Pattern classification predicts individuals' responses to affective stimuli

Since the successful demonstration of "brain reading" of fMRI BOLD signals using multivoxel pattern classification (MVPA) techniques, the neuroimaging community has made vigorous attempts to exploit the technique in order to identify the signature patterns of brain activities associated with different cognitive processes or mental states. In the current study, we tested whether the valence and arousal dimensions of the affective information could be used to successfully predict individual's active affective states. Using a whole-brain MVPA approach, together with feature elimination procedures, we are able to discriminate between brain activation patterns associated with the processing of positive or negative valence and cross validate the discriminant function with an independent data set. Arousal information, on the other hand, failed to provide such discriminating power. With an independent sample, we test further whether the MVPA identified brain network could be used for inter-individual classification. Although the inter-subject classification success was only marginal, we found correlations with individual differences in affective processing. We discuss the implications of our findings for future attempts to classify patients based on their responses to affective stimuli.