Secondary Logo

Journal Logo


Automatic Affective Evaluations of Physical Activity

Conroy, David E.1,2; Berry, Tanya R.3

Author Information
Exercise and Sport Sciences Reviews: October 2017 - Volume 45 - Issue 4 - p 230-237
doi: 10.1249/JES.0000000000000120
  • Free

Key Points

  • Interventions to promote physical activity frequently rely on effortful cognitive processes that are burdensome.
  • Pleasant affective processes can provide proximal intrinsic rewards that motivate physical activity.
  • Automatic affective evaluations of physical activity (“gut reactions”) are distinct from reflective attitudes toward physical activity.
  • People with more positive automatic affective evaluations of physical activity are more active.
  • Automatic affective evaluations are modifiable and represent a promising new target for interventions to increase physical activity levels.


Initiating physical activity can be difficult. Maintaining regular activity is even more challenging. Fewer than 5% of U.S. adults attain even 30 min of daily moderate-vigorous physical activity. People often experience physical activity — particularly exercise — as unpleasant, and unpleasant experiences during activity are associated with reduced engagement (21,46). Unfortunately, interventions to promote physical activity are based frequently on theories that de-emphasize affective processes in favor of cognitive processes involving self-monitoring, goals, feedback, and beliefs (about oneself and the behavior). The behavior change techniques used to target these motivational processes have demonstrated positive effects (40,41), but they often are effortful and impose a burden on people. This burden has become the Achilles’ heel of many contemporary public health strategies because it can decrease engagement and compromise long-term effectiveness. More importantly, these behavior change techniques ignore the importance of affective processes in regulating physical activity. In this article, we challenge the assumption that people are willing and able to maintain physical activity motivation using effortful cognitive processes alone.

An alternative mechanism of physical activity that does not require onerous assumptions about effortful self-regulation involves automatic motivational processes. We propose the hypothesis that automatic affective evaluations of physical activity play a role in regulating physical activity that is unique from the influence of reflective processes such as self-monitoring, goal-setting, behavioral feedback, and various beliefs about oneself or a behavior. As such, automatic affective evaluations offer promising new targets for physical activity interventions.

Automatic affective evaluations of physical activity reflect the affective experiences that arise rapidly and involuntarily when the concept of physical activity is activated in a person’s mind. They are based on associations learned over time and through experience with physical activity. In the context of the affect and health behavior framework, automatic affective evaluations are most similar to a form of automatic affective processing known as implicit attitudes (50). They reflect the immediate affective evaluations of a target — in this case, physical activity behavior — and can derive from direct or vicarious experience with that target (e.g., feelings evoked during physical activity or when observing others in physical activity). Unlike reflective affective associations, automatic affective evaluations occur rapidly and effortlessly; they do not require conscious processing or elaboration (35). They can influence both automatic motivation (e.g., prompting goal pursuit without conscious awareness) and reflective affective processes (e.g., anticipated affect or affective attitudes). These processes are thought to be the upstream determinants of affectively charged motivation for physical activity and other health behaviors.

This article introduces dual-process models that can be applied to physical activity motivation to establish the function of automatic affective evaluations. Measures of automatic evaluations and literature linking those measures with physical activity are reviewed. Finally, individual- and social-level intervention strategies to modify automatic affective evaluations of physical activity are introduced to frame opportunities for applying recent advances in this area.


The overarching feature of dual-process models is the identification of two systems that operate in parallel to regulate thoughts, feelings, and actions (20,23,27,34,48). System 1 is grounded in associations learned over time. It is relatively fast, effortless, and automatic. System 1 involves constructs such as habits and automatic evaluations. This system is sometimes referred to as an impulsive or automatic system.

In contrast, system 2 involves rule-based processing. It is relatively slow, effortful, and volitional. System 2 processes are exemplified by social-cognitive theories that invoke constructs such as intentions, efficacy beliefs, outcome expectations, and plans (e.g., theory of planned behavior, health action process approach, self-efficacy theory). This system is sometimes referred to as a reflective or controlled system. These systems have been integrated in dual-process theories.

Although system 2 processes have featured prominently in physical activity research, system 1 processes have attracted increased attention over the past decade. A recent review summarized evidence linking physical activity with a variety of system 1 processes including habits and nonconscious goal pursuit (44). That review also examined the response latency measures used to assess a variety of system 1 processes, including automatic affective evaluations, and concluded that the measures were too heterogeneous to draw meaningful conclusions about the measures. We take a different, construct-focused perspective here by focusing our review on automatic affective evaluations and excluding other constructs measured with similar tools. We further provide a theory-based lens to help the field move forward. We focus specifically on the role of system 1 (automatic) affective evaluations because pleasant affective experiences can provide the immediate rewards needed to expend effort toward initiating, and possibly maintaining, physical activity (22,50).

