Study 2: Rumination and reward consumption

The second study aimed at investigating a specific mechanism that is expected to have an important influence on depressed individuals’ behavior during reward consumption:

rumination. This quasi-experimental study compared two groups (dysphorics vs.

nondysphorics) during consumption. In accordance with previous studies (Franzen &

Brinkmann, 2015c), our first hypothesis suggests that dysphoric individuals would report and show reduced affective responses to reward. Our second and main hypothesis suggests that rumination would mediate the link between dysphoria and facial expressions. Specifically, we postulate that dysphoric individuals would show reduced facial expressions during reward consumption because they have the tendency to ruminate. In contrast, we postulate that nondysphoric individuals would show more positive facial expressions during reward

128 Partie empirique: Studies 5 and 6: Mechanisms consumption because they are less prone to rumination. Stated differently, for dysphoric individuals the negative valence of rumination is expected to interfere with the positive valence of the reward information.

5.3.1. Method

Participants. Participants in the second study were university students, and the final sample consisted of 31 participants, including 23 women and 8 men aged between 19 and 37 (M = 23.94, SD = 4.93). Fifteen participants were referred to as nondysphorics because they scored in the lower quartile of the CES-D (M = 5.6, SD = 2.95). Sixteen participants were referred to as dysphorics because they scored in the upper quartile of the CES-D (M = 30.94, SD = 8.31).

For the selection of participants, we used the same procedure as the in Study 1. Ninety-nine potential participants first answered an online version of the CES-D. Only participants who scored in the lower (≤ 10) or in the upper (≥ 16) quartile of the current CES-D distribution were invited to the experimental session, and the CES-D was assessed a second time during the experimental session. Of the 39 students who participated in the experimental session, only those whose CES-D scores stayed within the limits set by the initial distribution were kept for analyses (N = 33). Additionally, two participants were further excluded due to extremely high EMG reactivity of more than 3 SDs above overall means.

Experimental task and reward manipulation. Composed of 14 trials, the experimental task was adapted from the classic Sternberg task (Sternberg, 1966). For each trial, a varying number of black letters were successively presented in the middle of the screen for 1 second. Then, a target letter was presented in blue font. Participants had to decide whether this target letter had been part of the list of black letters presented previously by pressing one of two specified keys. Following this response, the message “answer recorded”

was presented in the middle of the screen. If participants did not answer after 2 seconds, the message “please answer more quickly” appeared in the middle of the screen for 1 second.

Then, a new trial began.

In order to create a task with an unclear level of difficulty in accordance with the predictions of motivational intensity theory (Brehm & Self, 1989; see also Study 1), the performance standard was given only at the end of the task. Moreover, each trial was composed of a varying number of letters (from 3 to 9 letters). Finally, participants were only informed about the general procedure of the task but were not given any further details, such as the number of trials, the number of letters in each trial (i.e., the difficulty), or the total duration of the task, which was about 5 minutes.

Partie empirique: Studies 5 and 6: Mechanisms 129 Regarding the reward manipulation, participants were informed that they could win 10 Swiss Francs (about 10 USD) at the end of the experiment if their performance was equal to or greater than the performance standard to be announced after the task. An individual performance standard was secretly created for each individual, and the reward was given to all participants in order to have a measure of affective responses for all participants. The computer calculated the individual performance standard by subtracting 2 from each participant's number of correct responses.

Facial EMG measures. In order to measure affective responses to reward, muscular measures were assessed during the 10-second period when participants were told that they had won the 10 Swiss Francs. The facial EMG signal was continuously sampled at 1000 Hz using MindWare Technologies LTD, Gahanna, OH. Three solid gel electrodes (3 SG3-N, Multi Bio Sensors Inc., MedCaT B.V, Netherland) were used, two of them for the activity of the zygomaticus major muscle and one isolated ground electrode which served as a reference.

The first electrode was situated in the middle of an imaginary line extending from the corner of the lip at rest (i.e., cheilion) to the corner of the ear (i.e., ipsilateral condylion). The second electrode was placed approximately 1 cm further along this imaginary line. The third electrode was attached to the forehead on the edge of the hair line, an electrically inactive site (Fridlund

& Cacioppo, 1986). Electrical activity was measured using a bipolar recording, and the EMG signal was amplified with a constant gain of 1000, as recommended by Fridlund and Cacioppo (1986). In order to maximize the signal-to-noise ratio, the EMG signal was filtered with a 10-500Hz passband, as recommended by Tassinary and colleagues (Tassinary et al., 2007).

