Experiment 1

3.3.3.1. Materials and Methods 3.3.3.1.1. Participants

Twenty-four female undergraduates of the University of Geneva were recruited. They were all right-handed (mean of Edinburgh Handedness Inventory = 81.26, SD = 15.52), healthy, had normal or corrected-to-normal vision, and ranged in age from 18 to 28 years (M = 21.42, SD = 2.13). Participants were paid 30 CHF plus the bonus money they won.

3.3.3.1.2. Experimental Paradigm and Procedure

After arrival at the laboratory, participants read and signed an informed consent form.

They further filled out questionnaires about their current health and demographic characteristics. The practice session and the experimental gambling task took place in a sound-attenuated room; both tasks were presented on a computer screen (17, resolution 1280

 1024) and contained only gray characters against a black background (Figure 3A). The distance between participants’ eyes and the computer screen was 60 cm. Participants used the numeric keyboard of a standard PC keyboard to give their choices for each gambling trial.

The experimental gambling task started with 30 Swiss francs (CHF) because this was the minimal guaranteed amount of money participants were paid for their participation. The gambling task consisted of the following course of events within one trial, shown in Figure 3A. Each trial started with a fixation cross (randomized duration between 400 and 800 ms;

1° high, 1° wide) in the center of the screen, followed by three circles (Figure 3A, screen

“selection of circle”; 3.8° high, 4.6° wide). Participants were told that the possible outcomes of a trial (a win, +0.25 CHF; a loss, −0.25 CHF; and a break-even, 0 CHF) were concealed under these three circles; no cues were provided that allowed the participants to tell where the win was hidden. The circle that the participant had chosen was highlighted (500 ms) before the feedback stimulus appeared at its center (Figure 3A, screen “feedback”; 500 ms).

CHAPTER 3.3:APPRAISAL PATTERNS IN FACIAL EXPRESSIONS 112

Figure 3. An example of a gambling task trial of (A) Experiment 1 and (B) Experiment 2.

The presentation times of each trial event are indicated below the corresponding screen. At feedback onset, the information about the goal conduciveness and the power checks was simultaneously presented via filled or nonfilled geometric shapes. RT = reaction time. A = accepting. R = rejecting. See text for details.

Feedback stimuli simultaneously conveyed information about the goal conduciveness check (i.e., the gambling outcome: wins [goal conducive event], losses [goal obstructive event], and break-even outcome [in-between condition]) and the power check (i.e., choice about the gambling outcome at the end of a trial: choice [high power] and no choice [low

CHAPTER 3.3:APPRAISAL PATTERNS IN FACIAL EXPRESSIONS 113 power]). Consequently, the six feedback-stimuli conditions were “win, high power”, “loss, high power”, even, high power”, “win, low power”, “loss, low power”, and “break-even, low power”. Feedback stimuli were geometric shapes, a hexagon, a square, and a diamond with either a grey colored or a black colored fill. The geometric shapes encoded the three levels of the goal conduciveness check manipulation, whereas the gray and black colored fills encoded the two levels of the power check manipulation. In Table 1 an example illustrates the manipulation of the appraisal checks. Feedback-stimulus probability was balanced for all feedback conditions, with equal probability and without replacement across all trials.

Table 1

Example of the Feedback Stimuli for the Operationalization of the Goal Conduciveness and the Power Checks

Note. GC = goal conduciveness appraisal check; Power = power appraisal check. The associations of the geometric shapes with the levels of GC as well as the meaning of filled and nonfilled shapes with the levels of Power were counterbalanced across participants.

Experiment 2 GC

In the gambling task, after the feedback-stimulus presentation the screen turned black (1 s). Next, participants were presented with the choice outcome options for that trial. At this point, the screen had one letter to the left and one to the right (Figure 3A, screen “choice about outcome”; A = accept, R = reject; 0.8° high, 6.6° wide; Arial font, size 28). In high

CHAPTER 3.3:APPRAISAL PATTERNS IN FACIAL EXPRESSIONS 114 power trials, participants could freely choose between two options: accepting (A) or rejecting (R) a win, a loss, or a break-even outcome (presentation of “A R” or “R A”: randomized order with the same number of presentations). In low power trials, they had to accept the chosen choice option of either rejecting (presentation of “R R”) or accepting (“A A”) the outcome (randomized selection with the same number of presentations). Then, participant’s decision was accentuated by highlighting the corresponding letter (Figure 3A, screen

“monetary outcome”; 500 ms; Arial font, size 52 bold); simultaneously, the monetary outcome of that trial was shown between these two letters. At the end of each trial, the total monetary outcome was updated to the amount of money won (+0.25 CHF) or lost (−0.25 CHF), or remained unchanged for break-even outcomes. Immediately after, the next trial started. In total, the gambling task consisted of 384 gambling trials (divided into 12 blocks of 32 trials lasted) and lasted about 30 minutes.

