths and limitations. For example, behaviour‐focused measures diminish the importance of cognitive processes, whereas self‐report measures are limited by the accuracy of self‐perceptions. For all instruments, evidence to support the validity of scores should be deliberately planned, collected and evaluated. 53 , 54

Third, researchers should test clear, practical applications of motivation theory. 50 , 55 , 56 Each of the theories discussed above has empirical evidence demonstrating theory‐predicted associations between a predictor condition (e.g. higher versus lower expectancy of success) and motivation‐related outcomes, but the cause‐effect relationship in these studies (often correlational rather than experimental) is not always clear. Moreover, the practical significance of the findings is sometimes uncertain; for example, does a change in the outcome measure reflect a meaningful and lasting change in the learner, or is it merely an artifact of the study conditions? Well‐planned experiments can strengthen causal links between motivational manipulations and outcomes. 57 We can find examples of interventions intended to optimise self‐efficacy, 28 task value, 5 attributions 17 and mindsets, 32 but research on motivational manipulations remains largely limited in both volume and rigour. 17 Moreover, moderating influences such as context (e.g. classroom, clinical or controlled setting) and learner experience or specialty can significantly impact results. Linking motivational concepts with specific cognitive processes may be instrumental in understanding seemingly inconsistent findings. 17 , 39 Finally, real‐world implementations of research‐based recommendations may be challenged by resource limitations, logistical constraints or lack of buy‐in from administrators and teachers; research on translation and implementation will be essential. 58

Lastly, we call for research that builds and extends motivation theory for education generally 50 and health professions education specifically. Theory‐building research should investigate ‘not only that the intervention works but also why it works (i.e., mediating mechanisms) as well as for whom and under what conditions (i.e., moderating influences)’. 17 Such research not only specifies the theoretical lens, interventions and outcomes, but also considers (and ideally predicts) how independent and dependent variables 2 interact with one another and with the topic, task, environment and learner characteristics. 59 Harackiewicz identified four possible relationships and interactions among motivation‐related variables:

• additive (different factors have independent, additive effects on a single outcome),
• interactive (different factors have complex effects on a single outcome),
• specialised (the impact of a given intervention varies for different outcomes) and
• selective (outcomes for a given intervention vary by situation, e.g. context or topic). 39

We encourage would‐be investigators to further explore theory‐specific literatures to understand conceptual nuances, current evidence, potential interactions, important outcomes and timely questions. 47 , 60

Only research grounded in such solid foundations will provide the theoretical clarity and empirical support needed to optimise motivation to learn in health professions education.

Contributors

DAC and ARA jointly contributed to the conception of the work, drafted the initial manuscript, revised the manuscript for important intellectual content and approved the final version. ARA is an employee of the US Government. The views expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the Uniformed Services University of the Health Sciences, Department of Defense, nor the US Government.

Conflicts of interest

the authors are not aware of any conflicts of interest.

Ethical approval

as no human subjects were involved, ethical approval was not required.

Supporting information

Table S1. Summary of practical applications of motivation theory.

Acknowledgments

we thank Kelly Dore for her contributions during the conceptual stages of this review and Adam Sawatsky and Dario Torre for their critiques of manuscript drafts.

The copyright line for this article was changed on 6 October 2016 after original online publication.

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here .

Loading metrics

Open Access

Peer-reviewed

Research Article

When research is me-search: How researchers’ motivation to pursue a topic affects laypeople’s trust in science

Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Validation, Visualization, Writing – original draft, Writing – review & editing

* E-mail: [email protected]

Affiliation Department of Psychology, Ludwig-Maximilians-Universität München, Munich, Germany

Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Resources

Roles Conceptualization, Supervision, Writing – review & editing

• Marlene Sophie Altenmüller, 
• Leonie Lucia Lange, 
• Mario Gollwitzer

• Published: July 9, 2021
• https://doi.org/10.1371/journal.pone.0253911
• Peer Review
• Reader Comments

Research is often fueled by researchers’ scientific, but also their personal interests: Sometimes, researchers decide to pursue a specific research question because the answer to that question is idiosyncratically relevant for themselves: Such “me-search” may not only affect the quality of research, but also how it is perceived by the general public. In two studies ( N = 621), we investigate the circumstances under which learning about a researcher’s “me-search” increases or decreases laypeople’s ascriptions of trustworthiness and credibility to the respective researcher. Results suggest that participants’ own preexisting attitudes towards the research topic moderate the effects of “me-search” substantially: When participants hold favorable attitudes towards the research topic (i.e., LGBTQ or veganism), “me-searchers” were perceived as more trustworthy and their research was perceived as more credible. This pattern was reversed when participants held unfavorable attitudes towards the research topic. Study 2 furthermore shows that trustworthiness and credibility perceptions generalize to evaluations of the entire field of research. Implications for future research and practice are discussed.

Citation: Altenmüller MS, Lange LL, Gollwitzer M (2021) When research is me-search: How researchers’ motivation to pursue a topic affects laypeople’s trust in science. PLoS ONE 16(7): e0253911. https://doi.org/10.1371/journal.pone.0253911

Editor: Lynn Jayne Frewer, Newcastle University, School of Natural and Environmental Sciences, UNITED KINGDOM

Received: December 4, 2020; Accepted: June 15, 2021; Published: July 9, 2021

Copyright: © 2021 Altenmüller et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: We provided all materials, the anonymized data and analyses, and supplementary materials online at the Open Science Framework via the following link: https://osf.io/phfq3/ .

Funding: The author(s) received no specific funding for this work.

Competing interests: The authors have declared that no competing interests exist.

Introduction

“Being a scientist is, at the most fundamental level, about being able to study what’s exciting to you”, says Jeremy Yoder, a gay man studying experiences of queer individuals in science [ 1 ]. Like Yoder, many researchers are passionate about their research and dedicated to their field. After all, they are free to choose research questions they deem important and are interested in. Freedom of science and research secures the independence of the academic from the political and other spheres. In return, researchers are expected to be neutral and objective and make their research process transparent to guarantee that this freedom is not exploited for personal gains.