Affective evaluations form the basis for peoples’ attitudes about objects or behaviors. Attitudes reflect whether a person likes or dislikes the behavior and, consequently, whether a person capitalizes on or bypasses opportunities to enact that behavior (13). The specific mechanisms linking affective evaluations of behavior with action vary between the two systems. For example, in system 2, positive affective evaluations often are expected to strengthen intentions for physical activity which in turn increases the likelihood of the intended physical activity (1). Some also argue that affective responses can influence behavior independent of reasoning about that behavior (30). The rest of this review focuses on the role of system 1 affective evaluations in regulating physical activity. As background, we highlight dual-process theories that propose system 1 mechanisms for affective processes as well as relations with system 2 affective processes. A number of models have been applied to explain social relationships (e.g., racism) or personality as a function of associations between concepts held in memory that automatically are activated when an attitude object (e.g., exercise) is encountered. We emphasize two models that have been used in physical activity research and can provide insight into the role of automatic affective evaluations in regulating physical activity.

First, the reflective-impulsive model focuses on evaluative and semantic associations that are activated by a cue or stimulus (20). These associations are the basis for affective responses to stimuli within the approach- or avoidance-oriented impulsive system (i.e., system 1). The reflective system (system 2) is thought to elaborate those automatic associations into propositions (wherein a “truth” value to the association is assigned after deliberation). This system is important for intentions and goal achievement. The reflective system can also “overcome habitual responses or [put] together actions plans in new situations when habits fail” (p. 73) (20). According to the reflective-impulsive model, no process is entirely reflective or impulsive, and these two systems interact to influence behavior. In general, the reflective-impulsive model charts how initial inputs are categorized, leading to factual and evaluative decisions and ultimately to a behavioral decision and intentions to act on that decision (“I’m going to the gym”). Schemata are an important construct within the reflective-impulsive model and are activated through spreading activation in associative networks.

Second, the associative-propositional evaluation model posits automatic affective responses that align with the emotional valence, be it positive or negative, of an encountered object (27). This model involves automatically activated associative pathways created through learning or motivational states stored in memory. They are automatically activated by external inputs (i.e., a cue in the environment), and explicit attitudes or similar constructs reflect the degree to which automatic associations are considered “true.” For example, if hearing about “exercise” automatically activates a feeling of discomfort (system 1), this feeling can turn into the proposition “exercise is not fun,” with which one can explicitly agree or disagree (system 2). Different associations can be activated by the same object or concept such that “exercise” may be associated with attributes such as fun, discomfort, disease risk, appearance, or weight, depending on preexisting memory structures and the nature of the cue that activated the association (e.g., friends wanting you to join them in a pick-up soccer game or a health pamphlet at the doctor’s office warning of the consequences of inactivity). The associative-propositional evaluation model further proposes that changes in the associative structure may occur through evaluative conditioning or a change in pattern activation (27). At the reflective (system 2) level, changes can occur because of a change at the associative level, alterations in propositions needing to be evaluated (e.g., because of new information), or the use of different strategies to achieve consistency about a set of propositions. Thus, the associative-propositional evaluation model proposes that automatically activated affective associations from system 1 provide the starting point for cognitive elaboration of system 2 attitudes.

System 1 and system 2 evaluations (attitudes) of a target will not always converge and, in some cases, may be discordant. The systems of evaluation model provides insight into how people respond in this situation (39). In general, system 2 evaluations respond relatively quickly to counter-attitudinal information, whereas system 1 evaluations take longer to develop. A person who evaluates physical activity favorably with system 1 (i.e., a favorable automatic affective evaluation) might be presented with information from a peer that a specific group exercise class is too demanding. This person’s system 1 evaluation of physical activity is unlikely to change in response to this information, but her or his system 2 evaluation could quickly become less favorable. In this case, which system takes over to guide affect, behavior, and cognition? In the systems of evaluation model, evaluative discordance between systems leads to feelings of ambivalence, weaker attitudes, and detracts from well-being (39). Thus, discordance between system 1 and system 2 evaluation has implications for behavior. The two systems are expected to regulate behavioral outcomes that correspond to the memory structures from which they originate (48). System 1 is based in associations so it will likely influence well-learned, proceduralized actions and more spontaneous, unplanned actions. On the other hand, system 2 is based on rules and propositions so it will likely influence novel actions that require explicit monitoring and volitional, planned actions. In the context of physical activity, system 1 will most likely influence light-intensity and lifestyle physical activities, whereas system 2 will most likely influence planned exercise behavior (or at least intentions that rely on declarative memory processes).

Figure 1 summarizes the general conceptual model that links contextual cues to automatic affective evaluations and physical activity within a dual-process framework. The figure is generic in some respects because a number of theories could be plugged in to the reflective portion of the figure. The theory of planned behavior is used here because it allows us to highlight the role of attitudes across the two systems. Similar constructs (e.g., values, outcome expectancies) appear in most, if not all, theories based on expectancy-value logic or social-cognitive assumptions. Readers should be able to consider explicit constructs within a number of psychosocial models without difficulty.