Self-reported pleasure. We measured subjective affective responses to reward with a self-report question (“To what extent the reward you received causes pleasure in you?”). The participants answered this question on a visual analogue scale ranging from 0 (no pleasure) to 100 (a lot of pleasure).

Depression questionnaire. In order to assess current depressive symptomatology, we used the CES-D as in Study 1 (Cronbach’s α = .96 during the experimental session).

Rumination. A French version of the Ruminative Responses Scale (RRS; Nolen-Hoeksema & Morrow, 1991) was used in order to measure dispositional tendencies to ruminate (Cronbach’s α = .94). This scale is composed of 22 items describing peoples’ responses to depressed mood that are focused on the self, on their symptoms, and on the possible causes and consequences of their negative mood. Each of the 22 ruminative thoughts or behaviors is

130 Partie empirique: Studies 5 and 6: Mechanisms answered on a scale ranging from 1 (almost never) to 4 (almost always). The total score was calculated by summing all items.

Procedure. The procedure of the present study was similar to the one in Study 1, with the exception that EMG measures and self-reported liking were assessed instead of cardiovascular measures and self-reported wanting. This second study began with potential participants answering the online CES-D. During the following experimental session, participants were welcomed and seated and filled in a consent form. The experimenter applied the sensors for the assessment of facial EMG measures and left the room in order to monitor the experiment from an outside control room. Participants were asked to answer two questionnaires (the CES-D and the RRS) ostensibly for an unrelated questionnaire validation study. When they had finished, the experimenter reentered the room, started the next part of the experimental session, and left the room. First of all, the participants read introductory information of the study, and answered some socio-demographic questions. Afterwards, participants watched an excerpt of a neutral documentary movie, which was presented for the sake of applying the same procedure as in Study 1. They read instructions about the task and the reward and worked on the 5-min. Sternberg task. After the task, participants were informed that they had obtained the reward and they evaluated the affective value of the reward (i.e., subjective measure of liking). Finally, the experimenter reentered the room, removed the electrodes, and thanked and debriefed the participants.

Data analyses. For electromyographic recordings, two reactivity scores were computed for the zygomaticus major muscle: A mean score and a maximum score. For the mean score, a baseline mean score was computed as the arithmetic mean of all data points assessed just after the task during the 2 seconds before participants were informed they had obtained the reward. A reward mean score was created by averaging all data points assessed during the first 2 seconds of the period when participants were told they had obtained the reward. The muscular reactivity mean score was then computed as the difference of the reward score minus the baseline score. For the maximum score, the procedure was exactly the same, the only difference being that instead of averaging all data points, we took the maximum value of each period to calculate the baseline maximum score, the reward maximum score, and the muscular reactivity maximum score (Fridlund & Cacioppo, 1986; Tassinary et al., 2007).

Concerning our hypotheses, mean and maximum reactivity scores are comparable, therefore, both are considered primary EMG measures. Independent samples t-tests were performed to compare dysphoric and nondysphoric participants on all physiological and self-report measures.

Partie empirique: Studies 5 and 6: Mechanisms 131 For the mediation hypothesis, the same mediation models as in the first study were computed using the same procedures (PROCESS macro for SPSS; Preacher & Hayes, 2008):

the direct effect procedure, the nonparametric bootstrapping procedure, and the Sobel test.

Simple mediation regression analyses were used in order to examine the mediating effect of rumination between dysphoria (dichotomous predictor variable) and facial expressions (i.e., zygomaticus major muscle reactivity; outcome variable).

5.3.2. Results

Preliminary analyses. Preliminary analyses revealed no significant group differences neither for the EMG mean baseline measure, t(29) = -1.44, p = .16, nor for the EMG maximum baseline measure, t(29) = -0.08, p = .94. These results demonstrate that dysphorics and nondysphorics did not significantly differ on muscular baseline measures. Means and standard errors of these measures are presented in Table 12.

Table 12

Means and Standard Errors of Muscular Baseline and Reactivity in Study 2

Baseline M (SE) Reactivity M (SE)

EMG mean EMG max EMG mean EMG max

Nondysphorics 0.0573 (0.1806) 0.1806 (0.0315) 0.0004 (0.0004) 0.0128 (0.0049) Dysphorics 0.1216 (0.0272) 0.3816 (0.0786) -0.0018 (0.0018) -0.0268 (0.0098)

Note. EMG is indicated in microvolts.