Just as people differ in what they are interested in in their personal lives, researchers differ in what they find more or less fascinating and worth studying. Such fascination can have multiple causes and is often rooted in a perceived personal connection to a topic. For instance, Sir Isaac Newton allegedly became interested in gravity after an apple fell on his head [ 2 ]. A specific type of personal connection exists when researchers study a phenomenon because they are directly (negatively) affected by that phenomenon. In 1996, Harvard alumni and neuroanatomist Jill Bolte Taylor suffered a rare form of stroke that made her undergo major brain surgery, affected her personal and academic life tremendously, and eventually awakened her interest in studying the plasticity of the brain [ 3 ]. In 2006, she published an award-winning book covering her research and her personal story that led her to pursue this path. The Jill Bolte Taylor case is, thus, a prototypical example for such “me-search”: researchers studying a phenomenon out of a particular personal affection by (or connection to) this phenomenon. “Me-search” thus means pursuing a scientific question when the answer to that question is idiosyncratically relevant for the individual researcher (as opposed to when the answer is relevant per se).

Being directly affected by a phenomenon provides researchers studying it with a high degree of expertise and motivation: Jill Bolte Taylor, for instance, claims to bring a deep personal understanding and compassion to her research and work with patients [ 4 , 5 ]. That said, being personally affected may also come at the cost of losing one’s scientific impartiality and neutrality for the subject: Jill Bolte Taylor was criticized for being overly simplistic in her scientific claims and mixing them with esoteric ideas, and for pushing her own agenda (i.e., selling her story) by dramatizing her own experiences [ 4 – 7 ].

While some criticized Jill Bolte Taylor heavily, the general public does not seem to have a problem with her research as “me-search”. Her book is currently translated into 30 languages, and thousands of people visit her talks and keynote addresses [ 4 – 6 ]. Does that suggest that the general public tends to turn a blind eye on conflicts of interest that may arise from a researchers’ personal affection by their research object? While the Jill Bolte Taylor case seems to suggest so, research on science communication and public understanding of science has shown that people are highly sensitive to potential conflicts of interest arising from researchers’ personal involvement: perceiving researchers as pursuing an “agenda” for personal reasons is a major factor predicting people’s loss of trust in researchers and science [ 8 – 11 ]. On the other hand, people may see personal (“autoethnographic”) experiences of researchers personally affected by their topic as valuable and laudable ‒ it may imply that “they know what they’re talking about” [ 12 – 14 ]. Similarly, revealing a personal interest or even passion for a particular research topic (e.g., due to being personally affected) could also overcome the stereotypical perception of scientists as distant “nerds in the ivory tower” [ 15 , 16 ]: researchers who openly disclose the idiosyncratic relevance of their research topic may appear more approachable, more likeable, and more trustworthy [ 17 – 19 ].

Thus, the public’s reaction to “me-search” seems to be ambivalent and contingent on certain boundary conditions. Thus, the question we are going to address in this article is whether and when ‒ that is, under which circumstances ‒ a researcher’s personal affection by a research topic (“me-search”) positively vs. negatively impacts public perceptions regarding the trustworthiness of the respective researcher (and the entire research area in general) and the extent to which this researcher’s findings are perceived as credible .

Perceivers’ motivated stance as a moderating variable

This potentially ambivalent perception of “research as me-search” can be understood from a motivated reasoning [ 20 ] perspective: Laypeople receive and process information in a manner biased towards their own beliefs, expectations, or hopes. This also applies to the reception of scientific information [ 21 , 22 ]: For example, laypeople are more likely to dismiss scientific evidence if it is inconsistent with their beliefs [ 23 , 24 ] or if it threatens important (moral) values [ 25 , 26 ] or their social identity, respectively [ 27 – 29 ].

However, identity-related and attitudinal motivated science reception might differ in their underlying mechanisms. For identity-related motivated science reception, biased perception of information, which is relevant to a social identity, is driven by a defense motivation to protect this positive social identity [ 30 ]. Thus, identity-threating scientific information is countered by identity-protection efforts, such as discrediting the findings and the source. These efforts will be more pronounced among strongly identified individuals [ 27 – 29 ]. For attitudinal motivated science reception, however, the mechanism might function as a broader perception filter. When confronted with new scientific information about the respective attitude object, the perceptual focus will be directed at clues helping to uphold prior attitudes (i.e., confirmation bias [ 31 ]): Potentially attitude-inconsistent information is attenuated, while potentially attitude-consistent information is accentuated. The ambivalent nature of “me-search” might allow to be easily bend in such a motivated manner and, thus, lead to biased perceptions of a researcher either way: when the findings are in line with one’s prior beliefs, being personally affected may be considered an asset–the respective researcher is perceived as more trustworthy and his/her findings as more credible (compared to no idiosyncratic relevance). However, when the findings are inconsistent with one’s prior beliefs, idiosyncratic relevance may be considered a flaw–the respective research is perceived as biased, untrustworthy, and less competent, and his/her findings are likely perceived as less credible than when idiosyncratic relevance is absent.

Prior research on motivated science reception mainly focused on laypeople’s reactions towards specific scientific findings: after learning about the outcome of a particular study, participants dismiss the research (and devalue the researcher) if these outcomes are consistent vs. inconsistent with their prior beliefs [ 23 – 25 , 27 – 29 ]. However, people might be prone to motivated science reception even before results are known, judging researchers proverbially just by their cover (e.g., by biographical data, personal and scientific interests and motivations). People who hold positive attitudes towards a certain research topic might perceive “me-searchers” as more trustworthy and anticipate their results to be more credible (before knowing the specific outcomes). By contrast, people who hold negative attitudes towards a certain research topic they might trust “me-searchers” less and expect their findings to be less credible.