Before playing the gambling task, participants completed a practice gambling session (48 trials, 5–7 min) in order to apprehend the gambling rules as well as the meaning of the gray or black filled shapes (Table 1). Practice trials were identical to experimental trials, except for a longer duration of the feedback-stimulus (900 ms), the presence of explanatory labels (presented with each feedback-stimulus: “win, choice”; “win, no choice”; “loss, choice”; “loss, no choice”; “none, choice”; or “none, no choice”), and the amount of play money (started with 0 CHF instead of 30 CHF). Participants were informed that no explanatory labels would be provided during the gambling task. To make sure that

participants responded correctly during the practice session, a performance cutoff criterion for high power trials was implemented (>80% of correct responses, i.e., accepting wins and rejecting losses). If the criterion was exceeded at the end of the practice session, the

experimental gambling task started after a short break; otherwise, a second practice session was run.

CHAPTER 3.3:APPRAISAL PATTERNS IN FACIAL EXPRESSIONS 115 The amount of bonus money won during the experiment relied upon the participant’s gambling performance. Participants were instructed to win as much money as they could, the maximum bonus amount possible (16.75 CHF) was not mentioned. They were assured that they would not end up losing money (getting less than 30 CHF) or owing money to the experimenter. They were not informed that the type of feedback was selected at random on each trial; they were only told that they would play a gambling game. At the end of the experiment, participants were informed about the experimental manipulations, and were paid their participation fee (30 CHF) plus the bonus money.

3.3.3.1.3. Data acquisition

The practice session and the experimental gambling task including the behavioral data acquisition were controlled by E-Prime 2.0 (Psychology Software Tools, Inc., Pittsburgh, PA). Facial EMG (placement of surface electrodes over the frontalis, corrugator, and cheek region according to guidelines, Fridlund & Cacioppo, 1986) and EEG¹⁵ (64 channel electrode cap) were recorded, and digitized (bandwidth 0.1 to 417 Hz, sampling rate: 2,048 Hz) with a BioSemi Active-Two amplifier system (BioSemi Biomedical instrumentation, Amsterdam, the Netherlands).

3.3.3.1.4. Data analysis

Preprocessing of the EMG data (Brain Vision Analyzer software, Brain Products, Gilching, Germany) followed the standard procedure (Fridlund & Cacioppo, 1986). First, bipolar montages were calculated of each pair of electrodes of each facial muscle region (frontalis, corrugator, and cheek region) by subtracting the recorded activity of one electrode with the activity of the neighboring electrode. Next, the continuous waveforms of the EMG data were bandpass filtered (20–400 Hz, 12 db/octave), full-wave rectified, low pass filtered (40 Hz, 12 db/octave), and cut into segments (including 500 ms baseline and 1.4 s

5 EEG results are reported in details elsewhere (Gentsch, Grandjean, & Scherer, 2013a; chapter 3.1)

CHAPTER 3.3:APPRAISAL PATTERNS IN FACIAL EXPRESSIONS 116 stimulus intervals) for each experimental condition (viz., “win, high power”, “loss, high power”, “break-even, high power”, “win, low power”, “loss, low power”, and “break-even, low power”). Then, the EMG data were down sampled to 512 Hz and exported to a

commercial software package (MATLAB R2012a, The MathWorks Inc., Natick, MA, 2012).

Separately for each facial muscle region, artifacts and outliers (deviating more than two standard deviations from the mean baseline activation of a given participant) were eliminated (2.24% of the trials of the entire EMG data). To examine the temporal profiles of facial EMG for 1.4 s after each stimulus, mean amplitude values for the subsequent 100-ms time intervals were calculated as a percentage change of the mean amplitude value of the baseline.

