IST Lunch Bunch

How is it that humans so quickly recognize objects and scenes? We use both behavioral methods and machine learning techniques applied to fMRI brain data to address this question. Traditional models of recognition usually proceed from the registration of low-level features, to perceptual grouping and segregation processes, and culminating in recognition and categorization processes that link the visual input with stored learned representations. However, we have evidence that this serial model may not be accurate, with object and category knowledge impacting rather than following early visual processing.  I will review these data as well as their implications for human vision.