Purchase this article with an account.
David R. Andresen, Kalanit O. Grill-Spector; View sensitivity of object representations in human object-selective visual cortex. Journal of Vision 2006;6(6):312. doi: https://doi.org/10.1167/6.6.312.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Humans can recognize familiar objects from almost any viewpoint. However, the neural representations underlying view-invariant recognition are not well understood. We conducted three fMRI experiments with 6 subjects to investigate the sensitivity of object-selective cortex to changes in object viewpoint. In Experiment 1, we examined whether all views of objects are represented equivalently, or whether different views elicit differential responses. For this experiment, blocks of animal and vehicle line drawings were shown in 15° (front), 75°, 135°, and 195° (rear) views. In Experiment 2 we used an immediate fMRI-adaptation paradigm, and in Experiment 3 we used a long-lagged fMRI-adaptation paradigm to characterize the sensitivity of neural populations to parametric changes in viewpoint. Subjects were adapted with either 15° or 195° views of vehicles and animals. Sensitivity to viewpoint was measured as the degree of recovery from adaptation after 0°, 60°, 90°, 120°, or 180° rotations in depth. Experiment 1 revealed that 15° views elicited a higher response than 195° views for animals, but not vehicles. Similarly, Experiments 2 and 3 revealed that fMR-adaptation effects differed with adapting view for animals, but not vehicles. These results suggest that different proportions of neurons are allocated to represent different views of animals. In addition, posterior regions in the lateral-occipital complex (LOC) recovered completely after small rotations, while anterior regions along the fusiform gyrus recovered more gradually with increasing rotation. Taken together, these results indicate that objects are represented by mixtures of view-dependent neural subpopulations within a hierarchy of increasingly view-invariant object representations.
This PDF is available to Subscribers Only