3.3.3.1.5. Statistical analyses temporal profiles of appraisal check effects

The EMG data of each facial region was submitted to a 2 (Power: high vs. low) × 3 (Goal conduciveness: loss vs. win vs. break-even) × 14 (Time: 100-ms time-intervals)

repeated measures ANOVA with Greenhouse-Geisser correction (cf. Vasey & Thayer, 1987).

The epsilon-values of the Greenhouse-Geisser correction are reported as ε. The degrees of freedom are reported with their uncorrected values. If p-values were adjusted to the Greenhouse-Geisser correction, they are labeled accordingly with pGG. All reported effect sizes are partial ŋ² (in the results simply noted as ŋ²). All tests were performed at an alpha level of 5%, and they were computed using IBM SPSS Statistics 19.

3.3.3.2. Results

Regarding prediction (1) that results of the goal conduciveness check and the power check trigger facial expressions, according to the component patterning theory and recent findings, we expected main effects of the goal conduciveness check over the corrugator and the cheek region. Main effects of the power check were expected over the frontalis, the corrugator, and the cheek region. Regarding prediction (2) that efferent effects of the goal conduciveness check precede efferent effects of the power check, goal conduciveness effects

CHAPTER 3.3:APPRAISAL PATTERNS IN FACIAL EXPRESSIONS 117 were predicted to start earlier (at around 800 ms after feedback-stimulus onset) than effects of the power check (starting earliest at around 900 ms). To test this predicted sequence of the temporal unfolding of appraisal check effects in facial EMG, comparisons were computed for every 100-ms time interval after feedback-stimulus onset. Although the Component Process Model makes no precise predictions about the nature of cumulative effects (i.e., of interaction effects) between appraisal checks and given that a previous study on sequential efferent effects of appraisal checks found interaction effects (Aue et al., 2007), the possibility of obtaining them was not ruled out. However, no direct a priori hypotheses were formulated about interaction effects of Goal conduciveness × Power ×Time or Goal Conduciveness

× Power. For each facial region, one three-factorial ANOVA was calculated; the results are summarized in Table 2.

Table 2

Results of the Repeated Measures ANOVA for each Facial Region of Experiment 1 Frontalis region Corrugator region Cheek region Note. N = 24. For each muscle region a Greenhouse-Geisser adjusted repeated measures ANOVA was performed with the within-subject factors of Goal conduciveness (GC: win vs.

loss vs. break-even outcome), Power (free vs. no choice about GC), and Time (Time: 14 post-stimulus 100-ms time intervals).

†p < .10. *p < .05. **p < .01. ***p < .001.

CHAPTER 3.3:APPRAISAL PATTERNS IN FACIAL EXPRESSIONS 118 3.3.3.3. Activity over the frontalis region

The three-factorial ANOVA solely yielded a main effect of time, F(13, 299) = 3.30, p_GG = .041, ŋ² = .13, ε = .168, and no significant interaction effects of Goal conduciveness, Power, and Time, or significant main effects of goal conduciveness and power (F-values <

.1.31, pGG -values > .231). This result indicates that the muscle activity over the frontalis region changed over time but that it was not significantly affected, as predicted, by efferent effects of the manipulated appraisal checks. Thus, over the frontalis region, no empirical support for the three predictions was found.

3.3.3.4. Activity over the corrugator region

The three-factorial ANOVA revealed a marginal significant interaction effect of Goal Conduciveness × Time, F(8, 184) = 1.91, p_GG = .076, ŋ² = .08, ε = .251, which suggests an unfolding of effects over the corrugator region. Furthermore, the ANOVA yielded a

significant main effect of goal conduciveness, F(2, 46) = 8.33, p_GG = .001, ŋ² = .27, ε = .920, and a marginal significant effect of time, F(4, 92) = 2.49, pGG = .078, ŋ² = .10, ε = .199. No significant main or interactions effects of the appraisal checks were found (F-values < 1.47, p_GG -values > .192). Notwithstanding, these ANOVA results support the prediction that results of the goal conduciveness check affect the corrugator region and that they unfold over time. However, these ANOVA results provide neither empirical support for the prediction that the power check results affect the corrugator region, nor that they unfold over time, nor that they show cumulative effects with the goal conduciveness check.

