The fundamental inspiration for the ART of physical inactivity and exercise was the social psychologist Lewin’s (1943, 1951) proposal that, in order to accurately explain behavior, psychological models must first consider the exact moment at which the behavior occurs. This proposition and some of its consequences will be described first. Then, the theoretical roots and concepts from which the ART of physical inactivity and exercise were developed will be explained. These are the hedonistic perspective on motivated behavior, the automatic evaluation effect, the evaluation-behavior link, and the dual-process view of behavioral decision-making.
Lewin’s force-field analysis
Lewin, an important figure in modern social psychology, developed his influential field theory in the 1940s. Field theory postulates that human behavior and behavior change should be understood in terms of the “forces” and “tensions” that move us to action (Lewin, 1943). Its three central assumptions are (1) that behavior is a consequence of the totality of the situation around us, (2) this situation can be described in terms of fundamentally interrelated factors that constitute a dynamic field, and (3) the dynamic field has more influence on behavior than past experience or future desires.
According to Lewin, the dynamic field or “life space” contains the person and his or her psychological environment (i. e., the environment as subjectively perceived by that person) as related to his or her needs. Needs (and quasi-needs, whose salience varies over time, depending on the situation) are the central motivational concept in Lewin’s field theory. Needs release energy and transform “regions” in the person into “tension systems” whenever they give rise to an intention or an intentional action (intentions and intentional actions arise from needs and at goals). “Vectors”, namely forces that act on a person, determine the direction in which a person moves through his or her psychological environment. Vectors “push” a person towards an attractive (positively valenced) region and “repulse” that person from a region he or she thinks will increase tension (negatively valenced). Typically, at least two vectors (the “force” and “counterforce”) act on a person at the same time, maintaining a “quasi-equilibrium” or creating a tension system. “Locomotion” through the psychological environment—in other words, behavior change—can thus be seen as resulting from the interaction of the two forces.
Force-field analysis (Lewin, 1951) is a simple conceptual framework that reminds us to take into account both the driving and restraining forces associated with the transition to another, perhaps opposite, behavior (Fig. 1). In order to change, an individual must first identify and “unfreeze” the driving and restraining forces holding him or her in the current state of quasi-equilibrium; the change (“transition”) is achieved by increasing the drivers, reducing the restraints, or both, in order to create an imbalance of forces. Finally, the system must be brought back into quasi-equilibrium (“refrozen”).
Applying Lewin’s ideas to the context of physical inactivity, let us imagine a situation in which someone is sitting on the sofa and is reluctant to do anything different. According to Lewin, that this person’s motivation for behavior change and exercise may not yet be strong enough can only be one part of the explanation. The other part is that a restraining force is holding this person back in this situation. The ART of physical inactivity and exercise suggests that momentary affective responses to a situation act as such a restraining force.
Affective responses to exercise and hedonistic thinking in exercise psychology
In the past few years, the voices arguing that the role of affect in exercise motivation might be substantially underestimated in contemporary theoretical work have grown louder (e. g., Ekkekakis & Dafermos, 2012; Sudeck, Schmid, & Conzelmann, 2016; Wienke & Jekauc, 2016). Affect is a broader concept than mood and emotion, and core affect is defined as a “neuro-physiological state consciously accessible as a simple, primitive, nonreflective feeling most evident in mood and emotion but always available to consciousness” (Russel & Feldman Barrett, 2009, p. 104). Most importantly for the work presented here, core affective valence characterizes all states in which a person feels good or bad, including free-floating pleasure and displeasure, as well as pleasant and unpleasant moods and emotions; it is experienced constantly although its nature and intensity can vary over time (Ekkekakis, 2013, p. 38).
Affective responses to exercise, especially core affective valence reported under different exercise intensities, and the link between these affective responses and future exercising, have received considerable research attention in the past few years. There is convincing experimental evidence (see the review by Ekkekakis, Parfitt, & Petruzello, 2011) that many people begin reporting negative changes in core affective valence (reduced pleasure or even displeasure) during exercise performed at an intensity that approximates the ventilatory or lactate threshold, (a level of intensity at which lactate begins to accumulate in the blood). Above this threshold, negative changes in core affective valence become universal, before variability re-emerges after exercise has ended. With regard to the behavioral consequences of these responses, a growing number of studies show that a positive change in core affective valence during exercise is reliably linked to future exercise (with small to medium effect sizes), whereas post-exercise affect is unrelated to future behavior.