Additionally, motivated reception processes can be extended over and above the specific information under scrutiny and lead to questioning the scientific method in itself–a phenomenon termed the “scientific impotence excuse” [ 32 ]. In line with that, critical evaluations of specific researchers and their findings are sometimes generalized to the entire field of research [ 27 ]. Thus, the fact that a researcher engages in “me-search” might be interpreted in a way that fits best to one’s preexisting convictions and may generalize to the entire field of research.

The present research

In two studies, laypeople read alleged research proposals concerning potentially polarizing research topics (i.e., LGBTQ issues and veganism) which were submitted by researchers who disclosed being either personally affected or not affected by the respective topic. We investigated whether ( Study 1 ) and when (i.e., moderated by preexisting positive attitudes towards the respective research topic, Studies 1 and 2) such “me-search” information increased or decreased laypeople’s perceptions regarding these researchers’ epistemic trustworthiness and the anticipated credibility of their future scientific findings. Of note, we use the term “credibility” to differentiate evidence-related trust/credibility from person-related trust/credibility (i.e. “trustworthiness”). Further, we test whether one researcher’s “me-search” impacts the evaluation of the entire respective field ( Study 2 ).

For both studies in this paper, we report how we determined our sample size, all data exclusions (if any), all manipulations, and all measures [ 33 ]. All materials, the anonymized data, and analyses are available online at the Open Science Framework (OSF; see https://osf.io/phfq3/ ). Before starting the respective study, informed consent was obtained. Participants read a GDPR-consistent data protection and privacy disclosure declaration specifically designed for the present study. Only participants who gave their consent could start the respective survey. According to German laws and ethical regulations for psychological research [ 34 ], gathering IRB approval is not necessary if (i) the data are fully anonymized, (ii) the study does not involve deception, (iii) participants’ rights (e.g., voluntary participation, the right to withdraw their data, etc.) are fully preserved, and (iv) participating in the study is unlikely to cause harm, stress, or negative affect. The present studies met all of these criteria; therefore, no IRB approval had to be obtained.

In our first study, we conducted an online experiment investigating the main effect of a researcher’s disclosure of being personally affected vs. not affected by their research on their trustworthiness and the credibility of their future research. Further, we tested whether laypeople’s preexisting attitudes towards the research topic moderate this effect.

Four-hundred and eleven German participants were recruited via mailing lists and social networks. Ninety-seven participants had to be excluded due to pre-specified criteria: Sixty-seven participants failed the manipulation check; 25 participants failed the pre-specified time criteria (viewing the manipulation stimulus less than 30sec, taking less than 3min or more than 20min for participation); 5 participants had apparently implausible response patterns (e.g., “straight-lining;” identical responses on every single item on more than one questionnaire page in a row). Eighty-five further participants failed the attention check. Excluding them did not change the overall results, so, for the sake of statistical power, we did not exclude these 85 cases. The final sample consisted of N = 314 participants. We conducted sensitivity analyses using G*Power [ 35 ] for determining which effect sizes can detected with this sample in a moderated (multiple) regression analysis: At α = 0.05 and with a power of 80%, small-to-medium effects (f 2 ≥0.03) can be detected with this sample. Participants were mostly female (74% female, 25% male, 2% other) and their age ranged between 16 and 68 years ( M = 26.79; SD = 10.18). Most participants were currently studying at a university (71%; working: 21%; unemployed or other: 8%). Participants who were currently studying or already had a university degree (93%) came from a variety of disciplines (law, economics, and social sciences: 49%; humanities: 16%; mathematics and natural sciences: 14%; medicine and life sciences: 11%; engineering: 4%).

Materials and procedure.

After obtaining informed consent, we asked participants to imagine they were browsing the website of a research institute and came across a short proposal for a new research project by a researcher named Dr. Lohr (no gender was indicated for greater generalizability and avoiding possible gender confounds). Next, participants read the beginning of the alleged proposal of a planned research project for which Dr. Lohr was allegedly applying for external funding. The text briefly introduced the planned project (i.e., investigating social reactions to queer employees at the workplace) and a statement of Dr. Lohr explaining why they were interested in conducting this project. Participants were randomly allocated to two groups. In the “not personally affected” condition, Dr. Lohr wrote:

“ I am interested in investigating this research topic in more detail not only out of scientific reasons but also because I–as someone who does not identify as homosexual and is not affected by my own research–really think we need more evidence-based knowledge about queer topics which we can implement in everyday life .”

In the “personally affected” condition, Dr. Lohr wrote:

“ I am interested in investigating this research topic in more detail not only out of scientific reasons but also because I–as someone who identifies as homosexual and is affected by my own research–really think we need more evidence-based knowledge about queer topics which we can implement in everyday life .”

We added a definition for the word “queer” below the proposal: “ Queer is a term used as self-description by people who do not identify as heterosexual and/or who do not identify with the gender assigned at birth . The term is often used as umbrella term for LGBTQ (lesbian , gay , bisexual , trans and queer) and describes all people who identify as queer .” After completing an attention check question (see pre-registration), we measured participants’ trust in Dr. Lohr with the Muenster Epistemic Trustworthiness Inventory (METI; [ 36 ]), which was constructed for measuring trust in experts encountered online. It consists of 14 opposite adjective pairs measuring an overall trustworthiness score (Cronbach’s α = .95) as well as the sub-dimensions expertise (e.g., competent–incompetent, Cronbach’s α = .92) and integrity/benevolence (e.g., honest–dishonest, Cronbach’s α = .93) on 6-point bipolar Likert scales. Factor analyses (see Appendix A in the supplementary materials, https://osf.io/phfq3/ ) suggest that a two-factor model (with expertise and integrity/benevolence) fit the data better than a three-factor model (as suggested by [ 36 ]), corroborating the idea of a cognitive-rational dimension and an affective dimension of trustworthiness [ 37 ]. Next, participants rated the extent to which they found Dr. Lohr’s research credible on a 6-point Likert scale ranging from 1 = “not at all” to 6 = “very much” (6 items, e.g., “I think Dr. Lohr’s future findings will be credible;” “I will be critical of Dr. Lohr’s research results” (reverse-coded); Cronbach’s α = .84).

