Acute affective response to a moderate-intensity exercise stimulus predicts physical activity participation 6 and 12 months later

Abstract

Objective

Examine the relationship between acute affective responses during a moderate-intensity exercise stimulus and future physical activity participation.

Design

Longitudinal, observational study in the context of a randomized controlled trial.

Methods

Healthy, sedentary adults (n=37) reported their basic affective response (i.e., feel good versus bad) prior to and during an acute, moderate-intensity exercise stimulus presented prior to randomization in a controlled physical activity promotion trial. At 6 and 12 months, 31 of the 37 participants reported their total weekly minutes of physical activity.

Results

As hypothesized, basic affective response to the moderate-intensity stimulus predicted 6-month physical activity (β=.51, p=.013) when controlling for baseline physical activity and self-reported affect prior to the initial exercise stimulus, and 12-month physical activity (β=.45, p=.047) when also controlling for 6-month physical activity.

Conclusions

Affective response to an acute moderate-intensity exercise stimulus predicted self-reported physical activity 6 and 12 months later. The findings could have implications for prescription of exercise intensity, as exercising at an intensity that yields a positive affective response may lead to greater participation in physical activity programs among previously sedentary adults.

Introduction

Participation in regular physical activity is associated with a lower risk of several chronic diseases (Knowler et al., 2002; McTiernan et al., 2003; Slattery & Potter, 2002; Thompson et al., 2003; Vuori, 2001) and all cause mortality (Lee & Skerrett, 2001; Mokdad et al., 2004). Yet, only 45% of US adults currently engage in the recommended level of physical activity (Macera et al., 2005), thus highlighting the need to identify determinants of physical activity adoption and maintenance (Baranowski, Anderson, & Carmack, 1998; Lewis, Marcus, Pate, & Dunn, 2002). Physical activity research has generally been based on one or more socio-cognitive models (Ajzen, 1991; Bandura, 1986; Rogers, 1983) or models that incorporate multiple constructs from various theories (Prochaska & DiClemente, 1984). Consistent with these models, a majority of research on physical activity determinants has focused on cognitive, behavioral, and more recently, environmental variables (e.g., Humpel, Owen, & Leslie, 2002; Trost, Owen, Bauman, Sallis, & Brown, 2002). Some of these models also include affective states as predictors of behavior (e.g., Bandura, 1986); however, little is known regarding how affective variables relate to physical activity participation.

Hedonic theory (Kahneman, 1999; Young, 1952), a derivative of learning theory (e.g., Thorndike, 1927), provides a framework for understanding relationships between affect and physical activity behavior. While behaviorist notions of learning focus on observable and instrumental consequences of behavior (Skinner, 1938; Watson, 1919), hedonic theory focuses on affective responses to behavior as determinants of future behavior (Kahneman, 1999; Young, 1949). Cabanac (1971), Cabanac (1992) has argued that hedonic responses (i.e., pleasure versus displeasure) provide an index of the usefulness of behavior and its immediate outcomes relative to existing internal states. For example, Kahneman, Fredrickson, Schreiber, and Redelmeier (1993) have shown that affective responses to a behavior may influence decisions regarding whether or not to repeat the behavior. This tendency for humans to maximize pleasure and minimize displeasure has been examined extensively as a mechanism for eating behavior (for reviews see Cota, Tschöp, Horvath, & Levine, 2006; Nasser, 2001), and has recently been posited in the context of exercise behavior (Cabanac, 2006; Ekkekakis, Hall, & Petruzzello, 2005). Consistent with hedonic theory, the purpose of this paper is to examine acute affective responses to exercise behavior as a predictor of future physical activity participation.