Psychological or motivational hedonism (partial psychological hedonism; Murphy & Eaves, 2016) posits that, in general, individuals tend to seek pleasurable experiences and avoid displeasure (Rozin, 1999). This must not be misunderstood as implying that seeking pleasure and avoiding displeasure is postulated to be the one and only ultimate source of all human motivation (Ekkekakis & Dafermos, 2012). Nevertheless, hedonistic theories differ significantly from most theories presently used in the study of exercise motivation, which are based on a cognitive core and assert that, once enough information is available (e. g., about the health benefits of exercise and personal relevance of goals), individuals will inevitably make the rational decision to change their behavior and will be motivated to do so, more or less regardless of any hardship they have to endure in the process.
We believe that negative core affective valence during exercise reflects such hardship (Ekkekakis, 2003), especially for sedentary, low-active, or obese individuals (e. g., Ekkekakis, Vazou, Bixby, & Georgiadis, 2016). The ART of physical inactivity and exercise adds a hedonistic perspective to theoretical accounts of exercise motivation by postulating that core affective valence can strongly influence deliberative reasoning about exercise engagement and effort; it further suggests that, under certain circumstances, core affective valence may have a direct, immediate impact on behavior through behavioral urges.
Affective evaluation, automatic evaluation, and the evaluation-behavior link
Investigating the role of evaluative mental processes in order to understand why people do the things they do has a long history in social psychology (Briñol & Petty, 2012). The contemporary view is that evaluations are the basis of attitudes (e. g., Fazio, 1990; Zanna & Rempel, 1988). Although it has long been recognized that affective evaluations contribute to attitude formation (e. g., Rosenberg & Hovland, 1960; Thurstone, 1946; Zajonc, 1980), discussion of the distinct, influential “pre-cognitive” effect that affective evaluations may have on evaluative judgment and behavior has, to date, been neglected or marginalized in exercise psychology (Bodur, Brinberg, & Coupey, 2000; Ekkekakis & Zenko, 2016). This may be partly an historical accident, in that the emergence of exercise psychology as a discipline (in the 1960s, e. g., see Buckworth & Dishman, 2002) followed the so-called “cognitive revolution” in psychology, which had began in the 1950s (e. g., see Miller, 2003).
The cognitive revolution brought a new and influential perspective to research on the psychological mechanisms of behavior change. Fishbein and Ajzen (1975) suggested that all mental evaluations are based on beliefs about the object of evaluation. They introduced cognitive constructs like “intention” and “planning” to bridge the gap between evaluative judgments (attitudes) and behavior, and, importantly, argued that behavior change was a product of reasoned action. Researchers in exercise psychology readily adopted this view (and similar cognitive approaches to behavior change, e. g., social cognitive theory; Bandura, 1977), probably because, at the time, it appeared as the most promising approach to behavior modification.
There were, however, social psychologists who resisted the zeitgeist and refused to abandon the notions of precognitive affect and affective evaluation. They began exploring the issues of when and under what circumstances affective evaluation would influence behavior. For example, Zajonc (1980, p. 151) insisted that “preferences need no inferences” and Fazio (1986, 1990) concurred that presentation of an attitude object automatically activates an individual’s memory of the evaluation associated with that object.
Today automatic evaluations are defined as the immediate affective (positive or negative) responses a person has towards an event or stimulus (Bargh, Chaiken, Raymond, & Hymes, 1996; Murphy & Zajonc, 1993). According to Fazio (2001), these automatic evaluations
...alert us to the presence of objects that have the potential for hedonic consequences and promote hedonically meaningful categorisations of such objects. We are likely to notice those objects that can provide reward or satisfaction, those that we have personally defined as likeable and can benefit from approaching. Likewise, we are likely to notice those objects toward which we have developed strongly associated negative evaluations, ones that we wish to avoid if at all possible (p. 129).
