Impact of videos targeting intrinsic versus extrinsic motivation on exercise experience and effort

ABSTRACT

Regular exercise is beneficial for physical and psychological health. Unfortunately, approximately one in four adults and the majority of adolescents do not meet recommendations for weekly exercise participation. A commonly cited cause for not meeting these recommendations is a lack of motivation, yet few studies have attempted to manipulate motivation. Thus, directional conclusions citing lack of motivation as a cause for insufficient exercise may not be warranted. The current project compares the efficacy of two experimentally manipulated motivational strategies on exercise behaviour in two groups of individuals: those meeting recommendations for exercise participation and those not meeting recommendations. Participants exercised at a self-selected pace after being randomly assigned to watch a video intended to cue either intrinsic or extrinsic motivation. Though effort was largely similar across subjects, effects of the cues on affective response to exercise and post-exercise motivational outcomes differed based on group and condition.

KEYWORDS:

• Physical activity
• health
• self-determination theory
• affect

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Results

Sample characteristics

Participants ranged in age from 18 to 26 (M = 19.42, SD = 1.64) years. Body mass index (BMI) ranged from 18.30–39.78 (M = 23.11, SD = 3.72) kg/m²; 1% of the sample was considered underweight, 82% were considered healthy weight, and 17% were considered overweight. Weekly moderate-intensity exercise minutes ranged from 5 to 720 (M = 166.00, SD = 162.27) for maintainers and from 0 to 420 (M = 87.97, SD = 99.72) for non-maintainers. Weekly vigorous-intensity exercise minutes ranged from 0 to 700 (M = 139.50, SD = 160.34) for maintainers and from 0 to 480 (M = 73.44, SD = 97.93) for non-maintainers.³ Also note that though some non-maintainers indicated that they had been engaging in a considerable amount of exercise at baseline, no minutes of exercise was the most common response across the different levels of exercise intensity (i.e., 22% of non-maintainers reported 0 min of mild-intensity exercise at baseline, 19% reported 0 min of moderate-intensity exercise, and 28% reported 0 min of high-intensity exercise).

During exercise measures

Effort

To examine group differences in heart rate across the exercise bout, we ran a mixed analysis of variance (ANOVA) with one within-subjects factor (time: 0, 10, 20, 30 min) and two between-subjects factors (group: maintainers, non-maintainers; and condition: aesthetic, enjoyment). The main effects of group [F(1, 68) = 0.68, η² = .010, p = .413] and condition [F(1, 68) = 0.15, η² = .002, p = .703] were not significant. The linear time by group interaction was significant [F(1, 68) = 4.81, η² = .066, p = .032], with maintainers showing a steeper increase in heart rate than non-maintainers (see Figure 1). We did not find evidence for the linear time by condition [F(1, 68) = 0.01, η² = .000, p = .944] or time by group by condition [F(1, 68) = 0.49, η² = .007, p = .486] interactions.

Figure 1. Heart rate means (at 0, 10, 20, and 30 min) by group (maintainers vs. non-maintainers).

Perceived exertion

To examine group differences in perceived exertion across the exercise bout, we ran a mixed ANOVA with one within-subjects factor (time: 10, 20, 30 min) and two between-subjects factors (group: maintainers, non-maintainers; and condition: aesthetic, enjoyment). The main effects of group [F(1, 68) = 0.14, η² = .002, p = .713] and condition [F(1, 68) = 0.00, η² = .000, p = .960] were not significant. In addition, the linear time by group [F(1, 68) = 1.01, η² = .015, p = .320] and linear time by condition [F(1, 68) = 0.33, η² = .005, p = .570] interactions were not significant, though the linear time by group by condition interaction approached significance [F(1, 68) = 3.53, η² = .049, p = .065]. Visual inspection of the means indicated that perceived exertion increased from 10 min to 20 min and from 20 min to 30 min among maintainers assigned to both conditions and non-maintainers assigned to the enjoyment condition. Among non-maintainers assigned to the aesthetic condition, perceived exertion increased from 10 min to 20 min but decreased slightly from 20 min to 30 min.

