Stop laughing! Humor perception with and without expressive suppression

Abstract

The neurophysiological study of emotion regulation focused on the strategy of reappraisal—i.e., the cognitive reinterpretation of a stimulus. Reappraisal reduces emotional expression, the experience of both negative and positive feelings, and the amplitude of an event-related potential (ERP)—the late positive potential (LPP). In contrast, the strategy of expressive suppression (ES), being the inhibition of emotional expression, has been reported to reduce subjective feelings of positive, but not negative emotion, and has not yet been investigated with ERPs. We focused on the LPP to assess the correlates of ES in the context of humor perception. Twenty-two female participants rated sequences of humorous (H) and non-humorous (NH) pictures, while their zygomaticus muscle was recorded. A spontaneous (SP) condition, in which participants attended naturally to the pictures, resulted in higher ratings of funniness, increased smiling, and increased LPP amplitude for H compared to NH stimuli. An ES condition, in which participants suppressed their facial reactions, resulted in reduced smiling, without affecting subjective ratings. LPP amplitude did not differ between H and NH stimuli during ES, suggesting equal allocation of processing resources to both stimuli. These results suggest that, similarly to reappraisal, ES modifies the way the brain processes positive emotional stimuli.

Keywords:

• Emotion regulation
• Expressive suppression
• Humor
• EEG
• EMG

Acknowledgments

All authors are supported by the Swiss Science Foundation. We thank Annekathrin Schacht for helpful comments on data analysis, and Corrado Corradi-Dell'Acqua and Guillaume Chanel for help with the figures.

Notes

¹ Other studies investigating ES in fact instructed their participants to suppress not only their emotional expression but also their subjective feelings (e.g., Levesque et al., 2003; Ohira et al., 2006).

² Mirth, exhilaration, and amusement are often used interchangeably in the humor literature and designate the positive response to humorous stimuli.

³ The second picture of each pair was rated in the pilot study on a 100-point visual analog scale ranging from 1 (not at all humorous) to 100 (very humorous). Out of the initial picture set, 294 pictures were retained after removal of outliers. Scores were standardized (this explains partly negative scores) and averaged over participants. A one-way ANOVA with the factor Stimulus Type (two levels: H and NH; these categories were based on an a priori categorization) resulted in a significant main effect: F(1, 292) = 702.5, p < .001. Humorous trials (M = 0.59, SE = 0.03) were rated as significantly more humorous than neutral trials (M = –0.6, SE = 0.03). Finally, six new picture pairs were added for the main experiment, which included 300 picture pairs in total. Note that the rating study only served to confirm that approximately half of the trials were generally perceived as H and the other half as NH. This allowed achievement of an equal distribution of H and NH trials. Analyses of the ERP and EMG data from the main experiment were instead based on each participant's subjective ratings on a 5-point Likert scale.

⁴ Blinks were identified by their topography with high loadings on bilateral fronto-polar sites, their low-frequency power spectrum, and their transient time-pattern. Artifacts elicited by saccades and lateral eye movements showed an anterior left–right dipole, high power in lower frequencies, and an abrupt (square-like) change in the component time series. Finally, components corresponding to blinks and eye movements were not time-locked to stimulus onset (for examples of corresponding topographies, see Onton, Westerfield, Townsend, & Makeig, 2006; McMenamin et al., 2010).

⁵ After removal of the ES block, the theoretical mean and median trial number corresponded to the number of SP trials after artifact rejection, divided by two. For each participant, the indices of trials at which true smiles occurred were used to calculate the actual mean and median trial numbers, and then compared via dependent-sample, two-tailed t-tests to the theoretical values.