When considering the relationship between affect and physical activity, it is important to first distinguish among affect-related terms. In this paper we distinguish between basic affect and distinct affective states. Consistent with Ekkekakis and Petruzzello's (2000) review of the use of affect-related terminology, we consider basic affect to be the most general valenced experiential response (i.e., good/pleasure versus bad/displeasure). Distinct affective states, such as emotions and moods, include this basic affective component (i.e., good/pleasure versus bad/displeasure), plus a cognitive appraisal process (for a discussion of differences between emotions and moods see Ekkekakis & Petruzzello, 2000; Larsen, 2000). Thus, basic affect (i.e., good/pleasure versus bad/displeasure) can occur in the absence of an appraisal process and thus outside the context of more distinct affective states, or, can underlie a distinct affective state. For example, anger and embarrassment are distinct affective states, which include an underlying basic affective response (i.e., bad/displeasure), as well as an appraisal process. Finally, we use the term affect as an umbrella term encompassing both basic affect and distinct affective states.

A significant body of research has examined changes in both distinct affective states (Berger & Motl, 2000; Gauvin & Rejeski, 1993; McAuley & Courneya, 1994), and, more recently, basic affect (Ekkekakis et al., 2005) in response to acute bouts of exercise. However, few studies have examined the link between affective response to exercise and future physical activity. Annesi (2005), Annesi (2002a), Annesi (2002b) conducted a series of studies examining relationships between affective responses prior to and following acute exercise bouts and subsequent attendance at an exercise facility among 69, 72, and 50 (respectively) healthy, previously sedentary adults. Participants were asked to complete 20–30-min moderate to vigorous intensity exercise sessions 3 times per week over a 12–15 week period and to complete the Exercise-induced Feeling Inventory (EFI; Gauvin & Rejeski, 1993) biweekly before and 5–10 min after the exercise session. The EFI assesses 4 distinct affective states, including positive engagement, revitalization, tranquility, and physical exhaustion. In each of the studies, differences in pre–post scores for each subscale were aggregated over time, and used to predict attendance at the exercise facility during the 15-week program period. There was no main effect of EFI responses to exercise in two of the studies (Annesi (2002a), Annesi (2002b)); however, in one of these studies there was an interaction between EFI responses and baseline self-motivation. Specifically, among participants with low self-motivation there were positive relationships between positive engagement, revitalization, and tranquility subscales and exercise attendance, and a negative relationship between the physical exhaustion subscale and exercise attendance, but these relationships were opposite among participants with high self-motivation (Annesi, 2002a). In the third study (Annesi, 2005), conducted only among women, exercise attendance was positively related to revitalization and negatively related to exhaustion. The other two subscales of the EFI were not administered.

Berger and Owen (1992) used a similar design to examine distinct affective responses to exercise, as measured by the state scale of the State-Trait Anxiety Inventory (STAI; Spielberger, Gorsuch, & Lushene, 1970) and the Profile of Mood States (POMS; McNair, Lorr, & Droppleman, 1971), as predictors of class attendance among 59 college students enrolled in either swimming or yoga classes. The STAI and POMS, which includes subscales for 6 mood states, were taken before and after three exercise classes at the beginning, middle, and end of the 14-week semester. Aggregated change scores for the STAI and each subscale of the POMS accounted for approximately 25% of the variance in classroom attendance when entered together in multivariate analysis. Although results for individual scales were not presented, the authors stated that “exercisers who reported greater mood benefits had fewer absences” (p. 1340).

Two additional studies measured affective states before and after a baseline exercise session. Klonoff, Annechild, and Landrine (1994) assessed levels of happiness and euphoria via single-item Likert scales administered prior to and following a 30-min aerobics session among 23 healthy, previously inactive women. Change scores for the two scales were found to be unrelated to number of aerobics sessions completed during a subsequent 10-week aerobics program. Similarly, Carels, Berger, and Darby (2006) assessed mood states via the POMS before and after a graded submaximal exercise test completed at baseline of a 24-week behavioral weight-loss program among 25 previously sedentary, obese, postmenopausal women. No relationship was found between change scores for each POMS subscale and subsequent exercise; however, scores from the subscales vigor–activity, depression–dejection, and anger–hostility taken after the baseline exercise test were positively related to one or more measures of subsequent physical activity, while negative relationships were found for fatigue–inertia and confusion–bewilderment.