Behavioural intrinsic motivation measure

Among maintainers, 30% in the aesthetic condition and 25% in the enjoyment condition opted to increase their speed, 65% in aesthetic condition and 50% in the enjoyment condition chose to keep their original speed, and 5% in the aesthetic condition and 25% in the enjoyment condition opted to decrease their speed. Among non-maintainers, 6% in the aesthetic condition and 24% in the enjoyment condition opted to increase their speed, 67% in the aesthetic condition and 65% in the enjoyment condition chose to keep their original speed, and 27% in the aesthetic condition and 11% in the enjoyment condition opted to decrease their speed. To examine group differences in our behavioural measure of intrinsic motivation, we ran a logistic regression model with type of speed change as the dependent variable (down = −1, none = 0, up = +1) and group (non-maintainers = −1, maintainers = +1), condition (aesthetic = −1, enjoyment = +1), and the group by condition interaction as regressors. There were no significant effects of group [F(1, 68) = 1.29, η² = .019, p = .261] or condition [F(1, 68) = 0.05, η² = .001, p = .818], though the group by condition interaction approached significance [F(1, 68) = 3.75, η² = .052, p = .057]. It appears as though non-maintainers assigned to the aesthetic condition drove this trending group by condition interaction; relative to the other groups, far fewer non-maintainers in the aesthetic condition chose to increase their speed, whereas non-maintainers assigned to the enjoyment condition looked more like maintainers in either condition.

Affective response

To examine group differences in affective response across the exercise bout, we ran a mixed ANOVA with one within-subjects factor (time: 10, 20, 30 min) and two between-subjects factors (group: maintainers, non-maintainers; and condition: aesthetic, enjoyment). Baseline affective valence was included as a covariate. The main effects of group [F(1, 67) = 0.20, η² = .003, p = .656] and condition [F(1, 67) = 0.11, η² = .002, p = .741] were not significant. In addition, the linear time by group [F(1, 67) = 2.40, η² = .035, p = .126] and linear time by condition [F(1, 67) = 0.41, η² = .006, p = .523] interactions were not significant. However, there was a significant linear time by group by condition interaction [F(1, 67) = 6.09, η² = .083, p = .016], such that maintainers in the enjoyment condition and non-maintainers in the aesthetic condition showed a linear increase in positive affect over time, maintainers in the aesthetic condition showed a linear decrease in positive affect over time, and non-maintainers in the enjoyment condition showed no increase or decrease (see Figure 2).

Figure 2. Affective valence means (at 10, 20 and 30 min) by group (maintainers vs. non-maintainers) and condition (enjoyment vs. aesthetic).

Post exercise measures

A 2 (group) x 2 (condition) factorial ANOVA was utilised to examine group differences in intrinsic and extrinsic motivation for exercise measured after the conclusion of the exercise bout. As anticipated, self-reported intrinsic motivation, as assessed by the interest/enjoyment subscale of the IMI, was significantly higher in maintainers (M = 5.56, SD = 0.92) relative to non-maintainers (M = 4.92, SD = 1.18) [F(1, 68) = 8.35, η² = .109, p = .005]. Intrinsic motivation was also marginally higher among participants assigned to the enjoyment condition (M = 5.42, SD = 0.91) relative to those assigned to the aesthetic condition (M = 5.13, SD = 1.23) [F(1, 68) = 2.90, η² = .041, p = .093]. Finally, there was a significant group by condition interaction [F(1, 68) = 9.52, η² = .123, p = .003] such that, as predicted, self-reported intrinsic motivation was significantly higher in non-maintainers assigned to the enjoyment condition relative to non-maintainers assigned to the aesthetic condition [F(1, 68) = 10.30, η² = .131, p = .002] but did not vary by condition for maintainers [F(1, 68) = 1.08, η² = .016, p = .303] (see Figure 3). Self-reported extrinsic motivation was marginally higher in those participants assigned to the enjoyment condition (M = 3.39, SD = 0.53) relative to participants assigned to the aesthetic condition (M = 3.17, SD = 0.41) [F(1, 68) = 3.74, η² = .052, p = .057] but did not vary by group [F(1, 68) = 0.66, η² = .010, p = .418], and the group by condition interaction was not significant [F(1, 68) = 0.18, η² = .003, p = .674].

Figure 3. Intrinsic motivation means by group (maintainers vs. non-maintainers) and condition (enjoyment vs. aesthetic).

