| The emotional value hypothesis predicts a positive relationship between P300 amplitude and the affective value of a stimulus. The following studies were designed to test that hypothesis in the context of facial perception, in conditions where probability and task relevance of stimuli (the two classical P3-eliciting variables) were controlled. In Experiment 1, the subject task consisted of giving an attractiveness rating to each of forty male and female faces using a Likert scale ranging from one to five. Two equally probable levels of facial proportions (short-jaw and average jaw) were selected as the experimental manipulation of attractiveness, since that factor has been shown to be effective on female faces (Johnston and Franklin, 1993). The EEG was recorded at frontal, central and parietal electrode sites. A Principal Component Analysis identified a clear late positive component that peaked at 500 milliseconds and had the typical P300 scalp distribution. Results showed a statistically significant positive correlation between amplitude of this component and attractiveness ratings only for the female faces at the parietal site. However, since the proportions variable was not effective in eliciting ERP differences, the results could be open to other interpretations.;Experiment 2 was designed to overcome these shortcomings. Two levels of hair, eyes, mouth, chin and jaw proportions were systematically manipulated as independent variables in the making of 32 new male and female faces. The quality of the pictures was improved, a longer stimulus exposure was used and the attractiveness ratings were postponed to a fourth session to rule out possible response selection factors. A similar late positive component to the previous study was identified and the significant correlation between its amplitude and attractiveness ratings for the female faces at the parietal site was replicated. This time, however, the facial feature manipulations were effective in eliciting significant attractiveness and ERP changes. An analysis of the proportions data revealed that neither experimental nor real world stimulus probabilities were plausible explanations for changes in P300-related late positive components. Systematic effects of facial feature configurations on both behavioral and electrophysiological measures are discussed, both when they matched and when they revealed a dissociation between these dependent variables. Finally, some experimental prospects for the emotional value hypothesis are considered, along with plausible suggestions on the neural bases for the P300 component and its relationship with other psychological processes. |