Figure 1
Figure 1:
Conceptual dual-process model of physical activity highlighting the role of automatic affective evaluations. The theory of planned behavior was selected to illustrate a system 2 model, but any motivation theory based on expectancy-value logic or social-cognitive processes could be used as well.

Over time, experiences accumulate and people create memories based on their idiosyncratic associations between the concept of physical activity and different affective experiences. These memories can be characterized by the valence and strength of the constituent associations (2). When a person is exposed to cues that activate the concept of physical activity, relevant memories are activated and activation can spread to closely related nodes in the network (e.g., characteristic affective experiences). Concepts with strong automatic affective evaluations have been shown to activate regions of the brain associated with emotions. For example, images of black males have been shown to coactivate brain areas associated with fear (e.g., amygdala) among people with stronger automatic anti-black affective evaluations (43,49). These affective experiences may not exceed a threshold to be noticeable, but they can nevertheless suffice to instantiate basic attentional biases and approach-avoidance tendencies toward or away from the target of the evaluation (16). In effect, without necessarily experiencing overt pleasure or displeasure, a “gut feeling” can arise that may tip the scales toward or away from physical activity. With apologies to Sir Isaac Newton, we propose that people at rest tend to stay at rest unless compelled to change, and the impulse of these subtle affective forces can be sufficient to disrupt such behavioral inertia. In other words, automatic affective evaluations of physical activity can potentially deliver an impulse that increases the likelihood of moving the momentarily inactive person.


Measures of implicit cognition have been applied to assess individual differences in the accessibility of implicit attitudes or automatic affective evaluations in system 1. These measures tend to be indirect (i.e., implicit) and often performance based. In contrast, system 2 attitudes and related cognitions are assessed typically with explicit methods (i.e., questionnaires, interviews) (47).

A number of implicit measures have been used to assess automatic evaluations of physical activity. These include the extrinsic affective Simon task, evaluative priming tasks, impulsive approach tendency tasks, and variants of the Implicit Association Test (12,15,16,19). All use response latencies to infer the strength of automatic evaluative processes. To illustrate this family of implicit methods, consider the oldest and most commonly used method, the evaluative priming task (25). In this test, participants sit at a computer, fixate their gaze at a point in the center of the screen, and place their left and right index fingers on buttons they will use to indicate their responses. At the beginning of each trial, the fixation point is replaced with a priming stimulus, such as an image or word representing physical activity or a neutral image or word. These priming presentations may be either supraliminal (slow enough for conscious perception; e.g., >200 ms) or subliminal (too rapid for conscious perception; e.g., <100 ms). Next, a target affective stimulus — either pleasant or unpleasant — is presented for a longer period of time (e.g., 1000 ms), and the participant must select and initiate the appropriate response to indicate the valence of that stimulus.

Response times are recorded over many trials and averaged separately for the pleasant and unpleasant stimuli. Positive automatic evaluations of physical activity are inferred from relatively faster responses to pleasant stimuli after being primed with the concept of physical activity. Negative automatic evaluations of physical activity are inferred from relatively faster responses to unpleasant stimuli after being primed with the concept of physical activity. Figure 2 depicts the conceptual model underlying each trial on this test. The theoretical premise is that the priming procedure activates the concept of physical activity and that activation spreads to associated affective experiences involving either pleasure or displeasure. This spreading activation increases the accessibility of those affective concepts and can accelerate responses to concordant stimuli (response facilitation), interfere with responses to discordant stimuli (response inhibition), or both. The degree to which such facilitation and inhibition occurs indexes individual differences in automatic affective evaluations.

Figure 2
Figure 2:
Conceptual model for a trial in the evaluative priming task commonly used to measure automatic affective evaluations. Constructs and paths in black represent discrimination response times based on the information-processing demands required to identify an affective stimulus. Constructs and paths in gray represent additional response time variance due to evaluative priming.

The specific procedures and scoring models vary for different implicit measures. Most scoring models yield either a single score reflecting the valence of automatic affective associations with physical activity or separate scores reflecting the level of pleasure or displeasure that automatically is associated with physical activity (although more complex approaches based on information processing models have also been proposed) (45). The distinctions between scoring models are not trivial because the conceptual model underlying response latencies directly influences the scope of hypothesizing that is possible. For simplicity, our analysis reduces automatic affective evaluations of physical activity to a single dimension representing a resultant approach-avoidance motivational tendency.