Reward consumption analyses. Regarding the objective measures of the consummatory component, an independent samples t-test revealed a significant result for EMG maximum reactivity, t(29) = 3.53, p = .001, ηp2 = .31. This result confirms our predictions about a lower EMG maximum reactivity in the dysphoric group (M = -0.03, SE = 0.04) in comparison to the nondysphoric group (M = 0.01, SE = 0.02), as illustrated in Figure 12A. The t-test for EMG mean reactivity was not significant, t(29) = 1.20, p = .24, contrary to our hypothesis suggesting a higher EMG mean reactivity in the nondysphoric group (see Table 12 for means and standards errors of muscular reactivity measures).

Concerning the self-reported liking measure, results revealed a significant difference, t(29) = 2.65, p = .01, ηp2 = .19. As illustrated in Figure 12B, these results indicate that dysphoric individuals reported less pleasure when they received the reward (M = 51.81, SE = 4.11) than did nondysphoric individuals (M = 70.27, SE = 5.69).

132 Partie empirique: Studies 5 and 6: Mechanisms

Means and standard errors of (A) EMG maximum reactivity and (B) self-reported liking during reward consumption in Study 2

Rumination and mediation analyses. The analysis concerning the rumination questionnaire revealed a significant group difference, t(29) = -4.04, p < .001, ηp2 = .37. As illustrated in Figure 13, nondysphoric individuals (M = 34.20, SE = 2.75) indicated lower trait rumination than dysphoric individuals (M = 50.81, SE = 3.03).

Figure 13

Means and standard errors of rumination in Study 2

Concerning the mediating effect of rumination between dysphoria and facial expressions (maximum score), results of the direct effect procedure showed that the conditions suggested by Baron and Kenny were all met (see Table 13 for the results of paths A, B, C, and

A B

Partie empirique: Studies 5 and 6: Mechanisms 133 C'). In addition, results of both the Sobel test analysis and the bootstrapping method showed a significant mediating effect of rumination.

Table 13

Impact of rumination (mediator) between dysphoria (predictor variable) and facial expressions (outcome variable) with unstandardized regression coefficients (β) of Paths A, B, C, and C', Sobel test statistics, and bootstrapping procedure statistics in Study 2

Path A Path B Path C Path C' Sobel test CI (95%)

β SE p β SE p β SE p β SE p z p Lower Upper

16.61 4.11 < .001 -0.0014 0.0004 < .01 -0.0396 0.01 .001 -0.0169 0.01 .18 -2.3961 .02 -0.0447 -0.0072

Note. CI = confidence intervals; The nonparametric bootstrapping procedures are based on 5000 bootstrapped re-samples

5.3.3. Brief discussion

The aim of Study 2 was to investigate the mediating impact of rumination on the relationship between dysphoria and facial expressions. First of all, result confirmed previous studies (e.g., Franzen & Brinkmann, 2015c) by demonstrating that, in comparison to nondysphoric individuals, dysphoric individuals reported reduced pleasure and showed reduced facial EMG reactivity when they received the reward. Secondly, the results are in accordance with past research (Nolen-Hoeksema, 2000) by demonstrating that dysphoric individuals are characterized by higher trait rumination. Finally, the mediation analyses with the direct procedure, the bootstrapping method, and the Sobel all converge on the conclusion that rumination fully mediates the relationship between dysphoria and facial expressions.

Taken together, these results substantiate the importance of rumination as a psychological mechanism that underlies the phenomenon of reduced affective responses during reward receipt in dysphoria. More specifically, during the reception of a reward, dysphoric individuals’ tendency to ruminate interferes with the positive valence of the reward information. As a result, the affective responses (i.e., facial expressions of pleasure) are attenuated for dysphoric individuals.

5.4. General discussion

The two present studies aimed at investigating the mediating impact of two psychological mechanisms on the relationship between dysphoria and incentive responsiveness during anticipation and consumption. Specifically, the aim of Study 1 was to test if locus of control played a role for the relationship between dysphoria and effort

134 Partie empirique: Studies 5 and 6: Mechanisms mobilization. Study 2 focused on the impact of rumination on the relationship between dysphoria and facial expressions.

Results of Study 1 confirmed that compared to nondysphoric individuals, dysphoric individuals showed a reduced PEP reactivity during punishment anticipation. Results of our secondary dependent variables (SBP, DBP, and HR) were not significant. Moreover, the results confirmed our hypothesis concerning self-reported wanting by demonstrating that dysphoric and nondysphoric individuals showed the same motivation to avoid the punishment.