To follow-up the goal conduciveness check effects, post hoc comparisons on the goal conduciveness main effect were calculated. They showed that the corrugator activity was higher in response to losses (M = 0.66, SD = 3.44) in comparison to wins (M = -1.52, SD = 3.83), t(23) = 3.31, p = .003. Corrugator activity was also higher in response to break-even outcomes (M = 0.38, SD = 2.70) compared with wins, t(23) = 3.67, p = .001. Muscle activity

CHAPTER 3.3:APPRAISAL PATTERNS IN FACIAL EXPRESSIONS 119 was similar in response to losses and break-even outcomes, t(23) = 0.51, p = .616. This

response patterning of the goal conduciveness check over the corrugator region supports the prediction of the component patterning theory. Furthermore, this response patterning is compatible with other facial EMG studies that found patterns of valence over the corrugator region (e.g., Cacioppo, Petty, Losch, & Kim, 1986; Dimberg & Karlsson, 1997; Larsen, Norris, & Cacioppo, 2003). These studies demonstrated that corrugator activity is increased following negatively and attenuated following positively evaluated events.

In order to identify the time interval when results of the goal conduciveness check started to unfold over the corrugator region, the significant interaction effect was followed-up between Goal conduciveness × Time. Therefore, a one-factorial ANOVA was calculated for each 100-ms time interval after feedback-stimulus onset. The unfolding of goal

conduciveness check effects is illustrated in Figure 4. The first goal conduciveness check effect was found at 400 ms after feedback-stimulus onset, F(2, 46) = 3.56, pGG = .040, ŋ² = .13, ε = .934. Post hoc tests in this time interval revealed that corrugator activity was higher following break-even outcomes than following wins, suggesting that break-even outcomes were initially evaluated as goal obstructive. Subsequent effects of the goal conduciveness check over the corrugator region started at 600 ms, this response patterning remained uniform until the end of the recorded time intervals (until 1,400 ms after feedback onset): Corrugator activity was similarly elevated following both losses and break-even outcomes than following wins. The response patterning between 600 and 1,400 ms suggests that losses and break-even outcomes were similarly evaluated as goal obstructive, whereas wins were evaluated as goal conducive. Contradicting the predicted time interval of first significant effects of the goal conduciveness check over the corrugator region, in Experiment 1, first results of the goal conduciveness check occurred ~400 ms earlier than expected. In

CHAPTER 3.3:APPRAISAL PATTERNS IN FACIAL EXPRESSIONS 120 comparison to previous findings, the present effects are in the same time range as effects of the relevance check and the intrinsic pleasantness check.

Activity over the cheek region

The three-factorial ANOVA yielded no significant interaction effects of Goal Conduciveness × Time, or Power × Time (F-values < 1.84, pGG -values >.157). It revealed only a marginal significant interaction effect of Power × Goal Conduciveness, F(2,

46) = 3.18, p_GG = .062, ŋ² = .12, ε = .831, indicating cumulative effects. No other effects were found (F-values < 2.19, pGG -values > .108). The only significant main effect was Time, F(13, 299) = 13.38, p_GG < .001, ŋ² = .37, ε = .130. Strictly speaking, these ANOVA results indicate that muscle activity over the cheek region was not affected by results of the goal conduciveness check and the power check. This contradicts prediction (2) that appraisal check effects unfold in time over the cheek region. However, the marginal significant interaction effect between the power check and the goal conduciveness check supports prediction (1) that the cheek region is somewhat affected by results of these appraisal checks and also prediction (3) that these results show cumulative effects.

In order to examine the validity of prediction (3) concerning cumulative effects, exploratory post hoc tests on the interaction between Power and Goal conduciveness were calculated (Figure 5). Results revealed that when power was low (bound choice about the outcome), activity over the cheek region was marginally differentially increased in response to wins, losses, and break-even outcomes, F(2, 46) = 2.96, pGG = .087, ŋ² = .11, ε = .642.