The automatic evaluation effect has been unequivocally established in experimental social psychology (De Houwer & Hermans, 2001). Only a few years ago, researchers in exercise psychology started to apply this idea in their research, exploring correlations between automatic evaluations and exercise behavior (e. g., Bluemke, Brand, Schweizer, & Kahlert, 2010; Calitri, Lowe, Eves, & Bennett, 2009).
Likewise, research in experimental social psychology showed that activated evaluations could stimulate immediate approach-avoidance motor responses (e. g., Chen & Bargh, 1999; Seibt, Neumann, Nussinson, & Strack, 2008). These studies substantiated the assumption that there is a direct evaluation-behavior link and concluded that valenced behavior can be activated automatically by affective stimuli (Krieglmeyer, De Houwer, & Deutsch, 2013; see Phaf, Mohr, Rotteveel, & Wicherts, 2014, for a more skeptical perspective). This line of research has not yet influenced theoretical thinking within exercise psychology. To the best of our knowledge, so far there has been only one experimental study that has examined the effect of activated automatic evaluations on complex exercise behavior (Antoniewicz & Brand, 2016a; reviewed below).
The ART of physical inactivity and exercise is at the core of this emerging line of exercise psychology research. The theory assigns central importance to automatic affective associations of exercise and postulates that these associations give rise to action impulses directly. Assumptions about the conditions under which these action impulses are translated into complex behavior were derived from the general framework of dual-process theories, which is described in the next section.
The dual-process approach to feelings, thought, and behavior
As described in cognitive and social psychology, dual-process theories suggest that behavioral phenomena are the results of two qualitatively different mental processes. Type-1 processing is supposed to be fast and automatic in the sense that it requires minimal cognitive resources and effort, whereas type-2 processing is supposed to be generally slower and reflective and takes the form of more controlled reasoning (Evans, 2008; Evans & Stanovich, 2013).Footnote 1
One reason why many different labels for type-1 and type-2 processes have been used by researchers in the field (e. g., type-1: implicit, contextualized, associative, heuristic, experiential, impulsive; type-2: explicit, abstract, propositional, analytic, rational, reflective) is that different models often emphasize different aspects of human thought, affect, and behavior. For example, the associative-propositional processes in evaluation (APE) model (Gawronski & Bodenhausen, 2006, 2011; see below) applies exclusively to these mental processes and uses the implicit-explicit terminology. On the other hand, the reflective-impulsive determinants model (Strack & Deutsch, 2004; see below) emphasizes the links with social behavior and thus refers to reflections and impulses as the processes that underlie such behavior. We will describe these two models, since they are most relevant to our work, using their original terminology, but also draw links between this terminology and the broader type-1 vs. type-2 distinction.
Associative and propositional processes in evaluation.
The APE model describes two kinds of evaluative responses (Gawronski & Bodenhausen, 2006, 2011). Implicit evaluation occurs spontaneously as the personal “default” response to an object and is defined as the outcome of associative processes, namely the activation of object-related mental associations in memory (type-1 process). The pattern of activation of object-related associations depends on the input stimulus and the preexisting structure of associations in memory (Smith, 1996). In other words, in different contexts, the same object may activate different subsets of associations. The net valence of the concepts activated by the object defines the valence of the individual’s “gut” affective reaction to the object and thus his or her implicit evaluation of it.
According to the APE model, this affective “gut” reaction forms the basis of propositional (type-2) processing from which the explicit evaluation of the object develops. Propositions are mentally represented statements, such as beliefs about facts and values. Propositional processing within the APE model is thus inherently concerned with translating the subjective valence of the affective “gut” reaction (e. g., an unpleasant feeling) into a propositional statement (such as “I dislike exercising”), which can then be related to other stored propositional beliefs that are considered relevant to an explicit evaluation. Sometimes the propositional evaluation implied by the “gut” reaction is consistent with other salient propositions (such as “everyone in my family hates exercise”) and endorsed in a verbally reported explicit evaluation. On other occasions, however, propositions that are incompatible with the “gut” reaction, but still judged as relevant at least at that moment in time (e. g., “just today my doctor advised me to start exercising regularly”), may lead the individual to reject (deliberately disregard, suppress, or override) his or her affective “gut” reaction. The result is an implicit-explicit evaluation discrepancy. Most likely at the subconscious level, individuals will try to avoid the aversive feelings induced by such discrepancies (Festinger, 1957) and restore consistency by rejecting one of the propositions (e. g., “disliking exercise is not rational”) or by searching for additional propositions that will help temporarily resolve the discrepancy (e. g., “starting exercise tomorrow is as good as starting today”).