Next, we measured participants’ own positive attitudes towards LGBTQ issues—the moderator variable in our design—with eleven statements developed from research on sympathy, group attitudes, and allyship [ 38 , 39 ] rated on a 6-point Likert scale ranging from “not at all” to “very much” (e.g., “I think that LGBTQ-related topics receive more attention than necessary” (reverse-coded); “I am open to learning more about concerns raised by LGBTQ people;” Cronbach’s α = .93). Next, we conducted a manipulation check by asking participants to indicate whether Dr. Lohr disclosed being personally affected by their research (“Dr. Lohr stated being personally affected;” “Dr. Lohr stated not being personally affected;” “Dr. Lohr did not say anything about being affected or not”).

Finally, we measured demographic variables (age, gender, occupation, academic discipline) and control variables: general perceptions of researchers’ neutrality (self-developed 6-point bipolar scale with 4 adjective-pairs, e.g. subjective–objective, and 6 distractor pairs, e.g. introverted–extraverted, Cronbach’s α = .81) and Public Engagement with Science (PES) with two measures adapted from a survey by the BBVA Foundation [ 40 ]: a 5-item scale measuring PES frequency (e.g., “How often do you read news about science?” 5-point Likert scale ranging from 0 =“never” to 5 =“almost daily,” Cronbach’s α = .78) and a multiple choice question measuring 15 potential PES experiences during the last 12 months (e.g., “I know someone who does scientific research;” “I visited a science museum”). Participants had the opportunity to participate in a lottery and sign up for more information and were debriefed.

Our randomized groups did not differ in regard to general perception of neutrality in science ( p = .924) or PES (PES frequency, p = .709; PES experiences, p = .533). Table 1 summarizes all means, standard deviations, correlations and internal consistencies of the measured variables.

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

https://doi.org/10.1371/journal.pone.0253911.t001

Main effect of being personally affected.

First, we tested the main effect of the researcher’s disclosure of being personally affected on epistemic trustworthiness and credibility of future findings. Laypeople trusted Dr. Lohr significantly more in the “personally affected” condition ( M = 4.92, SD = 0.75) than in the “not personally affected” condition ( M = 4.66, SD = 0.81), t (312) = 2.93, p = .004, d = 0.33, 95% CI d [0.11; 0.56]. For credibility, the difference between the “personally affected” condition ( M = 4.15, SD = 0.96) and the “not personally affected” condition ( M = 4.04, SD = 0.86) was not significant, t (312) = 1.02, p = .306, d = 0.12, 95% CI d [-0.11; 0.34]. Further exploring the two dimensions of epistemic trustworthiness, Dr. Lohr was perceived as higher on integrity/benevolence, t (312) = 3.19, p = .002, d = 0.36, 95% CI d [0.14; 0.59], and on expertise, t (312) = 2.17, p = .030, d = 0.25, 95% CI d [0.02; 0.47] when disclosing being personally affected.

Moderation by pre-existing attitudes.

Second, we tested whether the effect of being personally affected by the research topic on trustworthiness was moderated by participants’ pre-existing attitudes towards LGBTQ issues. Using standardized linear regression, we again found a main effect of condition on trustworthiness, beta = 0.15, p = .004, 95% CI beta [0.05, 0.26]. There was a significant main effect of participants’ pre-existing attitudes, beta = 0.30, p < .001, 95% CI beta [0.20, 0.40] and the condition × attitudes interaction effect was significant, beta = 0.19, p < .001, 95% CI beta [0.08, 0.29], increasing the amount of explained variance in trustworthiness by 3% to R 2 adj = .14. Table 2 summarizes the results. Fig 1A displays the interaction effect and standardized simple slopes analysis further qualifies it: Participants with more positive attitudes towards LGBTQ issues (+1 SD above sample mean) trusted Dr. Lohr more when the researcher was personally affected vs. not affected, beta = 0.34, p < .001, 95% CI beta [0.20, 0.49]. For participants with less positive attitudes towards LGBTQ issues (-1 SD below sample mean), this effect appears to be reversed, yet the simple slope was not significant, beta = -0.03, p = .646, 95% CI B [-0.18, 0.11]. The same pattern of interaction effects emerged for both, integrity/benevolence ( p = .009, total R 2 adj = .14) and expertise ( p < .001, total R 2 adj = .10); full analyses are reported in Appendix B (see https://osf.io/phfq3/ ).

Linear regression plots for the interaction effect of attitudes × condition on epistemic trustworthiness (Fig 1A) and credibility (Fig 1B) with 95% confidence intervals: Participants’ attitudes towards the research topic moderated how a researcher’s disclosure of being personally affected (vs. being not personally affected) by one’s own research was perceived.

https://doi.org/10.1371/journal.pone.0253911.g001

https://doi.org/10.1371/journal.pone.0253911.t002

Regarding our second dependent variable, credibility, we found no main effect of condition, beta = 0.04, p = .456, 95% CI beta [-0.06, 0.13]. However, there was a significant main effect of participants’ pre-existing attitudes, beta = 0.48, p < .001 95% CI beta [0.39, 0.58]: Participants with more positive attitudes anticipated a higher credibility of future research findings in this condition than participants with less positive attitudes. Similar to epistemic trustworthiness, there was a significant condition × attitudes interaction effect, beta = 0.21, p < .001, 95% CI beta [0.12, 0.31], increasing the amount of explained variance in credibility by 4% to R 2 adj = .26. Table 2 summarizes the results. Fig 1B displays this interaction effect: Again, participants with more positive attitudes towards LGBTQ issues (+1 SD above sample mean) anticipated Dr. Lohr’s future research findings to be more credible when the researcher was personally affected vs. not affected, beta = 0.25, p < .001, 95% CI beta [0.12, 0.38]. However, for participants with more negative attitudes (-1 SD below sample mean) this effect was significantly reversed: They rated the future research as less credible when the researcher was personally affected vs. not affected, beta = -0.18, p = .009, 95% CI B [-0.31, -0.04].