Discussion

Given that a lack of motivation is one of the most commonly reported barriers to exercise participation (Brownson et al., 2001), we found it important to expand upon previous studies by exploring how motivational cueing impacts exercise behaviour in a controlled setting. To this end, we used visual cues (i.e., videos) to encourage participants to think more deeply about the source of their motivation and then assessed the impact of these cues on critical measures related to exercise initiation and maintenance. Perhaps the most interesting result of the current study was that of affective response to exercise, for which we had two general hypotheses. The first was that maintainers would have a more positive affective response to exercise across conditions. Though we saw significant variability in affective valence patterns for the different groups and conditions, this first hypothesis was, in general, not supported. Closer examination of the affective valence means at each time point (see Figure 2) reveals that, across conditions, maintainers were higher in self-reported positive affect only at the 10 min mark. Further, affective valence was approximately equal across the groups at 20 min, and non-maintainers reported, on average, more positive affect at 30 min than maintainers.

Our second hypothesis was that participants exposed to the intrinsic cue would show more positive responses across the exercise bout than participants exposed to the extrinsic cue. This prediction was partially supported in that maintainers assigned to the enjoyment condition showed a more positive affective response to exercise than did maintainers assigned to the aesthetic condition. In fact, the negative response of maintainers assigned to the aesthetic condition likely contributes to the aforementioned unexpected pattern of affective valence means. There are at least two potential explanations for this pattern of results. First, focusing on an extrinsic motivation like the appearance-related benefits of exercise may make affective response to exercise more negative. This is consistent with studies showing that providing extrinsic rewards for behavioural engagement can reduce intrinsic motivation (e.g., Deci et al., 1999). It is possible that this “undermining” phenomenon extends beyond extrinsic rewards and into the realm of motivational cueing and that the observed decreases in affective valence were caused by reductions in inherent intrinsic motivation among maintainers exposed to the extrinsic cue. Second, and perhaps more plausibly, results for affective response among maintainers could reflect changes in exercise intensity. Recall that a greater proportion of maintainers in the aesthetic condition opted to increase or keep their original speed at the halfway point, while a greater proportion of maintainers in the enjoyment condition opted to decrease their speed. Given past research suggesting reduced pleasure in the context of taxing exercise (Ekkekakis et al., 2008), it is perhaps logical that maintainers in the aesthetic condition would have a less positive affective response given that many of them increased or maintained the intensity with which they were exercising.

In the case of non-maintainers, those exposed to the extrinsic cue showed more positive affective responses to exercise than did those exposed to the intrinsic cue. Past investigations examining the factor structure and correlates of the Reasons for Exercise Inventory (Silberstein et al., 1988) may shed some light on this result. Cash et al. (1994) found management of weight and appearance to be the strongest behavioural motivator of exercise frequency among college-aged women. We speculate that the consistency of the aesthetic cue (with its focus on the appearance-related benefits of regular exercise) with their existing extrinsic motivation for exercise may have led non-maintainers to feel elevated or inspired.

Results for intrinsic motivation supported our hypotheses. That is, across conditions, maintainers reported significantly higher levels of post-exercise intrinsic motivation, relative to non-maintainers. These results are consistent with past studies suggesting that regular exercisers display higher levels of intrinsic motivation (Duncan et al., 2010). Additionally, across groups, post-exercise intrinsic motivation was marginally higher among participants assigned to the enjoyment condition relative to those in the aesthetic condition, as expected. Also consistent with predictions was the significant group by condition interaction, indicating that self-reported intrinsic motivation was higher in non-maintainers assigned to the enjoyment condition relative to non-maintainers assigned to the aesthetic condition but did not vary by condition for maintainers. However, the intrinsic motivation and affective valence results taken together are perplexing. It is unclear as to why non-maintainers assigned to the aesthetic condition would report steady increases in positive affect over time, while non-maintainers assigned to the enjoyment condition would report higher levels of post-exercise intrinsic motivation, especially given the established positive relationship between affect and intrinsic motivation (Isen & Reeve, 2005). It suggests that there are other important mechanisms driving motivation aside from affective response to exercise. Further research on this question is clearly warranted.

A surprising result is that, across groups, self-reported post-exercise extrinsic motivation was marginally higher in those participants assigned to the enjoyment condition relative to participants assigned to the aesthetic condition. While the script for the enjoyment condition was intended to encourage intrinsic motivation, some of the content pertained to enjoying activities and sports with friends, which may have also encouraged some of the more autonomous forms of extrinsic motivation (e.g., identified regulation). This could explain the elevated levels of extrinsic motivation in participants assigned to the enjoyment condition, but it does not account for the fact that extrinsic motivation was marginally lower in participants assigned to the aesthetic condition, the condition that was specifically designed to induce extrinsic motivation.

