SUPPLEMENTARY STUDY 1: CHOICE OF STIMULI FOR PAIN AND GENDER DISCRIMINATION TASKS
This preliminary experiment was conducted in order to select the stimuli for the tasks of the pain and gender discrimination experiment. Indeed, we wanted participants to discriminate pain among other negative emotions and we wanted participants to avoid seeing the same faces for both tasks of the experiment.
METHODS
Participants
Two hundred and three healthy students (90 males and 113 females; mean age = 20 ± 4.31) were recruited from two universities in Lyon, France (Claude Bernard Lyon 1 and Lumière Lyon 2). All participants were volunteers and were not paid for their participation. Medical students were excluded because it has been documented that expertise with pain can modulate pain perception in others 13. All participants were naive to the face stimuli used.
Stimuli
The stimuli used were extracted from the battery detailed in the main pain and gender discrimination experiment (Stimuli).
Study design and procedure
Participants had to rate the intensity of pain expressed by each pictureusing a Visual Analogue Scale (VAS; 10 cm long) from “no pain at all” (level 0) to “maximal pain possible” (level 10). Pictures were presented randomly for 3-second on a projection screen in an auditorium, followed by a 3-seconds gray screen allowing participants to write down their rating on an answer sheet. The duration of the experiment was approximately 10 minutes.
Data analysis
Pictures to which participants did not respond were discarded. For each participant, ratings for pictures depicting the same category of facial expression were averaged to obtain one individual score for each emotion.
Statistical tests
We used the D’Agostino-Pearson omnibus normality test to determine whether data sets were normally distributed. Depending on the results of this first analysis, parametric or non-parametric tests were used. The significance criterion for all analyses was set at p < .05. All analyses were conducted with the GraphPad Prism® 6.0 software.
The effect of participants’ gender on the rating of pain intensity was tested using an independent samples t-test comparing female and male participants’ responses. Pain intensity ratings for the different emotional facial expressions were compared using the Friedman non parametric “analysis of variance” test, with picture emotional valences as repeated measures. Post hoc analyses were performed using Dunn's multiple comparison test.
RESULTS
Effect of participants’ gender on the rating of pain intensity
There was no effect of the participants’ gender on the rating of pain intensity depicted in all facial expressions (t = 0.59; p = .56). Therefore, the ratings from both genders were pooled together.
Rating of pain intensity for the different emotional facial expressions
There was a significant difference between the pain intensity ratings for the eight different categories of facial expressions (Friedman X 2 (7) = 1226; p < .0001).
Post hoc analyses showed thatparticipants attributed the highest pain intensities to the faces expressing pain. As shown in Figure A, this rating was significantly higher (p < .0001) than those attributed to all other emotions and neutral faces. There was no significant difference between the pain intensity ratings for disgust versus fear, sadness versus anger and surprise versus neutrality.
Based on these results, anger faces were selected to serve as the noisein addition to neutral faces, for the Pain Discrimination Task (PDT) because they were rated with the lowest pain intensity scores among the four negative facial expressions (disgust, fear, sadness and anger) and were thus the most distinct from pain faces in terms of subjective pain perception. Faces expressing disgust, fear and sadness were used for the Gender Discrimination task (GDT).
Figure A (1.5-column fitting) Mean pain intensity ratings (from 0 “no pain” to 10 “maximal pain possible”) for each emotional facial expression with standard deviations. Significant differences between the ratings for two emotions are notified with stars (*** for p < .0001).
SUPPLEMENTARY STUDY 2: CONTROL STUDY FOR GENDER DISCRIMINATION
Results from the pain and gender discrimination experiments showed that participants could discriminate pain but not gender from a face presented for 100ms. Even if the battery of stimuli used is suitable for gender discrimination 70,71, the faces are deprived of external features that might have slowed down the gender discrimination. Moreover, because we did not want the participants to process the same pictures for both PDT and GDT, the GDT was proceededwith emotional faces (sadness, disgust and fear faces), which could have interfered with the gender discrimination. This supplementary experiment aimed to test whether gender would have been discriminated in faces presented for 100ms when they expressed no emotion, i.e. neutral faces.
METHODS
Participants
Two hundred and ten healthy students (40 males and 170 females; mean age 18.90 ± 1.69) who did not participate at any of the previous experiments and thus naive to the face stimuli used, were recruited from the University Lumière Lyon 2, France. All participants were volunteers and were not paid for their participation.
Stimuli
The stimuli used were extracted from the battery detailed in the pain and gender discrimination experiment (Stimuli). Thirty male and thirty female pictures were selected from this set of images, expressing neutral, pain, fear, disgust, anger and sadness facial expressions. A mask was created by averaging pixel-to-pixel of all of these pictures using PsychoPy® software (version 1.8) 63(Figure 2 D).
Study design and study procedure
Participants had to answer (yes or no) for each presented face, whether it was that of a man or not; male faces being the signal and female faces the noise. The pictures were randomly presented on a projection screen in an auditorium. First, a 1000ms cross-fixation appeared, then pictures were presented for 100ms preceded and followed by the 200ms mask. After that, a gray screen appeared for 5000ms where was projected the question in French “Man?” with a “Yes” written on the left and a “No” written on a right. This guided participants who had to tick yes or no on an answer sheet (yes on the left, no on the right).
Statistical tests
As detailed earlier in the pain and gender discrimination experiment we applied a Fisher’s exact test (with α =.05) to hits, misses, false alarms and correct rejections numbers obtained at the group level.
RESULTS
Participants could not discriminate gender on all types of emotional faces (neutral, pain, anger: p Fisher = 1.00; disgust, fear:p Fisher = .52; sadness: p Fisher = .57).
These results showed that it is not possible to discriminate gender on emotional faces presented for 100ms between masks, which backed up results from the “Pain and Gender discrimination experiment”. This non-discrimination is not due to emotional component of the faces that might impair gender discrimination because gender was not discriminable for neutral faces neither.
1