Results from Study 1 suggest that LGBTQ researchers are perceived as more trustworthy and their future findings as more credible when they disclose being personally affected by their research topic (i.e., being queer themselves), but only if perceivers hold positive attitudes towards LGBTQ issues. By contrast, holding less favorable attitudes towards LGBTQ issues lead to more skeptical reactions towards personally affected vs. unaffected researchers. This finding shows that learning about a researcher’s personal affection by their research can, indeed, go both ways, as suggested by our theoretical reasoning. On a more general level, our research suggests that public reactions towards “me-search” is a matter of pre-existing attitudes, and, thus, a case of motivated science reception [ 21 , 22 ].

There are some limitations to this first study: As most people in our sample held rather positive attitudes towards the LGBTQ community ( M = 4.93, SD = 1.02; on a scale from 1 to 6), predicted values on trustworthiness and credibility at the lower end of the attitude spectrum are probably less reliable. Also, we did not control for participants’ own identification as belonging to the LGBTQ community. Thus, we cannot differentiate clearly between attitudinal and identity-related effects, which is important because attitudes and identity concerns have a psychologically distinguishable impact on motivated science reception [ 27 , 28 ]. Additionally, replicating our results in a different domain is necessary to be able to generalize our findings. Another question of generalizability that is left unanswered is how such individual experiences with one personally affected researcher might impact laypeople’s perception of the entire field. This calls for more research on the double-edged nature of the moderating effect of preexisting attitudes.

In our preregistered second study (see https://osf.io/c9r4e ), we aimed to replicate our findings in a more diverse sample and with a different research topic that has the potential of polarizing participants even more strongly. We used the same design as in Study 1, but changed the proposed research topic to perceptions of vegans and introduced a vegan vs. non-vegan researcher. Again, we hypothesized that laypeople’s attitudes towards veganism moderate the effects on trustworthiness as well as credibility of future research. Additionally, we tested whether the effect of one researcher being personally affected by their own research generalizes to the broader perception of their entire field. Furthermore, we also explored whether the moderation by attitudes towards veganism prevailed when controlling for self-identification as being vegan (not included in preregistration).

We conducted an a-priori power analysis using G*Power [ 35 ] for detecting the hypothesized interaction effect in a moderated multiple regression analysis ( f 2 = 0.04, based on Study 1, with 1- β = 0.90 and α = 0.05, which resulted in a total sample of N = 265. Anticipating exclusions (see specified criteria) of comparable size as in the previous study, we aimed for a sample of at least 350 participants.

We collected data from 364 participants via mailing lists and social media. Fifty-seven participants had to be excluded due to our preregistered criteria (see https://osf.io/c9r4e ): one participant was younger than 16 years, 31 failed the manipulation check, 10 took less than 20sec viewing the proposal, 12 took less than 3min or more than 20min for participation, 3 had apparently implausible patters of response (i.e., “straight-lining;” identical responses on every single item on more than one questionnaire page in a row). The final sample consisted of N = 307 participants (76% female, 23% male, 1 other) who were between 18 and 79 years old ( M = 33.55, SD = 13.92). Approximately half of the sample (50%) was currently studying at a university, further 40% were working and 10% not working, one person was currently in training. Eighty-five percent were currently studying or already held a university degree (social sciences: 49%, humanities: 17%, natural sciences: 14%, life sciences: 8%, engineering: 6% and other 6%). Most participants did not consider themselves as vegans (89%).

We used the same materials and procedure as in Study 1 (see OSF for full materials: https://osf.io/phfq3/ ). However, we changed the research topic to “perceptions of vegans”. Participants were randomly assigned to two conditions. In the “not personally affected” condition, the researcher Dr. Lohr wrote:

“ I was interested in investigating this research questions not only out of scientific reasons but because , as someone who is not living as a vegan and , thus , not personally affected by my own research , I think we have a need for more evidence-based knowledge regarding the social embedding of vegan lifestyles , which we can acknowledge in everyday life .”

“… because , as someone who is living as a vegan and , thus , personally affected by my own research , I think we have a need for more evidence-based knowledge regarding the social embedding of vegan lifestyles , which we can acknowledge in everyday life . ”

As dependent variables, we again used the 14-item METI [ 36 ] to measure epistemic trustworthiness, but we expanded the measure for credibility of future research by adding one more item (“I would express skepticism towards Dr. Lohr’s future findings”) to better capture the behavioral aspects of credibility (now: 7 items; Cronbach’s α = .86). We also added a measure of participants’ evaluation of the entire field (not the specific researcher) as a third dependent variable. This 12-item scale was adapted from a related study [ 28 ] (e.g., “I think researchers who do research on that topic sometimes lack competence,” “I think it is difficult to apply results from this line of research to reality;” 6-point Likert scale ranging from 1 = “not at all” to 6 = “very much;” Cronbach’s α = .85). Next, participants’ attitudes towards veganism (i.e., the moderator variable) were measured with a 14-item scale adapted from the attitude measure in Study 1 by changing and adding items (e.g., “I think veganism is exaggerated” (reverse-coded) and “I can imagine being a vegan myself;” 6-point Likert scale ranging from 1 = “not at all” to 6 = “very much;” Cronbach’s α = .95).

