4.2.1.1 Introduction
As I mentioned in this Chapter’s introduction, it is possible that the effect of perceptual load on the social judgment perceptions in the prior Chapter were due either to the delayed response or attenuated short term memory imprint for the facial stimuli. To address these issues, in the present study, we set out to examine the effects of perceptual load on the detection of trustworthiness
subjective judgment faces with a slightly modified paradigm (where participants performed the facial social judgment task before the perceptual load search task).
As Experiment 4 was a follow-up to that of Experiment 1, we hypothesised that the reversal in participants' response order would not prohibit high perceptual load from impacting the judgment accuracy of trustworthiness faces and therefore predicted that facial judgment accuracies would be reduced under high perceptual load.
4.2.1.2 Method
4.2.1.2.1 Participants
13 new participants were included in Experiment 4, Mean Age = 21.69, SD 2.72 (range 18 - 26) (6 males) (3 outliers removed). All of the participants in this experiment had normal or corrected-to- normal vision, as in the prior experiment, any subject with less than 50 % accuracy on the Low Load face or search tasks was classified as an outlier and removed from the subsequent analyses, participants participated for course credit or were paid £3 pounds.
4.2.1.2.2 Stimuli and procedure
The apparatus and procedure were the same as experiments 1-3 except that the participants were instructed to make their response on the faces identities before that of the search stimuli (the perceptual load manipulation). For brevity, no face rating task was performed at the end in this series of corollary
experiments (and nor were reaction times recorded as they were contaminated by order effects as the load task was now performed after the face judgment.
As before, the faces were generated using FaceGen Modeller 3.2 (Singular Inversions, 2007), according to the methods described in (Oosterhof et al., 2008). Mirroring Experiment 1, in Experiment 4 participants were instructed in this instance to make their response on 24 faces identities that vary on 4 levels of trustworthiness. Also as before, in advance of starting the experiment, participants
completed one practice block of 12 example trials to familiarise themselves with the keys.
Figure 4-1 Example screen display of Perceptual Load – “Role Reversal” manipulation (high load condition illustrated)
4.2.1.2.3 Design
The Design was very similar to Experiment 1. Mean percentage search accuracy rates were the dependent variable and were entered into a 2 x 4 repeated measures ANOVA, with perceptual load (Low, High) and Valence (-3, -2, +2, +3 - SD faces) as within subject independent variables (IVs).
Subsequently the mean percentage trustworthiness accuracy rates were treated as the dependent variable and entered into the same 2 x 4 repeated measures ANOVA, again with perceptual load (Low, High) and Valence (-3, -2, +2, +3 - SD faces) as within IV’s, excluding trials in which the search response was incorrect.
This same design was employed for Experiment 5 (using dominance faces) and Experiment 6 (using threatening faces).
4.2.1.3 Results and Discussion
Search task Mean percentage search accuracy rates were significantly better in the low
perceptual load condition (M = 96 %) than in the high perceptual load condition (M =92 %), F(1, 12) = 8.08, MSE =.005, p = .013, η2 = .41. There was no significant effect of valence (F<1) F(3, 36) =0.19,
MSE = .003, p = 0.96, η2 =.002. Additionally there was also no significant interaction between valence and load (F<1) F(3, 36) =.25, MSE =.004, p = .86, η2 =.02, this confirms again that the perceptual load manipulation was effective (see Table 4-1).
Perceptual load
Low High Differential Effect of load
Search Accuracy (%) 96(4) 92(5) 4*
Face Accuracy (%) 83(6) 80(6) 3*
SDs are listed in parenthesis. * = significant NS = not significant
Table 4-1 Mean percentage search accuracy rates and mean face accuracy rates in the reversed trustworthiness task as a function of load in Experiment 4
Face classification. Mean percentage trustworthy face accuracy rates were weakly but significantly higher in the low perceptual load condition (M = 83 %) compared to the high perceptual load condition (M =80 %), F(1, 12) = 3.58, MSE = .09, p = .041(one-tailed), η2 = .23, indicating a moderate effect of perceptual load. There was a significant effect of valence F(3, 36) =17.90, MSE = .014, p < .001, η2 = .60, specifically, accuracy decreased from the high untrustworthy faces to the high trustworthy faces. As Figure 4-2 illustrates (and confirmed by post –hoc testing), the overall effect of valence was driven by -3SD valence being significantly greater than all the other valences, - 2SD valence being significantly greater than +2SD and less then -3SD valence, and +2SD being less than +3SD valence.
Figure 4-2 Trustworthiness mean percentage face accuracy rates in the task reversed assignment (valence ranges from high untrustworthy valence -3SD on the left to high trustworthy +3SD on the right)
Finally, there was no significant interaction between valence and load F(3, 36) = 1.07, MSE =.01, p = .374, η2 = .082.
This result seems tentatively encouraging in that, we have a weak main effect of load (one- tailed given our predications about the direction of perceptual load), with a similar pattern to that of Experiment 1 (see Figure 3-4) with a dip in the mid-range valence faces (Figure 4-2 above). However in contrast to Experiment 1 there are higher overall accuracies for the untrustworthy valenced faces as opposed to untrustworthy ones here. This suggests that the button order may slightly impact the
participants' evaluations. Additionally, all the low load accuracies are numerically greater than the high load accuracies except for the mid-range valence -2SD category, where high load is very slightly (although non-significantly) greater than low-load.
Overall, however, in both instances, Experiments 1 and 4, it seems the effect of load is being driven by trustworthy faces, which follow-up testing revealed to be borne by +2SD valence faces being lower under high load (t(12) = 2.38, p = .035) here, rather than +3SD valence faces as in Experiment 1. To ascertain if the patterns that we observed in the last Chapter are reflected in the other facial judgment domains, we return to facial dominance evaluations in the next experiment.