Theoretically, the implicit measures described previously are sensitive to responses that occur before there is time for conscious reflection or deliberation. One frequently cited advantage of implicit measures is they overcome the issue of socially desirable responses. This advantage may be overstated because responses to implicit tasks can be faked or modified, albeit to a lesser degree than explicit measures (26). Social desirability also did not moderate the relationship between automatic associations and explicitly measured motives related to exercise (9). Implicit measures also do not necessarily reveal the “true self” as they are, by and large, measures of associations stored in memory rather than what one truly believes (26). The psychometric properties of popular scores from these measures have also been criticized via questions regarding how much an observed score reflects a true score, how random error distributions impact extreme scores, and whether the relationship is linear or nonlinear (10,11). A related argument is that Implicit Association Test scores have been given meaning without any empirical rationale to identify cut points that indicate some level of preference or bias. These critiques focused on research using implicit scores to study racial bias but warrant careful consideration in research on physical activity as well.

Notwithstanding those critiques, implicit measures of system 1 processes can be very useful in predicting aspects of physical activity that are not regulated by system 2 processes because they capture affective responses that arise before one reflects on feelings or the possible consequences of being physically active. Implicit measures have not been deployed widely enough to draw conclusions about automatic affective evaluations of physical activity at the population level (whereas implicit measures of racism have been deployed more widely), and some of the psychometric critiques noted previously present barriers to adoption as a psychological assessment tool (10). Psychometric work within the physical activity domain is needed to establish what implicit measures actually measure and how they relate to explicit measures of related processes (e.g., (9,45)). Notwithstanding those measurement challenges, specific predictions about which system will regulate specific behaviors in specific conditions are possible for when system 1 and system 2 processes conflict (26).


Contemporary theories of physical activity motivation have heavily emphasized cognitive processes and given less attention to the role of affective processes (46). The field has begun to self-correct as accumulating results link affective processes with physical activity adherence (36,46). A subset of that literature has focused on automatic affective evaluations of physical activity that derive from system 1. This emerging literature on automatic affective evaluations of physical activity supports several theoretical propositions.

Proposition 1: Automatic Affective Evaluations Are Distinct From, But May Nevertheless Overlap With, Reflective Affective Evaluations of Physical Activity

Automatic evaluations presumably evolved to facilitate rapid processing of threats and rewards in the environment. Accelerated processing of threats and rewards confers an advantage by increasing potential response time. Reflective affective evaluations also are thought to derive from people’s experiences with activity, but these evaluations are vulnerable to shift as people reprocess or rationalize their experience with an activity from the broader perspective afforded by temporal distance. Psychometric questions withstanding, in general, automatic and reflective evaluations show small positive associations with an average correlation of r = 0.24 and half of that variation is attributable to moderators (31).

Several studies have specifically tested associations between automatic and reflective affective evaluations of physical activity. None have found evidence of a significant association (3,14,15,32); however, samples have been limited in size and do not represent the general population. Some have questioned whether system 1 processes may be artifacts in explicit measures of system 2 affective attitudes because (a) intentions do not always mediate relations between affective attitudes and health behaviors and (b) implicit measures do not always add incremental validity to behavioral predictions (18). Our working conclusion based on the available evidence related to physical activity specifically is that automatic and reflective affective evaluations of physical activity are not equivalent. At face value, this conclusion conflicts with predictions from both the reflective-impulsive model and the associative-propositional evaluation model (but is consistent with principles of the systems of evaluation model). It is possible that these individual studies did not have enough statistical power to detect the small associations that would be expected based on meta-analyses in other content areas (31). Alternately, it is possible that automatic and affective evaluations are concordant for some people more than others (e.g., for people with stronger exercise-related self-beliefs) or in some contexts more than others (e.g., for leisure-time but not occupational physical activity). One recent study with 455 middle-aged adults showed that the relationship of automatic associations of exercise with health and explicit health motives was moderated by affective attitudes such that stronger automatically activated associations of exercise with health were related to lower explicit health motives in participants with weaker attitudes (9). These authors speculated that even if someone holds a strong automatic association between exercise and health, they may not particularly value it. The boundary conditions that define stronger associations or link a specific system with behavioral outcomes warrant investigation because they will illuminate the interface between these parallel aspects of system 1 and system 2.

Proposition 2: Automatic Affective Evaluations Positively Influence the Quantity of a Person’s Physical Activity

Quantity in this sense reflects the frequency, duration, or total volume of activity. When a person thinks of physical activity — or encounters a contextual cue that activates the concept of activity — the activation of that concept spreads to associated affective memories, generating a pleasant or unpleasant impulse (similar to a gut reaction) that motivates action toward or away from the concept that evoked that impulse. This process has been outlined by Williams and Evans (50) when they situated automatic affective processing as a result of previous affective responses to a health behavior. In this way, the automatic affective association orients a person toward or away from physical activity. In practice, this mechanism can account for the many unplanned microdecisions about physical activity that people make on a daily basis. For example, one study found that having positive automatic evaluations of exercise was related to decision to engage in planned exercise in the face of competing alternatives such as being invited out for a drink with friends (14). The power of automatic affective evaluations is that they occur rapidly and may involve minimal awareness or effort (presumably because the attentional architecture for monitoring threat enhanced survival and fitness in evolutionary processes). These automatic affective evaluations can be overridden by reflective processes, but, all else being equal, they are capable of tipping the scales toward or away from physical activity in any of the low-stakes, microdecisions that people make about physical activity every day.