To reduce exclusions after data collection, participants could proceed only if they answered all attention checks correctly (4 items; multiple choice). We added self-identification as vegan as a control variable (“Do you presently consider yourself a vegan?” yes/no); and an open-ended question about participants’ opinion regarding the researcher being personally affected to explore how laypeople rationalize their opinion. These responses were later coded for valence (positive, negative, mixed, or neutral) and content (deductive and inductive coding) by two raters blind to the specific research question (see Appendix C in the supplementary materials, https://osf.io/phfq3/ ; interrater reliability for valence, Cohen’s κ = .86, p < .01; and for content, Cohen’s κ = .74, p < .01). Again, the questionnaire closed with a sign-up for a lottery and more information as well as a debriefing.

Our randomized groups did not differ in regard to PES (PES frequency, p = .147; PES experiences, p = .101). However, they did differ significantly in regard to the general perception of neutrality in science ( p = .049). Possible implications are addressed in the Discussion. Table 3 summarizes all means, standard deviations, correlations and internal consistencies. In the following, we report our findings for all three dependent variables (trustworthiness, credibility, evaluation of the entire field), consecutively.

https://doi.org/10.1371/journal.pone.0253911.t003

Trustworthiness.

First, we ran the standardized regression model for epistemic trustworthiness. There was neither a significant main effect of condition on epistemic trustworthiness, beta = 0.04, p = .482, 95% CI beta [-0.07, 0.15] nor a significant main effect of attitudes towards veganism, beta = 0.07, p = .205, 95% CI beta [-0.04, 0.18]. However, the hypothesized condition × attitudes interaction effect was significant, beta = 0.22, p < .001, 95% CI beta [0.11, 0.34], increasing the amount of explained variance in trustworthiness by 4% to R 2 adj = .05. Table 4 summarizes the results. Fig 2A and standardized simple slopes analyses show that participants with more positive attitudes towards veganism (+1 SD above sample mean) trusted Dr. Lohr more when personally affected vs. not affected, beta = 0.26, p = .001, 95% CI beta [0.11, 0.42]. This conditional effect was reversed for participants with more negative attitudes (-1 SD below sample mean), who trusted Dr. Lohr less when personally affected vs. not affected, beta = -0.19, p = .020, 95% CI beta [-0.34, -0.03]. The interaction effect remained significant when controlling for participants’ self-identification as being vegan ( p < .001, total R 2 adj = .06). In secondary analyses, we explored the effects on the two facets of epistemic trustworthiness, separately. The same pattern of interaction effects emerged for both integrity/benevolence ( p < .001, total R 2 adj = .08) and expertise ( p = .005, total R 2 adj = .02); full analyses are reported in Appendix D in the supplementary materials (see https://osf.io/phfq3/ ).

Linear regression plots for the interaction effect of attitudes × condition on epistemic trustworthiness (Fig 2A), credibility (Fig 2B) and critical evaluation of the entire field (Fig 2C) with 95% confidence intervals: Participants’ attitudes towards the research topic moderated how a researcher’s disclosure of being personally affected (vs. being not personally affected) by one’s own research was perceived.

https://doi.org/10.1371/journal.pone.0253911.g002

https://doi.org/10.1371/journal.pone.0253911.t004

Credibility.

On credibility, there was no significant main effect of condition, beta = -.07, p = .146, 95% CI beta [-0.17, 0.03] but a significant main effect of attitudes towards veganism, beta = .35, p < .001, 95% CI beta [0.25, 0.45]. As predicted, the condition × attitudes interaction effect was also significant for credibility, beta = 0.25, p < .001, 95% CI beta [0.15, 0.35], increasing the amount of explained variance in credibility by 6% to R 2 adj = .21. Table 4 summarizes these results. Fig 2B and standardized simple slope analyses qualify the interaction effect: In line with the results for trustworthiness, participants with more positive attitudes (+1 SD above sample mean) anticipated Dr. Lohr’s future findings to be more credible when personally affected vs not affected, beta = 0.18, p = .016, 95% CI beta [0.03, 0.32], while the conditional effect for participants with more negative attitudes (-1 SD below sample mean) changed its sign, beta = -0.32, SE ( B ) = 0.14, p < .001, 95% CI beta [-0.47, -0.18]. As before, the interaction effect remained significant when controlling for self-identification as being vegan ( p < .001, total R 2 adj = .21).

Evaluation of the field.

Third, we investigated whether this moderation effect generalizes to the evaluation of the entire field of veganism research. There was no significant main effect of condition, beta = -.00, p = .989, 95% CI beta [-0.10, 0.10] but a significant main effect of attitudes, beta = -.41, p < .001, 95% CI beta [-0.51, -0.31]. Again, we found the hypothesized condition × attitude interaction effect, beta = -.27, p < .001, 95% CI beta [-0.37, -0.18], increasing the amount of explained variance in critical evaluation by 7% to R 2 adj = .27. Again, Table 4 summarizes these results and Fig 2C and standardized simple slopes analyses further qualify the interaction effect: Participants with more positive attitudes towards veganism (+1 SD above sample mean) were less critical of research on veganism when Dr. Lohr was personally affected vs. not affected, beta = -0.28, p < .001, 95% CI beta [-0.41, -0.14]. By contrast, this conditional effect was reversed for participants with more negative attitudes towards veganism (-1 SD below sample mean), beta = 0.27, p < .001, 95% CI beta [0.14, 0.41]. This interaction effect also remained significant when controlling for self-identification as being vegan ( p < .001, total R 2 adj = .28).

Participants’ opinion.

