Object learning and visual category representations in the human brain

One of the peculiar things about human behavior is the ability to process a vast amount of visual information within a split second. What happens in the brain when humans see an object and a face? To answer these questions, we investigate how the human brain represents visual information and how this visual information processing relates to perceptual learning and memory.

The first study draws attention to learning effects of complex objects related to memory reactivation. Our study implemented two training protocols: A full-practice training paradigm and a short-reactivation training paradigm. Participants were trained to recognize a set of backward-masked objects for multiple days. After the initial training to consolidate the memory, a group was trained with repeated practice, while the other had a few, at-threshold reactivation trials. Our findings support that perceptual learning happens with memory reactivation but suggest that recognition performance is improved much more by large amounts of practice.

The second study is of a more methodological nature, and focuses upon an apparent inconsistency in where the strongest face selectivity is found depending on whether results are analyzed in individual subjects or at the group level. More specifically, we investigate why the amygdala sometimes exhibits a strong selectivity for faces, but only in group analyses. The results show the discrepancies between individual and group analyses can be explained by a smaller inter-subject variability of peak’s coordinates in the amygdala.

In the final experiment of this dissertation, we explored the spatiotemporal dynamics of object perception. This study utilizes multivariate pattern analysis (MVPA) and representational similarity analysis (RSA) to distinguish the EEG activation patterns corresponding to animate, lookalike zoomorphic, and inanimate objects. The initially activated representations to lookalike zoomorphic objects are very similar to the representations activated by animal pictures and very different from the neural responses to regular objects. Neural responses that reflect the true identity of the zoomorphic objects as inanimate objects are weaker and appear later, as do effects of task context.