This proposition has received consistent support in an emerging literature from multiple laboratories using a variety of research designs. At the most basic level, automatic affective evaluations discriminate between groups of more and less active people. People who engage in more regular physical activity have more pleasant automatic evaluations of physical activity (3,12,24). Automatic affective evaluations also are positively associated with recent physical activity levels in retrospective research designs. People with greater levels of recent or typical physical activity have more favorable automatic evaluations of physical activity (12,15,24). This finding also holds up in prospective tests. People with more favorable automatic evaluations of physical activity engage in more physical activity after the assessment of their automatic affective evaluations (19,33,45). Relations with retrospective and prospective physical activity also seem to be robust after adjusting for relations between physical activity and a variety of reflective motivational processes (15,19). Automatic evaluations have even been linked with adherence and drop-out patterns over time — people with more unfavorable automatic evaluations of physical activity are more likely to drop out earlier from a supervised physical activity program (4). Not all studies provided sufficient information to estimate an effect size, but, when sufficient information was available, associations tended to be small to medium size (3,12,15,19). Although that general effect size is similar to the average across other domains, caution is warranted when interpreting it because of design limitations and the heterogeneity of independent and dependent measures used (29). Nevertheless, even small effects of automatic affective evaluations may accumulate to very meaningful impacts on behavior and health if a person experiences those effects frequently (28). At this point, our working conclusion is that automatic affective evaluations play a role in regulating physical activity, particularly the maintenance of activity levels over time.

Proposition 3: Automatic Affective Evaluations Alter the Qualities of Physical Activities in Which a Person Engages

Quality in this sense refers to the intensity, mode, or any contextual property of activity. For example, Eves et al. (24) proposed that automatic affective evaluations may be less relevant for regulating walking than other activities. Conroy et al. (19) suggested that automatic affective evaluations may be more relevant for nonintentional (or spontaneous) than intentional (or planned) physical activities. Consistent with that proposal, automatic affective evaluations have been associated with spontaneous effort during a handgrip task in controlled laboratory conditions (17). Our working conclusion is that automatic affective evaluations are selectively associated with more routine and familiar physical activities that do not require advance planning or extensive self-regulatory effort.

Proposition 4: Automatic Affective Evaluations Have Both Stable and Modifiable Components

Nearly all of the available research has treated automatic affective evaluations as an individual difference and focused on variation between people. Recent work has shown that automatic evaluations have both time-invariant and time-varying components (33). In addition, changes in automatic evaluations covaried positively with changes in physical activity over a 2-wk period. It also has been found that messages targeting explicit attitudes can influence implicit attitudes (6). Thus, automatic affective evaluations may be modifiable and profitable targets for behavioral interventions. Of course, research in this area — especially on automatic affective evaluations of physical activity — is in its infancy and requires replication and extension. Longitudinal designs with more intensive and longer sampling periods will both increase sensitivity to the effects of fluctuating automatic evaluations over time and clarify the role of these evaluations on behavioral regulation. Pending replication of these initial findings, our working conclusion is that automatic affective evaluations of physical activity vary both between people as well as within people over time.

Proposition 5: Automatic Affective Evaluations Are Modifiable via Individual- and Social-Level Interventions

If automatic affective evaluations of physical activity have a time-varying component, they may be suitable targets for interventions to increase physical activity. As noted previously, automatic affective evaluations are responsive to explicit messaging (6). Another strategy for modifying automatic affective evaluations involves mental imagery. Markland et al. (37) reported differences in implicit exercise-related attitudes after a 3-min guided imagery session about a positive exercise experience compared with a control condition in which participants visualized relaxing at home after work. This effect was found in both frequent and less frequent male and female exercisers. The size of effects from these interventions ranged from medium to medium to large.

Another well-established technique that has been applied recently to enhance automatic affective evaluations is evaluative conditioning (30,38,42,47). As argued earlier, human behavior is influenced by one’s likes or dislikes and evaluative conditioning has been used extensively to increase liking for targeted stimuli (30). When objects and events repeatedly occur at the same time and corresponding mental representations are coactivated, associative links are formed (i.e., evaluative conditioning occurs). Imagine a clever exercise promoter who repeatedly presents “exercise” in fun contexts (regardless of whether the exercise itself is depicted as fun); an association between these concepts is made, increasing the likelihood that when someone encounters an exercise situation, they will associate it with the concept of “fun.” In this example, exercise is a conditioned stimulus and fun is an unconditioned stimulus. Repeatedly pairing concepts in this way promotes associative learning (27,30). Meta-analysis has indicated that evaluative conditioning effects are seen in adults (but not children), and there are no differences between men and women; evaluative conditioning effects also occur irrespective of whether the stimuli are visual, auditory, or verbal (30). It is also of interest that larger effects occur with explicit compared with implicit measures, which as noted by the authors, may be because the associations created through evaluative conditioning are considered “true.”