Overall, participants who responded to the open-ended question expressed mostly negative opinions about the researcher being personally affected by his own research (negative: 48%, neutral: 21%, positive: 17%, and mixed: 14%). The most frequently mentioned (negative) remark was that a “me-searcher” might be biased towards their research (60%; e.g., “ By introducing himself as being affected , I fear he cannot evaluate the results of his research objectively ”). The second most frequently mentioned remark was that such idiosyncratic relevance is irrelevant (24%; e.g., “ It wouldn’t make a difference ”). Positive remarks were mentioned less frequently: Participants ascribed more motivation (11%; e.g. “ I think interest , also personal interest , is an important prerequisite for determined research ”) or knowledge about the topic (8%; e.g. “ Very good , most likely , he thus is knowledgeable about the subject matter and can conduct the study in a more purposeful manner ”) to the “me-searcher”, or recognized the transparency (7%; e.g., “ The main thing is transparency . People are always biased , perhaps even unconsciously ”; for more details, see Appendix C in the supplementary materials: https://osf.io/phfq3/ ).

In Study 2, we replicated the moderation effect of preexisting attitudes on the effect of a researcher disclosing being personally affected (vs. not affected) by their own research on participants’ epistemic trustworthiness and credibility ascriptions regarding the research and researcher’s future findings. Further, we showed that this effect generalizes to the evaluation of the entire research area. Here, positive attitudes towards veganism determined how learning about an openly vegan researcher impacted participants’ perceptions of trustworthiness and credibility as well as the evaluation of the entire field of veganism research compared to learning about a non-vegan (i.e., non-affected) researcher. Participants who held more positive attitudes towards veganism reported more trust, higher anticipated credibility of future findings, and a less critical evaluation of the field when confronted with a vegan researcher. Conversely, for participants with less positive attitudes this effect was reversed. The moderation by positive attitudes towards veganism persisted when controlling for participants’ self-identification as vegans. Overall, the interaction effects observed in Study 2 explained similar amounts of variance as in Study 1 (epistemic trustworthiness: 3% vs. 4%, and credibility: 4% vs. 6%). Further, qualitative analyses revealed that most participants reported negative–or, at least, mixed–perceptions of a “me-searcher” (e.g., “me-searchers” may be biased, but also highly motivation and knowledgeable), which corroborated our theoretical prediction that “me-search” may be a double-edged sword. Interestingly, these qualitative findings seem somewhat contradictory to the quantitative findings, according to which there was no main effect of researchers’ idiosyncratic affection by their research topic.

In Study 2, one caveat is that the groups differed significantly in regard to participants’ general expectations of neutrality in science. Participants who read about the personally affected researcher had weaker expectations of neutrality; yet, when added to the regression model as a control, the pattern of results remained unchanged (see Appendix E in the supplementary materials, https://osf.io/phfq3/ ). Further, as a second caveat, we show that participants generalized their perceptions to the overall field of veganism research. However, this research area might be considered quite narrow and, thus, future research should investigate how far such generalization processes stretch out to perceptions of broader areas of research (e.g., health psychology).

General discussion

In two studies, we show that laypeople’s perception of researchers who disclose being personally affected by their own research can be positive as well as negative: The effect of such “me-search” was moderated by laypeople’s preexisting attitudes. Queer or vegan researchers were perceived as more trustworthy and their future findings were anticipated to be more credible when participants had positive, sympathizing attitudes towards the related research object (i.e., LGBTQ community or veganism). When participants’ attitudes were less positive, this pattern reversed. In Study 2, we extended our research from individualized perceptions of single researchers and their findings to evaluations of the entire field of research. Participants who were confronted with a personally affected researcher seemed to consider this person a representative example and generalized their judgment to their evaluation of the entire (though here quite narrow) research area.

We explored epistemic trustworthiness in more detail in both studies, namely the cognitive-rational facet of expertise and the affective facet of integrity/benevolence: Both were impacted by researchers’ disclosure of being personally affected, although effect sizes for expertise were descriptively smaller than for integrity/benevolence. This points to “me-search”–when received positively–possibly adding to the perception of competence-related aspects like a deeper knowledge of a phenomenon (e.g., via anecdotal insights) [ 12 – 14 ] and, even more so, warmth-related aspects like seeming more sincere, benevolent, transparent and, thus, approachable [ 15 , 16 , 41 ]. Disclosing such personal interest in a scientific endeavor might be able to bridge the stereotypical perception of cold and distant “science nerds” by revealing passionate, human and, thus, more relatable side of a researcher. When received negatively, however, “me-search” might be regarded as harboring vested interests, which casts doubts on a researcher’s neutrality and objectivity [ 8 – 11 , 42 ].

In general, the main models tested here explained between 5% and 28% of variance which may not appear impressive at first glance. However, our studies posed a very strict test of the effects of “me-search” by only using a subtle manipulation sparse in information followed by measures of very specific perceptions which might have contributed to an understatement of the real-world impact.

“Me-search” neither automatically sparks trust nor mistrust in laypeople, even if their explicit opinions seem rather negative. In line with assumptions from motivated science reception [ 22 , 43 ], our findings suggests that the ambivalence of the fact that a researcher is personally affected can be seized as an opportunity to interpret the situation in a manner that best fits to preexisting attitudes: Researchers, their findings and even their entire field of research are evaluated–even before learning about specific findings–based on prior attitudes towards the research topic. We show in Study 2 that the moderation effect of participants’ positive attitudes towards the respective research topic (i.e., veganism) prevails when controlling for self-identification with the topic (i.e., being a vegan). This suggests that, indeed, in motivated reasoning attitudinal and identity-related processes can be differentiated: Here, social identity protection could be ruled out as alternative explanation for the effects of pre-existing attitudes. Noteworthily, we demonstrate that motivated science reception already operates when the results are not (yet) known. This points towards a perceptual filter made up of pre-existing attitudes that is activated when confronted with scientific information and leads to biased pre-judgments: Ambivalent cues (i.e., “me-search”) are prematurely interpreted in line with prior attitudes without actually knowing whether the new scientific information will be attitude-consistent or inconsistent (when, later, results are reported).