Importantly, the type-2 process involved in rejecting the proposition implied by the affective “gut” response in one particular situation does not necessarily permanently deactivate or eliminate the automatic association that gave rise to it. However, as a result of the presumed interactions between spontaneous associations and propositional reasoning, every activation of an evaluative process leaves a trace in memory. Therefore, the propositional processes involved in generating an evaluation can create new associations or attenuate existing ones. In the same way, every momentarily activated association that is translated into and processed as propositional information can increase the salience and relevance of stored propositions. The results of these interactions can be understood in terms of fundamental learning principles, such as a reinforcement mechanism. Consequently, evaluative processes tend towards self-reinforcement unless new information (e. g., behavioral experience, cognitive information) is available for activation in the future (i. e., in a situation, in which a new evaluative process is triggered).
Experimental evidence supporting the fundamental postulates of the APE has been compiled into two comprehensive reviews (Gawronski & Bodenhausen, 2006, 2011). Numerous empirical social-psychological studies draw on the APE as a broad explanatory framework for the role of type-1 processes in social cognition and behavior but only a few recent articles in exercise psychology have done so (e. g., Berry, Rodgers, Markland, & Hall, 2016; Brand & Antoniewicz, 2016).
Reflective and impulsive determinants of behavior.
Another influential framework for interpreting the complex relationship between type-1 and type-2 processes is the reflective-impulsive model (RIM; Strack & Deutsch, 2004). Although the RIM shares certain important assumptions with the APE model (e. g., about the fundamentally associative nature of type-1 processes), there are also notable differences between the two. The RIM, for example, assumes that the two processes are invoked simultaneously but independently (i. e., not in the default-interventionist logic of the APE, in which the implicit association gives the default value for further propositional processing such that all type-2 processing depends on initial type-1 input). Most of these differences are the subject of unresolved theoretical debates (see Evans & Stanovich, 2013, for a critical review of the dual-process/dual-system approach in general), and are beyond the scope of this article. What makes the RIM especially relevant to the present theoretical proposal (the ART) is its focus on how the two types of processes affect behavior.
According to the RIM, the interplay between the impulsive and the reflective systems can be generally described as competition for control over the overt response. It is assumed that information entering the perceptual gates will always be processed in the impulsive system, that this processing is mediated by an approach-avoidance motivational orientation, and that a behavioral schema (i. e., sensory-motor cluster) will be activated as a result. The impulsive system is viewed as a system of experiential primacy, in which automatic thoughts and feelings arise spontaneously through activation of learned associations. Depending on its intensity and the availability of self-control resources (Baumeister & Heatherton, 1996), a stimulus may also enter and be processed in the reflective system. In this case, by weighing beliefs and knowledge, an intention can be formed in the reflective system and an appropriate behavioral schema in the impulsive system may be activated.
As a result of independent processes in the reflective and impulsive systems, two behavioral schemata can be activated at the same time. These schemata may be concordant (directed towards the same or similar goals, e. g., approach-approach) or divergent (e. g., approach-avoidance). When divergent behavioral schemata are activated and self-regulatory resources are depleted (Muraven, Tice, & Baumeister, 1998), the schema from the impulsive system will likely prevail and be expressed behaviorally. In contrast, when self-regulatory resources are available, reflective operations are possible (e. g., distracting attention from the tempting stimulus, emphasizing the rational consequences of one behavioral option), thus, enabling the reflective system to control overt behavior (Hofmann & Friese, 2017). Evidence supporting the validity of the central assumptions and predictions of the RIM has been comprehensively reviewed by Strack and Deutsch (2004) and, more recently, by Deutsch, Gawronski, and Hofmann (2017).