Evaluative conditioning has been applied successfully to modify diet and alcohol use (38). One study in the exercise domain changed explicit but not implicit “fitness and fatness” bias toward persons with obesity who were active using evaluative conditioning. Images of persons with obesity were shown exercising and repeatedly paired with words such as “fit” and “healthy.” Persons with obesity were subsequently rated more positively on a questionnaire (7). The first study to evaluate evaluative conditioning effects on physical activity found that a brief evaluative conditioning task had a large effect that increased the self-selected intensity of physical activity (5). Based on this work, our working conclusion is that targeting automatic affective evaluations via messaging, mental imagery, or evaluative conditioning can increase either the quantity or quality of physical activity.

In addition to these individual intervention strategies, social learning also is likely to influence automatic affective evaluations of physical activity. The concept of “exercise” can be associated with multiple attributes including appearance, health, or social opportunities. People can learn about exercise directly from health or fitness professionals or indirectly from popular publications that portray idealized models of what it means to be healthy, fit, or attractive. These models generally entail larger muscles for men and a thin, toned physique for women. A sizable literature points to the effects of these media portrayals on system 2 motivational processes, such as self-beliefs and attitudes. Repeated exposure is likely to produce (unintended) associations that shape system 1 processes, including automatic affective evaluations, but these effects of multiple exposures have yet to be investigated. Although previous exposure was not measured, others have found that automatically “believing” messages about exercise and appearance was related to lower intentions to exercise, even when the automatic associations were unrelated to explicit believability (8). Implicit and explicit measures of believability about exercise and health as an outcome were positively correlated.

Environmental cues also can influence situated conceptualizations, which are cognitive structures, shaped by experience, that underlie our impulses, habits, and goals (42). For example, if you play a fun game of soccer in the park with friends, you might store memories of physical exertion combined with the positive feelings of being with friends, perhaps even the smell of freshly cut grass, and the feeling of warm sunlight on your skin. Collectively, these experiences can result in a rewarding memory or a pleasant association with physical activity. Conversely, walking into an unfamiliar fitness center and seeing confident people who know how to use the machines, followed by a feeling that one does not belong, can create negative memories. These memories can result in the automatic association of “physical activity” with people who look a certain way, the smell of sweat, and a feeling of shame. When these scenarios occur often or repeatedly, they can generate stronger memories, increasing the likelihood that one’s automatic response, when cued with the idea of “physical activity,” will be positive (if you have had many positive experiences of being active with friends) or negative (if you have had many negative experiences of feeling you do not belong in a fitness center). Placing stimuli (e.g., posters) that activated concepts of health and slimness has been shown to increase purchasing of healthy snacks from a vending machine (42). Likewise, placing stimuli that activate the pleasant concepts people associate with physical activity may be a means of increasing physical activity via system 1.


The pendulum for physical activity motivation has begun to swing away from narrowly focused cognitive approaches. Affective processes are increasingly recognized as critical for physical activity motivation (36,46). In this review, we differentiated between affective processes in two psychological systems: one based on learned associations (system 1) and the other on rules and propositions (system 2). We sought to build on momentum for linking system 1 processes with physical activity behavior specifically (38,44,47). Automatic affective evaluations of physical activity in system 1 were hypothesized to play a role in regulating physical activity. Our review of the literature led to conclusions that these automatic (system 1) affective evaluations are distinct from reflective (system 2) evaluations of physical activity, are associated with both the quantity and quality of physical activity, and are potentially modifiable targets for physical activity interventions. This literature has relied on relatively young and generally healthy samples. Results should be extended to older samples and populations with chronic diseases who can benefit from increased physical activity. Effect sizes should be reported consistently in future work.

Successful proof-of-concept studies have been published for three individual intervention methods targeting automatic affective evaluations (5,6,37). Multilevel intervention approaches warrant consideration as well because they can increase the odds of improving distal health outcomes by initiating clinically meaningful behavior change. Another challenge in extending this research is that existing interventions targeting automatic affective evaluations have been delivered in lab settings and may not be well suited for population-level dissemination and implementation. Approaches with greater reach that would be adopted broadly are needed to improve population health.

Collectively, these findings highlight the potential for promoting physical activity via system 1 processes. These processes impose less self-regulatory burden than the workhorse behavior change techniques used to modify system 2 processes. Therefore, automatic affective evaluation processes may be especially critical intervention targets if people have a limited tolerance for engaging with interventions that require them to expend effort without immediate rewards.


This research was undertaken, in part, thanks to funding from the Canada Research Chairs program for Tanya R. Berry.


