Participants. One hundred thirty-nine undergraduate students (34 women, 105 men) at the University of Chicago participated in this study in return for monetary compensation.
Procedure. This study employed a 2 (mood: positive vs. neutral) 2 (goal: health improvement vs. mood management) between-subjects design. An experimenter approached participants and invited them to participate in a study investigating whether demographic factors are related to overall fitness levels among students. Following a procedure that Muraven et al. (1998) developed, the experimenter took two handgrip measures of each participant, one before and one after the experimental manipulation.
Participants read that the handgrip task measures the strength of muscles of the forearm and hand and is a useful indicator of overall levels of fitness. They further read that recent evidence has suggested that overall muscle tone is related to healthfulness. The purpose of these instructions was to emphasize the long-term value of performing well on the task for a person’s health; that is, performing well on the handgrip task provides positive prognosis and helps maintain good health. The experimenter then assessed participants’ first handgrip measure. We chose a particularly tight handgrip for this task, and pretests indicated that although the task appeared interesting to participants before they attempted it, participants found the task to be difficult and unpleasant once they tried it. Specifically, participants were to squeeze the ends of the handgrip together and hold a piece of paper between the two ends. The experimenter measured time from the moment the experimenter applied the paper to the grip until the paper fell away from the grip. The experimenter noted the reading and explained that a second handgrip measure would be taken again after a while because the participant needed to rest. In the interim, participants completed two supposedly unrelated tasks to be used in future studies.
In order to induce mood, the first task was a word association task, in which participants provided their first associated word to each of ten positive words (e.g., “smooth,” “beautiful”), or neutral words (e.g., “desk,” “number”). This task has been shown to manipulate mood state effectively (e.g., Isen, Johnson, Mertz, & Robinson, 1985). Participants’ responses included associations such as “satin” and “soft” to the word “smooth” and “table” and “wood” to the word “desk.” After participants completed this task, they rated how interesting the task was (1 = not at all, 7 = very) and how they felt at that moment (1 = very bad, 7 = very good). We assumed that when the source of the mood is not apparent, rating it would not attenuate the actual experience.
Next, in order to prime the relevant health improvement or mood management goal, participants completed a supposedly unrelated, language survey, which consisted of a scrambled-sentence goal-priming task. The scrambled-sentence test (Bargh & Chartrand, 1999; Srull & Wyer, 1979) activated nonconsciously a self-improvement or a mood management goal. By employing a nonconscious priming procedure (as opposed to conscious priming, as in Studies 1 and 2), we attempted to demonstrate the effect of mood on performance in self-control tasks outside participants’ conscious awareness of their accessible goals.
The task was to decipher eight scrambled sentences, which activated either health improvement goals (e.g., “aim for better shape,” “fitness is a virtue”) or mood management goals (e.g., “aim for most pleasure,” “cheerfulness is a virtue”). Each sentence consisted of five words. Participants had to form a grammatically correct four-word sentence from the five words presented in a scrambled order and to circle any sentence that took more than 30 seconds to unscramble. All participants completed this task successfully without circling any sentences (see appendix for a complete list of 4-word priming sentences).
After completing both tasks, participants completed the second part of the handgrip task; again, as experimenter measured their times for squeezing the handgrip device. They were then debriefed and dismissed. None of the participants identified the relationships among the different parts of the experiment.
Results and Discussion
Manipulation checks. A 2 (mood: positive vs. neutral) 2 (goal: health improvement vs. mood management) ANOVA conducted on the self-report of mood measure revealed only a main effect of mood (M = 4.72, SD = 1.43 for happy participants vs. M = 4.23, SD = 1.50 for neutral participants), F(1, 135) = 3.92, p < .05 (all other Fs < 1). A similar mood goal ANOVA conducted on the interestingness rating of the word association task did not reveal any significant effects (all Fs < 1). These results suggest that the word association task was equally interesting across conditions, but participants who provided their first associated word to each of ten positive words reported feeling happier than participants who provided their first associated word to each of ten neutral words.
Performance on handgrip task. The times participants spent squeezing the handgrip were positively skewed; therefore, we log transformed the times using natural log transformation (Bargh & Chartrand, 2000; Fazio, 1990). There was no significant difference in the initial response times across conditions. To assess the effect of the manipulations, we computed changes scores by subtracting each participant’s initial handgrip duration time from his or her final duration time. A 2 (mood: positive vs. neutral) 2 (goal: health improvement vs. mood management) ANOVA conducted on the changes scores revealed a main effect of mood, indicating greater improvement in performance for happy participants than for neutral participants (M = 14.76 sec, SD = 42.18 vs. M = 1.44 sec, SD = 41.62), F(1, 135) = 10.03, p < .01. In line with our hypothesis, this main effect was qualified by the predicted interaction between mood and goal, F(1, 135) = 4.68, p < .04 (see Figure 3). Planned contrasts indicated that happy participants who held an accessible health improvement goal showed improvement in performance at the handgrip task relative to neutral participants with the same goal (M = 25.49 sec, SD = 43.63 vs. M = –6.17 sec, SD = 45.95), t(135) = 3.75, p < .01. However, happy participants who held an accessible mood management goal showed no improvement relative to neutral participants who held a mood goal (M = 3.72 sec, SD = 38.16 vs. M = 8.62 sec, SD = 36.29), t(135) < 1.2
These data support our hypothesis that positive (vs. neutral) mood facilitates performance on a self-control task only when the relevant health improvement goal is accessible, but when happy people strive to maintain their positive mood, performance drops back to baseline levels, suggesting that when the task was incompatible with the goal, there was no advantage for happy (vs. neutral mood) participants, who all avoided the task to the same extent. Notably, increased adherence to accessible goals among happy (vs. neutral) participants did not seem to require conscious awareness of the goals, which we primed nonconsciously in this study.
Studies 1–3 explored the effect of positive mood on adherence to subsequently primed goals, but we did not design them to distinguish between the effect of mood on noticing goals (i.e., attending to contextual cues for the goals) and the effect of mood on adherence to these already recognized goals. To assess whether positive mood directly facilitates goal adherence, another study tested the effect of positive mood on the pursuit of preexisting goals (i.e., goals that were primed before inducing mood).
In contrast with our previous studies, in Study 4, we primed goals before we induced mood, allowing us to rule out the possibility that positive mood only enables people to notice the primed goal. If this is indeed the case, mood should not have an effect if it is evoked after a goal is primed. Otherwise, as we predicted, positive mood should increase goal adoption and thus promote success at self-control toward preexisting goals. This study used a modified version of Raghunathan and Trope’s (2002) self-control task to test whether participants in a positive mood attend to information on the health consequences of drinking caffeine. Raghunathan and Trope induced a positive or negative mood among participants (caffeine users vs. nonusers) and then asked the participants to read an essay on the adverse effects of caffeine consumption. These authors reported that happy participants recalled more negative information than did unhappy participants but only if they deemed the information to be self-relevant, that is, they were caffeine users. For our study, we recruited only caffeine drinkers, and therefore all our participants are similar to Raghunathan and Trope’s participants in the self-relevant condition. We expect to extend our previous results insofar as happy participants should spend more time recalling and should ultimately recall more information than should unhappy participants when primed with a health goal but not when primed with a mood management goal.