Specifically, muscle activity was more elevated following losses than following wins, t(23) = 2.12, p = .045. However, when power was high (free choice about the outcome), no

differential effects in response to the gambling outcomes were obtained, F(2, 46) = 1.47, p_GG = .240, ŋ² = .06, ε = .952. These results suggest that when power was low,

CHAPTER 3.3:APPRAISAL PATTERNS IN FACIAL EXPRESSIONS 121

Figure 4. The EMG amplitude variations over the corrugator region (expressed as percentage change scores according to baseline) illustrate an unfolding over time of the goal conduciveness check results. The goal conduciveness check was manipulated in form of wins, losses, and break-even outcomes (Experiment 1). Shown are also the standard errors of the mean. *p < .05, **p < .01.

CHAPTER 3.3:APPRAISAL PATTERNS IN FACIAL EXPRESSIONS 122 goal conduciveness check effects differentially affected muscle activity over the cheek

region, which was not the case when power was high. This response patterning indicates cumulative effect of both the goal conduciveness check and the power check. Moreover, these cumulative effects seem to be multiplicative.

Figure 5. EMG amplitude variation over the cheek region (Experiment 1). Additionally, standard errors of the mean are provided. *p < .05.

3.3.3.5. Discussion

The aim of Experiment 1 was to test the following predictions of the Component Process Model: (1) that Coping appraisal checks have efferent effects on facial expressions, (2) that the efferent effects of these appraisal checks are subsequent to efferent effects of the goal conduciveness check, and (3) that the manipulation of appraisal checks of two appraisal objectives (i.e., Implication and Coping) results in cumulative effects. According to the Component Process Model, appraisal checks results trigger the response patterning in facial muscle regions when the processing of an appraisal check achieves preliminary closure. The result of each consecutive appraisal check that achieves preliminary closure is predicted to differentially and cumulatively affect the state of the emotion components, of facial

CHAPTER 3.3:APPRAISAL PATTERNS IN FACIAL EXPRESSIONS 123 expressions for example. The predicted efferent effects of the goal conduciveness check and the power check were investigated. Thus, facial EMG was placed over the frontalis, the corrugator, and the cheek region to record participants’ responses to feedback stimuli, presented in a gambling task. Feedback stimuli conveyed information about the goal conduciveness and the power checks at the same time.

In the concurrently recorded brain activity, the results on event-related potentials (ERP) demonstrated sequential effects of the goal conduciveness check and the power check (see chapter 3.1, or Gentsch et al., 2013). On the feedback-related negativity (an ERP

component sensitive to reward processing, occurring 230–300 ms after feedback, that indicated an unfavorable outcome, negative deflections were largest over frontal scalp locations) goal conduciveness check effects were solely found, suggesting that break-even outcomes were evaluated as goal obstructive, losses as in-between, and wins as goal conducive. On the subsequent component, the P300 (an ERP component associated with context updating, occurring ~350–600 ms after feedback, the P300 is sensitive to favorable outcomes such as wins and high monetary magnitudes compared with unfavorable feedback with low monetary magnitudes, these positive deflections are largest over parieto-occipital scalp locations) paralleling main effects of the goal conduciveness check, the monetary magnitude of the feedback-stimuli, and the power check were found. The results on the P300 indicated that wins were evaluated as positive compared with losses and break-even

outcomes, high monetary magnitudes were evaluated as more relevant than low monetary magnitudes, and having high power was more positively evaluated than having low power.

The results on the ERPs demonstrated that the manipulated information about the goal conduciveness and the power check in feedback stimuli was differentially processed.

Consequently, similar differential results were expected in facial expressions.

CHAPTER 3.3:APPRAISAL PATTERNS IN FACIAL EXPRESSIONS 124 Regarding prediction (1), efferent effects of the power check were predicted to occur over the frontalis, the corrugator, and the cheek regions; goal conduciveness check effects were expected over the corrugator and the cheek regions. Efferent effects of the power check were solely found over the cheek region, interacting with the goal conduciveness check.

Nonetheless, this interactive pattern supports prediction (3) that each consecutive appraisal check cumulatively affects the state of the facial expression component. When power was low, the efferent effects over the cheek region were more differentiated by efferent effects of the goal conduciveness check compared with when power was high. In contrast to when

Nonetheless, this interactive pattern supports prediction (3) that each consecutive appraisal check cumulatively affects the state of the facial expression component. When power was low, the efferent effects over the cheek region were more differentiated by efferent effects of the goal conduciveness check compared with when power was high. In contrast to when