1. Ajzen I. The theory of planned behavior. Organ. Behav. Hum. Decis. Process. 1991; 50:179–211.
2. Anderson JR. A spreading activation theory of memory. J. Verbal Learn. Verbal Behav. 1983; 22(3):261–95.
3. Antoniewicz F, Brand R. Automatic evaluations and exercise setting preference in frequent exercisers. J. Sport Exerc. Psychol. 2014; 36(6):631–6.
4. Antoniewicz F, Brand R. Dropping out or keeping up? Early-dropouts, late-dropouts, and maintainers differ in their automatic evaluations of exercise already before a 14-week exercise course. Front. Psychol. 2016; 7:838.
5. Antoniewicz F, Brand R. Learning to like exercising: evaluative conditioning changes automatic evaluations of exercising and influences subsequent exercising behavior. J. Sport Exerc. Psychol. 2016; 38:138–48.
6. Berry TR. Changes in implicit and explicit exercise-related attitudes after reading targeted exercise-related information. Psychol. Sport Exerc. 2016; 22:273–8.
7. Berry TR, Elfeddali I, de Vries H. Changing fit and fat bias using an implicit retraining task. Psychol. Health. 2014; 29(7):796–812.
8. Berry TR, Jones KE, McLeod NC, Spence JC. The relationship between implicit and explicit believability of exercise-related messages and intentions. Health Psychol. 2011; 30(6):746–52.
9. Berry TR, Rodgers WM, Markland D, Hall CR. Moderators of implicit-explicit exercise cognition concordance. J. Sport Exerc. Psychol. 2016; 38(6):579–89.
10. Blanton H, Jaccard J, Burrows CN. Implications of the Implicit Association Test D-transformation for psychological assessment. Assessment. 2015; 22(4):429–40.
11. Blanton H, Jaccard J, Strauts E, Mitchell G, Tetlock PE. Toward a meaningful metric of implicit prejudice. J. Appl. Psychol. 2015; 100(5):1468–81.
12. Bluemke M, Brand R, Schweizer G, Kahlert D. Exercise might be good for me, but I don’t feel good about it: do automatic associations predict exercise behavior? J. Sport Exerc. Psychol. 2010; 32(2):137–53.
13. Bohner G, Dickel N. Attitudes and attitude change. Annu. Rev. Psychol. 2011; 62:391–417.
14. Brand R, Schweizer G. Going to the gym or to the movies? situated decisions as a functional link connecting automatic and reflective evaluations of exercise with exercising behavior. J. Sport Exerc. Psychol. 2015; 37(1):63–73.
15. Calitri R, Lowe R, Eves FF, Bennett P. Associations between visual attention, implicit and explicit attitude and behaviour for physical activity. Psychol. Health. 2009; 24(9):1105–23.
16. Cheval B, Sarrazin P, Isoard-Gautheur S, Radel R, Friese M. Reflective and impulsive processes explain (in)effectiveness of messages promoting physical activity: a randomized controlled trial. Health Psychol. 2015; 34(1):10–9.
17. Cheval B, Sarrazin P, Pelletier L. Impulsive approach tendencies towards physical activity and sedentary behaviors, but not reflective intentions, prospectively predict non-exercise activity thermogenesis. PLoS One. 2014; 9(12):e115238.
18. Conner M, Prestwich A, Ayres K. Using explicit affective attitudes to tap impulsive influences on health behaviour: a commentary on Hofmann et al. (2008). Health Psychol. Rev. 2011; 5(2):145–9.
19. Conroy DE, Hyde AL, Doerksen SE, Ribeiro NF. Implicit attitudes and explicit motivation prospectively predict physical activity. Ann. Behav. Med. 2010; 39:112–8.
20. Deutsch R, Strack F. Building blocks of social behavior: reflective and impulsive processes. In: Gawronski B, Payne BK, editors. Handbook of Implicit Social Cognition: Measurement, Theory, And Applications. New York (NY): Guilford Press; 2014. p. 188–203.
21. Ekkekakis P, Parfitt G, Petruzzello SJ. The pleasure and displeasure people feel when they exercise at different intensities: decennial update and progress towards a tripartite rationale for exercise intensity prescription. Sports Med. 2011; 41(8):641–71.
22. Ekkekakis P, Hargreaves EA, Parfitt G. Envisioning the next fifty years of research on the exercise-affect relationship. Psychol. Sport Exerc. 2013; 14(5):751–8.
23. Evans JS. Dual-processing accounts of reasoning, judgment, and social cognition. Annu. Rev. Psychol. 2008; 59:255–78.
24. Eves FF, Scott EJ, Hoppé R, French DP. Using the affective priming paradigm to explore the attitudes underlying walking behaviour. Br. J. Health Psychol. 2007; 12(Pt 4):571–85.
25. Fazio RH, Sanbonmatsu DM, Powell MC, Kardes FR. On the automatic activation of attitudes. J. Pers. Soc. Psychol. 1986; 50(2):229–38.
