Purchase this article with an account.
Jason Haberman, Julie Belkova, George Alvarez; Individual face representation limits the precision of average face representation. Journal of Vision 2013;13(9):423. doi: 10.1167/13.9.423.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
A wealth of evidence suggests that the visual system can compress redundant information into ensemble representations. These representations are derived efficiently and across multiple visual domains (e.g., average orientation, average faces). An unresolved question is whether ensemble representations are computed in a separate visual processing stream from individual object representations. We examine this question using an individual differences approach, assessing the relationship between single item representations and average representations. Observers viewed sets of four faces varying subtly in identity. On individual trials, a box surrounded one of the faces, indicating the single target they should encode. On ensemble trials, a box surrounded the whole set, and observers had to encode the average of all the faces. Using continuous report, observers then adjusted a test face to match what they had just encoded: either a single face from the set or the average face of the set. Individual item precision and ensemble representation precision were assessed using average error (i.e., how far an observer’s response was from the actual correct answer, on average) – the smaller the error the more precise the representation. Because we were interested in individual differences, each participant received identical trials (i.e., same displays in the same order), so that differences across individuals cannot be explained by display differences. Our measures were highly reliable, as indicated by Cronbach’s alpha (individual r=.76; ensemble r=.64). Moreover, the correlation between individual and average face representations was strong (r=.72), approaching the maximum possible given the reliability of our measures. These results suggest that average face representations are computed over individual face representations and are therefore limited in large part by factors influencing individual face perception. While future work must account for general factors that drive these correlations, the current work lays a framework for understanding the mechanisms supporting ensemble processing.
Meeting abstract presented at VSS 2013
This PDF is available to Subscribers Only