Future research

Future research on the motivated reception of “me-search” should focus on three open questions. First, while we consider it a strength of our studies that the results of the proposed research project were not yet known, it might be interesting to see how being personally affected or not interacts with the perceived direction of the communicated scientific results (e.g. supporting vs. opposing a certain position): To what extent can the first, premature evaluation of a “me-searching” researcher be adapted if the actual results are inconsistent with this pre-judgment?

Second, the investigation of what specific characteristics of “me-search” are instrumentalized by benevolent or skeptical perceivers might not only provide practical tips on how to handle being personally affected (e.g., in science communication) but also important theoretical insights on the building blocks of trust in science and researchers (see discussion above regarding the effects on the facets of epistemic trustworthiness). As one example, knowing that a qualitative level of knowledge is highly valued could further research on the trust-benefit of enriching statistical evidence with anecdotal and narrative elements [ 44 , 45 ]. As second example, we argue that researchers’ self-disclosure of being personally affected by their research might signal transparency and, thus, improve the perception of the trust facets integrity and benevolence. Yet, even the disclosure of not being personally affected could have such an effect on a researcher’s reputation and, at the same time, it might be less ecologically valid (as, presumably, it is rather unusual to explicitly state to not be affected by something). Introducing a control group without any information about a researcher’s relation towards their research object might bring light to this.

Third, we demonstrated the moderation effect of preexisting attitudes for two research areas (i.e., LGBTQ and veganism) and in different populations. Yet, further research should investigate whether this effect will hold up for other areas, more diverse samples and different kinds of “me-search”, as well. For example, in some research fields being personally affected by the research might be perceived as more morally charged than in others and, thus, having stronger polarizing effects [ 46 ]: While, in veganism-research, “me-search” might be grounded in an ideological choice (e.g., thinking its morally wrong to consume animal products and, thus, being vegan), having a stroke and, following, studying stroke-related brain plasticity is likely perceived as less ideological. Also, different scientific methods (typically) used in a field might impact the perceptions of “me-search” depending on how prone for subjectivity these methods are perceived to be (e.g., qualitative “me-search” like autoethnographic analyses might be perceived more critically than when using seemingly objective, quantitative methods like physiological measures). Further, researchers who are not directly personally affected by their research but “merely” interested in something for personal reasons (e.g., being highly empathetic towards queer concerns without identifying as queer) might not profit from disclosure of such personal motivations: Such researchers might be perceived as impostors [ 47 ] lacking the expertise stemming from directly firsthand experiences.

Practical implications

Finally, for the applied perspective on public engagement with science, it should again be noted, that motivated reasoning processes are activated even before specific results are presented (e.g. before hearing a talk or reading about a study). This might be important, as judgments are quickly formed and remembered [ 48 , 49 ] and, therefore, the first impression of a researcher might set the tone for further interactions and, particularly, for the acceptance and implementation of their findings. This emphasizes the importance of researchers knowing their audience (and their attitudes) when engaging in science communication.

Of course, there are also ethical considerations concerning “me-search”: Researchers should always declare any conflict of interests when conducting research [ 50 , 51 ]. Failing to disclose being personally affected by one’s own research might backfire severely on researchers’ reputation–especially concerning their trustworthiness and the credibility of their findings–and in particular, when this information is disclosed by someone else and not themselves. At least for achieving positive reputational effects, it seems researchers need to freely initiate the disclosure of limitations and problems themselves [ 41 , 52 ]. A possible solution for reaping all the benefits and protecting against the potential harms of engaging in “me-search” might be to actively seek out mixed research teams. Including affected as well as non-affected individuals in research projects might be worth considering from the stance of the public’s trust in science: It enables deep, even personal insights to the studied phenomenon, while still securing balanced perspectives and impartiality.

Neuroanatomist Jill Bolte Taylor became famous for turning her “stroke of fate” into productive and well-selling “me-search”. Yet, she was praised as well as heavily criticized for mixing her personal and scientific motivations: When research is also “me-search”, it can be perceived positively as well as negatively depending on laypeople’s preexisting attitudes towards the research object. Researchers who disclose being personally affected by their own research can benefit from this disclosure in terms of trustworthiness and credibility when it is perceived by laypeople with positive attitudes; however, for audiences with more negative attitudes this effect is reversed and disclosure can be harmful. One experience with a personally affected researcher might be enough to impact the evaluation of the whole field. Thus, openly acknowledging “me-search” in one’s research is an ambivalent matter and its communicative framing as well as the targeted audience should be well considered.

• 1. Hu JC. Studying the science of science. Science | AAAS [Internet]. 2016 Mar 23 [cited 2020 Sep 7]; Available from: https://www.sciencemag.org/careers/2016/03/studying-science-science
• View Article
• Google Scholar
• 3. Taylor JB. My Stroke of Insight. Hachette UK; 2009. 140 p.
• 4. Kaufman L. A Superhighway to Bliss. The New York Times [Internet]. 2008 May 25 [cited 2020 Oct 28]; Available from: https://www.nytimes.com/2008/05/25/fashion/25brain.html
• PubMed/NCBI
• 33. Simmons JP, Nelson LD, Simonsohn U. A 21 Word Solution [Internet]. Rochester, NY: Social Science Research Network; 2012 Oct [cited 2020 Jul 23]. Report No.: ID 2160588. Available from: https://papers.ssrn.com/abstract=2160588
• 34. German Psychological Society, German Association of Psychologists. Berufsethische Richtlinien des Berufsverbandes Deutscher Psychologinnen und Psychologen e.V.und der Deutschen Gesellschaft für Psychologie e.V. 2016.
• 40. Foundation BBVA. International Study on Scientific Culture. Understanding of Science. Bilbao, Spain: BBVA; 2011.
• 48. Ames DL, Fiske ST, Todorov AT. Impression formation: A focus on others’ intents. In: Decety J, Cacioppo JT, editors. The Oxford handbook of social neuroscience. New York, NY, US: Oxford University Press; 2011. p. 419–33. (Oxford library of psychology).