Page 39 - Demo

37How attractiveness affects implicit cognition 2symmetrized vs. asymmetrized) and Probe Location (behind symmetri,cal/behind asymmetrical face). Furthermore, in contrast to Experiment 1and 3, this experiment did not include a stimulus validation.ResultsSimple modelTo test our main prediction that facial symmetry would significantly influ,ence RT, we ran a Bayesian mixed model with by-subject mean-centered RTper trial as dependent variable, and the interaction between Condition andProbe Location as independent variables (see Table 2). We found no effectof facial symmetry on reaction time in any of the three conditions (see Fig,ure 4); in each condition the differences in RT between the probe locationswere negligible (asymmetrized vs. original: median difference = -1.01 [3.05],89% CI [5.92, 3.82], pd = .63; symmetrized vs. original: median difference =0.99 [2.91], 89% CI [-3.66, 5.66], pd = .64; symmetrized vs. asymmetrized:median difference = 1.67 [2.97], 89% CI [-3.14, 6.32], pd = .71).Age and SexWe investigated whether adding either Age or Sex to the model did improvethe predictive accuracy relative to the simple model. Both the model in,cluding Sex (∆elpdLOO = 4.4 [1.7]) and the model including Age (∆elpdLOO= 0.5 [2.9]) had a slightly lower predictive accuracy than the simple model.Altogether, this suggests that including age or sex did not improve the pre,dictive accuracy of the model.Experiment 3MethodParticipantsExperiment 3 included 150 new participants (73 females, mean age = 30.98years old, SD = 12.65, ranging from 18 to 70 years old). Participants werevisitors at the Apenheul Primate Park (Apeldoorn, The Netherlands). Allparticipants self-reported normal or corrected-to-normal vision and were het,erosexual. The experimental procedures were in accordance with the Decla,ration of Helsinki and the study was reviewed and approved by the Psychol,ogy Ethics Committee of Leiden University (CEP18-0531/272). Participantswere not compensated for their participation.Figure 4. By-subject mean-centered RTs per Condition and per ProbeLocation. Dots indicate the median reaction times (RT), while error barsrepresent the 89% Credible Interval. As can be seen, symmetry did notsubstantially affect reaction time in any of the three conditions.Experiment design & procedureThe experiment held a randomized within-subjects design, where independent variables comprised congruence (looking direction congruent with dotor not), attractiveness category of the stimulus (attractive, intermediate,unattractive), age and sex. The dependent variable was RT (in ms).StimuliFaces were selected from the Oslo Face Database (Chelnokova et al., 2014).This database includes 200 faces (100 females) with a neutral expressionand with three gaze directions: left, center, and right. All stimuli have beenrated for attractiveness. Based on these ratings, we chose 10 attractive, 10intermediate and 10 unattractive faces of each sex.The ages of the people in the photographs were not recorded, so it wasnot possible to analyse whether age differed between the stimulus categories.However, because the database consists of pictures of students, it is likelythat they are in the same age range.Iliana Samara 17x24.indd 37 08-04-2024 16:35