Taken together, the findings from previous studies examining the relationship between affective response to acute exercise and future physical activity behavior have been inconsistent and difficult to characterize. While the studies reviewed are among the first to examine this issue, the mixed findings may be attributed to a number of conceptual and methodological weaknesses. First, all of the studies reviewed measured distinct affective states rather than basic affect. When examining the relationship between affective responses to acute exercise and future physical activity participation, there are a number of reasons to prefer basic affect over distinct affective states at the present stage of knowledge development. For example, distinct affective states, such as tranquility, tension, or confusion–bewilderment do not lend themselves to any clear hypotheses concerning the relationship between affect and future behavior. Alternatively, consistent with hedonic theory (Kahneman, 1999; Young, 1952), differences in positive–negative valence of basic affective responses to exercise among previously sedentary individuals may serve to reinforce or punish the behavior and in turn make future physical activity participation more or less likely. Moreover, as pointed out by Ekkekakis and Petruzzello (2000), it is not clear which distinct affective states would have the greatest influence on subsequent physical activity, and since not all distinct affective states can be measured at one time, there would be a risk of failing to measure one or more crucial affective states.

Second, due to the multi-item nature of measures of distinct affective states (e.g., Gauvin & Rejeski, 1993), assessments were administered prior to and following, but not during the exercise task; thus, these studies measured the affective response to completing the bout of exercise, rather than the affective response during exercise (Ekkekakis & Petruzzello, 1999; Hall, Ekkekakis, & Petruzzello, 2002). As a result, studies, such as these, assessing affective responses before and after but not during exercise support the conclusion that acute bouts of exercise uniformly improve affective states (Yeung, 1996), while studies that have measured affective response during exercise (e.g., Feeling Scale; Rejeski, Best, Griffith, & Kenney, 1987) have shown significant interindividual variability in acute affective responses during moderate-intensity exercise (e.g., brisk walking; Ekkekakis, Hall, Van Landuyt, & Petruzzello, 2000), leaving open the possibility that this variable affective response may predict future physical activity participation (Ekkekakis, Hall, & Petruzzello, 2004).

Third, these studies exhibited a number of methodological and/or analytical problems involving timing of assessments and lack of control for potential confounders. Although all of the studies were longitudinal, in some studies (Annesi (2005), Annesi (2002a), Annesi (2002b); Berger & Owen, 1992), the measurement of affect temporally overlapped with the measurement of physical activity participation, making the possibility of confounding more likely. That is, a third variable, such as a favorable social atmosphere, may have caused both positive affective responses and increased exercise participation. Moreover, Berger and Owen (1992) did not use a previously sedentary sample and did not control for baseline physical activity levels; thus, the relationships found between mood changes and exercise class attendance could be an artifact of the influence of baseline physical activity levels on both affective response to exercise and class attendance. Finally, Carels et al. (2006) did not control for pre-exercise affective states when examining the association between post-exercise affect and future physical activity, and thus it is possible that baseline affect influenced both post-exercise affect and subsequent physical activity behavior.

In the present study, we sought to overcome these limitations by examining the association between basic affective responses to a moderate-intensity exercise stimulus with future physical activity participation among previously sedentary adults participating in a moderate-intensity physical activity promotion intervention. Specifically, we examined affective response to moderate-intensity exercise during a baseline graded submaximal exercise test in order to determine if these responses could predict physical activity participation at 6 and 12 months. In order to reduce the chances of confounding, we controlled for affective responses recorded prior to the graded exercise test, as well as baseline levels of physical activity. Consistent with hedonic theory, we hypothesized that those with a more positive affective response to exercise would engage in more minutes of physical activity. Because previous studies have shown that positive affective responses are related to lower ratings of perceived exertion (RPE; e.g., Hardy & Rejeski, 1989), we also measured RPE in order to determine the association between affective responses to exercise and future physical activity behavior independent of RPE.