These results are in line with previous studies focusing on dysphoric individuals in the framework of motivational intensity theory (Brinkmann et al., 2009; Franzen & Brinkmann, 2015b, 2015c). Moreover, results of Study 1 demonstrated that locus of control played an important role for the reduced cardiovascular responsiveness during punishment anticipation.

Taken together, these results suggest that dysphoric individuals are motivated to avoid a punishment, like nondysphoric individuals. However, because their locus of control is less internal and more controlled by chance or powerful others, they develop an inappropriate response to avoid the punishment. Indeed, they show a passive avoidance behavior, which is manifested by a reduced cardiovascular response during punishment anticipation. In contrast, nondysphoric individuals show an adaptive active avoidance behavior, which is manifested by a high cardiovascular response during punishment anticipation because their locus of control is more internal and less controlled by chance or powerful others.

Results of Study 2 demonstrated that in comparison to nondysphoric individuals, dysphoric individuals showed a lower zygomaticus major muscle reactivity and reported less pleasure during reward consumption. These results thus confirm the notion of subjectively and objectively reduced affective responses to reward in dysphoria. Our findings are in line with previous studies run with dysphoric individuals and focusing on facial expressions in response to emotional material (Sloan et al., 2001, 2002). The present results also replicate those of a previous study investigating the anticipatory and the consummatory components of reward and punishment processing in a dysphoric sample (Franzen & Brinkmann, 2015c). Moreover, the results of Study 2 on the psychological mechanism of rumination confirm our predictions.

Indeed, the results demonstrate that rumination plays an important role for the relationship between facial expressions and dysphoria during reward consumption. Taken together, these findings show that because of their tendency to ruminate, dysphoric individuals have a reduced facial expression of pleasure during reward reception. In contrast, because they are less prone to rumination, nondysphoric individuals show normally happy facial expressions during reward reception.

In sum, these two studies focused on two important psychological mechanisms: locus of control (Study 1) and rumination (Study 2). It is of note that these mechanisms do not exhaustively explain why depressed individuals show a reduced responsiveness to reward and

Partie empirique: Studies 5 and 6: Mechanisms 135 punishment. Other mechanisms may contribute to this phenomenon as well. However, we focused on these specific mechanisms because of their important role in literature and in the theories of depression (e.g., Abramson et al., 1978; Beck et al., 1979; Nolen-Hoeksema, 2000).

The results of these two studies revealed for the first time that these psychological mechanisms explain a part of the variance of reward and punishment responsiveness. Future studies might continue the present line of research by investigating the two remaining components of reward and punishment processing, which are punishment consumption and reward anticipation.

Moreover, rumination and locus of control have been assessed as traits in the two present studies. In future studies the present paradigm should be extended by the investigation of state rumination and state locus of control.

By investigating the mechanisms underlying specific impairments of depressed and dysphoric individuals in punishment anticipation and reward consumption, the present studies are a first step to build the basis for specific and targeted therapies to cope with these impairments. Concerning reward consumption, positive psychotherapy can be used in order to directly increase affective responses to reward. This method aims to elicit positive emotions, to build engagement in work, intimate relations, or leisure, and to find one’s own strengths and talents (Seligman et al., 2006). Moreover, rumination-focused cognitive behavioral therapy (Watkins et al., 2007) is an intervention focusing on a cognitive-behavioral treatment that explicitly targets depressive rumination. Based on our findings, treating rumination can be useful to indirectly improve affective responses to reward during consumption. Concerning punishment anticipation, behavioral therapy (Carvalho & Hopko, 2011) can have an impact on the specific impairment of passive avoidance behavior. In this intervention, the individual identifies a problematic behavior and learns to find an alternative behavior, which would be more adaptive in the given situation. Moreover, cognitive therapy (Beck et al., 1979) helps people to develop skills for modifying inappropriate beliefs and distorted thoughts. This intervention is thus suitable for depressed individuals who can learn to modify their beliefs about the control they have over their actions and over the consequences of their actions. In this way, this particular method can help them to develop a more internal locus of control and, indirectly, to respond in a more appropriate manner to the anticipation of a punishment.

In conclusion, the results of the present studies underline the importance of investigating the psychological mechanisms underlying reward and punishment responsiveness in dysphoria and depression. Study 1 demonstrated that locus of control can explain why dysphoric individuals show a passive behavior in anticipation of a punishment, even if they are motivated to avoid the punishment. Study 2 showed that rumination is an important mechanism that can explain why affective responses during reward reception are reduced in dysphoria.

136 Partie empirique: Study 7: Reward and punishment sensitivity in a clinical sample

6. Study 7: Reward and punishment sensitivity in a clinical