Characterizing the representational similarity space in OTC
Processes in the occipitotemporal cortex (OTC) support the recognition of objects, but the nature of object representation isn this cortex and how it relates to object similarity judgments is an active area of study. In this project, we're measuring neural responses to real world objects and relating them to different behavioral tasks that vary in the way in which they track implicit versus explicit judgments.
Exploring the large-scale neural organization of objects
Object representation in the occipitotemporal cortex seems to follow a segregation based on animacy and, within animacy, by size (Konkle and Oliva, 2012; Konkle and Caramazza, 2013). But, what do we mean by “size” and “animacy” representations in the occipitotemporal cortex? Specifically, what is the nature of the representation? Do neural responses reflect an abstract interpretation of a stimulus as being big or animate, or does it reflect something more visual in nature, such as the typical shapes of objects and animals?
size and motor-relevance
In one project exploring these questions, real-world size is contrasted to another, correlated high-level construct: motor-relevance (i.e. the degree to which an object elicits action movements), to test whether a more abstract, functionally-relevant property might partly drive the organization by size in the occipitotemporal cortex.
Magri C., Konkle T., Caramazza A. (2016). Visual object responses of the ventral stream reflect both size and motor-relevance. Vision Sciences Society, May 13-18, St. Pete Beach, Florida. [pdf]
size and curvature
In another project related to this topic, we're relating real-world size to a correlated, mid-level construct: curvature, which has been widely explored as a tuning property in the visual cortex (Pasupathy and Connor, 1999; Yue et al., 2014; Shirasam et al., 2014). It is possible that curvature alone might drive the organization by size of inanimate objects in the ventral stream; alternatively, real-world size information might be preserved even when correcting for curvature.