26. Gawronski B. Ten frequently asked questions about implicit measures and their frequently supposed, but not entirely correct answers. Can. Psychol. 2009; 50(3):141–50.
27. Gawronski B, Bodenhausen GV. Associative and propositional processes in evaluation: an integrative review of implicit and explicit attitude change. Psychol. Bull. 2006; 132(5):692–731.
28. Greenwald AG, Banaji MR, Nosek BA. Statistically small effects of the Implicit Association Test can have societally large effects. J. Pers. Soc. Psychol. 2015; 108(4):553–61.
29. Greenwald AG, Poehlman TA, Uhlmann EL, Banaji MR. Understanding and using the Implicit Association Test: III. Meta-analysis of predictive validity. J. Pers. Soc. Psychol. 2009; 97(1):17–41.
30. Hofmann W, De Houwer J, Perugini M, Baeyens F, Crombez G. Evaluative conditioning in humans: a meta-analysis. Psychol. Bull. 2010; 136(3):390–421.
31. Hofmann W, Gawronski B, Gschwendner T, Le H, Schmitt M. A meta-analysis on the correlation between the implicit association test and explicit self-report measures. Pers. Soc. Psychol. Bull. 2005; 31(10):1369–85.
32. Hyde AL, Doerksen SE, Ribeiro NF, Conroy DE. The independence of implicit and explicit attitudes toward physical activity: introspective access and attitudinal concordance. Psychol. Sport Exerc. 2010; 11(5):387–93.
33. Hyde AL, Elavsky S, Doerksen SE, Conroy DE. The stability of automatic evaluations of physical activity and their relations with physical activity. J. Sport Exerc. Psychol. 2012; 34(6):715–36.
34. Kahneman D. Thinking Fast and Slow. New York (NY): Farrar, Straus & Giroux; 2011. 499 p.
35. Kiviniemi MT, Voss-Humke AM, Seifert AL. How do I feel about the behavior? The interplay of affective associations with behaviors and cognitive beliefs as influences on physical activity behavior. Health Psychol. 2007; 26(2):152–8.
36. Lee HH, Emerson JA, Williams DM. The exercise-affect-adherence pathway: an evolutionary perspective. Front. Psychol. 2016; 7:1285.
37. Markland D, Hall CR, Duncan LR, Simatovic J. The effects of an imagery intervention on implicit and explicit exercise attitudes. Psychol. Sport Exerc. 2015; 17:24–31.
38. Marteau TM, Hollands GJ, Fletcher PC. Changing human behavior to prevent disease: the importance of targeting automatic processes. Science. 2012; 337(6101):1492–5.
39. McConnell AR, Rydell RJ. The systems of evaluation model: a dual-systems approach to attitudes. In: Sherman JW, Gawronski B, Trope Y, editors. Dual Process Theories of the Social Mind. New York (NY): Guilford Press; 2014. p. 204–17.
40. McEwan D, Harden SM, Zumbo BD, et al. The effectiveness of multi-component goal setting interventions for changing physical activity behaviour: a systematic review and meta-analysis. Health Psychol. Rev. 2016; 10:67–88.
41. Michie S, Abraham C, Whittington C, McAteer J, Gupta S. Effective techniques in healthy eating and physical activity interventions: a meta-regression. Health Psychol. 2009; 28(6):690–701.
42. Papies EK. Health goal priming as a situated intervention tool: how to benefit from nonconscious motivational routes to health behaviour. Health Psychol. Rev. 2016; 10(4):408–24.
43. Phelps EA, O’Connor KJ, Cunningham WA, et al. Performance on indirect measures of race evaluation predicts amygdala activation. J. Cogn. Neurosci. 2000; 12(5):729–38.
44. Rebar AL, Dimmock JA, Jackson B, et al. A systematic review of the effects of non-conscious regulatory processes in physical activity. Health Psychol. Rev. 2016; 10(4):395–407.
45. Rebar AL, Ram N, Conroy DE. Using the EZ-diffusion model to score a single-category Implicit Association Test of physical activity. Psychol. Sport Exerc. 2015; 16(3):96–105.
46. Rhodes RE, Kates A. Can the affective response to exercise predict future motives and physical activity behavior? A systematic review of published evidence. Ann. Behav. Med. 2015; 49:715–31.
47. Sheeran P, Gollwitzer PM, Bargh JA. Nonconscious processes and health. Health Psychol. 2013; 32(5):460–73.
48. Smith ER, DeCoster J. Dual-process models in social and cognitive psychology: conceptual integration and links to underlying memory systems. Personal. Soc. Psychol. Rev. 2000; 4:108–31.
49. Stanley D, Phelps E, Banaji M. The neural basis of implicit attitudes. Curr. Dir. Psychol. Sci. 2008; 17(2):164–70.
50. Williams DM, Evans DR. Current emotion research in health behavior science. Emot. Rev. 2014; 6(3):277–87.

nonconscious; maintenance; burden; pleasure; implicit measurement

Copyright © 2017 by the American College of Sports Medicine