Despite the differences in affective response across the 30-minute treadmill bout, heart rate and perceived exertion did not vary by group or condition. Further, the percentage of participants who accepted the option of a speed change, a behavioural measure of intrinsic motivation, also largely did not depend on group or condition. There are at least two explanations for these results. First, it could be the case that videos watched before exercising are not a strong enough manipulation to reliably affect effort during an exercise bout. Past studies that have successfully manipulated motivation have presented cues throughout the relevant task (e.g., Scarapicchia et al., 2013). Relatedly, the pattern of results from the behavioural intrinsic motivation analysis suggests that maintainers in particular may not have responded behaviourally to the manipulation. It could also be the case that both videos were equally effective at manipulating exercise effort. Given significant group by condition interactions in other constructs of interest (e.g., affective valence), this possibility is perhaps less likely.

As with any study, some limitations need to be addressed. First, the current sample was comprised of young, introductory psychology students in the very fit and homogenous city of Boulder, Colorado, limiting generalizability. Second, to reduce demand characteristics, we only assessed motivation for exercise at post-test and thus cannot make causal assertions about the impact of our manipulation on intrinsic and extrinsic motivation. Third, it isn’t clear whether the short-term effects seen here will endure and contribute to the development of self-determined motivation.

Relatedly, this investigation took place in the context of a single exercise session. It could certainly be the case that the effects of our cues are better revealed in the context of a multi-session study or even long-term exercise involvement. In the case of affective response, for example, though positive affect was relatively stable for non-maintainers assigned to the enjoyment condition, we would expect steady increases in affective response across multiple sessions as non-maintainers gain exercise experience (as well as additional exposure to the intrinsic cue) and presumably transform the motivational content into self-determined motivation. Clearly, additional studies incorporating multiple exercise sessions and a longer follow-up to assess maintenance are needed to better understand the effects of our cues. Such research could also speak to the question of whether positive affect can be used as an index to distinguish the effects of intrinsic versus extrinsic cues in the context of exercise.

A final limitation relates to the content of the video scripts. The aesthetic script, for example, contains material that pertains to the dual motives of avoiding failure and approaching success, as put forth by self-worth theory (Covington, 1992). Phrases like “I lacked self-esteem and wanted to look better” relate to the motive of avoiding low contingent self-worth, while phrases like “As my muscle mass began increasing, my self-worth surged” relate to the motive of enhancing contingent self-worth. Unfortunately, without a more structured manipulation check inquiring about which features of the video certain participants found inspiring, it is difficult to assess how these dual motives may have impacted our findings. Additional studies employing stronger manipulation checks or additional assessments that carefully probe how participants respond to the various motivational content covered by the scripts are needed to shed light on this issue. Relatedly, and as previously mentioned, though the enjoyment script was intended to enhance intrinsic motivation, some of the content may also have enhanced the more autonomous forms of extrinsic motivation. For example, the phrase “In addition, [exercise] is consistent with my lifestyle” reflects integrated regulation, while “I never have to worry about being fit enough to complete the hike” reflects introjected regulation. Subsequent studies should aim to either revise the current scripts or create new scripts that embody content pertaining exclusively to either intrinsic or extrinsic motivation.

In conclusion, the current findings offer new information about the effects of intrinsic versus extrinsic motivational cueing on exercise behaviour. While the current manipulation did not differentially affect effort for maintainers versus non-maintainers during an exercise bout, it appears to have affected other constructs related to effort, particularly affective response to exercise. As past studies have shown that intrinsic motivation predicts exercise adherence (Segar et al., 2011), results from the current study contribute to a body of literature attempting to uncover and reduce the various barriers that prevent people from engaging in regular exercise. Stronger manipulations, though with similar content, might be effective in manipulating effort during an exercise bout.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Post-exercise measures of motivation further support this assumption, as self-reported intrinsic motivation was significantly higher among maintainers relative to non-maintainers.

2 Though we did not perform a formal qualitative analysis on the free-text data from the writing task, informal screening revealed that all participants provided an on-topic response and thus no exclusions were made.

3 The low minimum values for moderate- and vigorous-intensity weekly exercise minutes among maintainers are likely due to one or multiple individuals within this group reporting engaging exclusively in exercise of one intensity or the other (e.g., an individual reporting 0 weekly minutes of moderate-intensity exercise but 200 weekly minutes of vigorous-intensity exercise).

Additional information

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.