Purchase this article with an account.
Jiye G. Kim, Irving Biederman; Greater modulation of LO responses to changes in nonaccidental than metric relations between simple shapes.. Journal of Vision 2012;12(9):1066. doi: 10.1167/12.9.1066.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Variation in the relations between objects is made explicit, often uniquely, at the same cortical locus—the lateral occipital cortex (LO)--where object shape is made explicit (Kim & Biederman, 2011; Kim, Biederman & Juan, 2011). Given the co-localization of shape and relations, do variations in the relations between shapes exhibit the greater sensitivity to viewpoint invariant (or nonaccidental) properties (NAPs) than metric properties (MPs) that are evident when discriminating individual shapes? In an adaptive staircase design, we compared the sensitivity for detecting a NAP relational change between two shapes, e.g., from a pyramid centered above a brick to the pyramid slightly off centered over the brick, and an MP change, e.g., where a slightly misaligned pyramid is varied to a pyramid further misaligned over the brick. Detecting the MP changes required more than double the stimulus display time compared to detecting the NAP changes. This held true even though the NAP and MP variations were matched in retinal and V1 physical similarities (Kim, Biederman & Amir, 2011). In two fMRI experiments, we show that the NAP relational changes consistently produce greater BOLD modulation than MP changes in LO, implicating LO as the potential neural locus for where the greater detectability of NAPs over MPs is made explicit. None of the other regions tested (V1-V4, posterior fusiform gyrus, and middle temporal cortex) showed this pattern of results. The similarity scaling of our stimuli with HMAX, a model of ventral pathway cell tuning (Mutch & Lowe, 2008), did not consistently yield greater dissimilarity values for the NAP variations that could have accounted for the greater NAP advantage witnessed in the behavioral performance and LO responses. There is thus a need for further development of this model to reflect the greater sensitivity to NAPs which are critical for view invariant object recognition.
Meeting abstract presented at VSS 2012
This PDF is available to Subscribers Only