Page 43 - Demo

41How attractiveness affects implicit cognition 2ResultsValidation of stimuliThe ordinal mixed model showed that subjects rated as the central-facingstimuli classified as attractive as substantially more attractive, and the stim,uli classified as unattractive as less attractive (Figure 6). This effect wassimilar for both women (∆estimateattractive-intermediate = 1.81 [0.34], 89% CI[1.26, 2.38], pd = 1.00; ∆estimateunattractive-intermediate = -2.25 [0.35], 89%CI [-2.83, -1.68], pd = 1.00) and men (∆estimateattractive-intermediate = 2.01[0.34], 89% CI [1.46, 2.54], pd = 1.00; ∆estimateunattractive-intermediate = -2.25[0.35], 89% CI [-2.83, -1.68], pd = 1.00).Simple modelTo test our main prediction that attractiveness would significantly influencegaze cueing, we ran a Bayesian mixed model with by-subject mean-centeredRT per trial as dependent variable, and the interaction between Attractive,ness Category and Gaze Congruency as independent variables (see Table3). We found a robust main effect of Gaze Congruency on RT (see Figure7); suggesting that people responded faster when the probe appeared onthe side that was congruent with the gaze direction of the stimulus (mediandifference = 32.16 [1.33], 89% CI [30.01, 34.32], pd = 1.00).We found no clear effect of Attractiveness Category on RT for congru,ent and incongruent trials. Specifically, on incongruent trials, there wasno substantial difference in RT between attractive and intermediate stim,uli (median difference = -1.68 [2.33], 89% CI [-5.39, 2.09], pd = .76), aswell as for unattractive and intermediate stimuli (median difference= 3.22[2.39], 89% CI [-0.52, 6.92], pd = .91). However, people responded slightlyfaster when the stimulus presented was attractive than unattractive (mediandifference = 4.84 [2.35], 89% CI [1.13, 8.56], pd = .98). Regarding congru,ent trials, we found no substantial difference in RT between attractive andintermediate (median difference = -0.61 [2.26], 89% CI [-4.29, 3.06], pd =.60), unattractive and intermediate (median difference = - 1.25 [2.38], 89%CI [-5.04, 2.45], pd = .70), or attractive and unattractive stimuli (mediandifference = 0.67 [2.36], 89% CI [-3.11, 4.37, pd = .61).Age and sexWe investigated whether adding either Age or Sex to the model improved thepredictive accuracy relative to the simple model. When comparing the modelthat included the 3-way interaction between Age, Attractiveness Categoryand Gaze Congruency to the simple model, we found that the predictiveaccuracy of the simple model was slightly better (∆elpdLOO = 4.6 [1.8]).The results were similar for the model that included the 3-way interactionResultsValidation of stimuliThe ordinal mixed model showed that subjects rated as the central-facingstimuli classified as attractive as substantially more attractive, and the stim,uli classified as unattractive as less attractive (Figure 6). This effect wassimilar for both women (∆estimateattractive-intermediate = 1.81 [0.34], 89% CI[1.26, 2.38], pd = 1.00; ∆estimateunattractive-intermediate = -2.25 [0.35], 89%CI [-2.83, -1.68], pd = 1.00) and men (∆estimateattractive-intermediate = 2.01[0.34], 89% CI [1.46, 2.54], pd = 1.00; ∆estimateunattractive-intermediate = -2.25[0.35], 89% CI [-2.83, -1.68], pd = 1.00).Simple modelTo test our main prediction that attractiveness would significantly influencegaze cueing, we ran a Bayesian mixed model with by-subject mean-centeredRT per trial as dependent variable, and the interaction between Attractive,ness Category and Gaze Congruency as independent variables (see Table3). We found a robust main effect of Gaze Congruency on RT (see Figure7); suggesting that people responded faster when the probe appeared onthe side that was congruent with the gaze direction of the stimulus (mediandifference = 32.16 [1.33], 89% CI [30.01, 34.32], pd = 1.00).We found no clear effect of Attractiveness Category on RT for congru,ent and incongruent trials. Specifically, on incongruent trials, there wasno substantial difference in RT between attractive and intermediate stim,uli (median difference = -1.68 [2.33], 89% CI [-5.39, 2.09], pd = .76), aswell as for unattractive and intermediate stimuli (median difference= 3.22[2.39], 89% CI [-0.52, 6.92], pd = .91). However, people responded slightlyfaster when the stimulus presented was attractive than unattractive (mediandifference = 4.84 [2.35], 89% CI [1.13, 8.56], pd = .98). Regarding congru,ent trials, we found no substantial difference in RT between attractive andintermediate (median difference = -0.61 [2.26], 89% CI [-4.29, 3.06], pd =.60), unattractive and intermediate (median difference = - 1.25 [2.38], 89%CI [-5.04, 2.45], pd = .70), or attractive and unattractive stimuli (mediandifference = 0.67 [2.36], 89% CI [-3.11, 4.37, pd = .61).Age and sexWe investigated whether adding either Age or Sex to the model improved thepredictive accuracy relative to the simple model. When comparing the modelthat included the 3-way interaction between Age, Attractiveness Categoryand Gaze Congruency to the simple model, we found that the predictiveaccuracy of the simple model was slightly better (∆elpdLOO = 4.6 [1.8]).The results were similar for the model that included the 3-way interactionFigure 6. Validation of the stimuli of Experiment 3. Probability of receiving high attractiveness ratings was higher for stimuli categorized as “attractive” (A). This is also depicted in (B), which treats the ratings as acontinuous variable for visualization purposes.Figure 6. Validation of the stimuli of Experiment 3. Probability of receiving high attractiveness ratings was higher for stimuli categorized as “attractive” (A). This is also depicted in (B), which treats the ratings as acontinuous variable for visualization purposes.Iliana Samara 17x24.indd 41 08-04-2024 16:35