Cortical Response Similarities Predict which Audiovisual Clips Individuals Viewed, but Are Unrelated to Clip Preference
Individual [channel × time] segments were assigned to one of the video clips using the separate time series approach (a), the aggregate time series approach (b) or the aggregate correlations approach (c). Within (a), electrode-wise correlations are computed between the withheld segment and the reference segments (i.e. the segments recorded within all other sessions and clips). The reference segment which contributes to the highest correlation was identified, and the clip that corresponds to that segment was assigned to the withheld segment. Within b, the reference segments are averaged electrode-wise across identical clips prior to computing correlations, and within c, the electrode-wise correlations are averaged across sessions that correspond to the same clip. The clips are assigned in the same manner as a. The process is repeated using each segment as the withheld segment, generating an assignment for every segment. These approaches emphasize different aspects of similarities across subjects, where (a) is advantageous in instances where only a subset of subjects demonstrate a similar response on a given clip, (b) is advantageous in instances where subjects demonstrate similar cortical responses within a clip, and when these similarities would be obscured when averaging across time series (as in c). C preserves the similarities in the time series across individuals, and is thus advantageous when the majority of individual’s demonstrate similar responses to the clips, and averaging time series enhances these similarities (i.e. the signal).