Purchase this article with an account.
Tim Preston, Zoe Kourtzi, Andrew Welchman; Context shapes estimation of 3D structure in human visual cortex. Journal of Vision 2008;8(6):845. doi: 10.1167/8.6.845.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
In estimating the three-dimensional (3D) properties of the world the brain combines ambiguous sensory signals with prior information about the structure of the environment. The abundance of depth aftereffects suggests that the brain constantly adapts to the changing statistics of the visual input. Here, we show that observers' perceptual judgements of disparity-defined shapes are influenced by the adaptive 3D context in which these shapes are viewed, and that fMRI responses in extrastriate visual cortex are modulated by this context. Observers viewed Mach card stimuli that were disambiguated with differing amounts of binocular disparity and reported their perception of 3D shape (convex or concave). Manipulating the context in which these shapes were viewed by presenting a concave or convex Mach card (adaptor stimulus) prior to each test stimulus affected the observers' reports of perceived 3D shape. In a concave context, perception of the test stimuli was biased towards a convex interpretation whereas in a convex context perception was biased towards a concave interpretation. We used an event-related fMRI adaptation paradigm to measure fMRI responses relating to this aftereffect in human visual cortex. Changing the 3D context resulted in fMRI selective adaptation in higher dorsal (V3A, V3B/KO, V7 and hMT+/V5) and ventral areas (VP/V3, V4 and LO) but not early visual areas. That is, we observed higher fMRI responses for test stimuli with different rather than the same 3D shape as the adaptor stimulus. In accordance with the behavioural aftereffect we observed similar fMRI adaptation effects in these areas when the test stimulus contained zero disparity. These findings suggest that fMRI activity in higher visual areas relates to perceived depth and is shaped by contextual information about the structure of the environment.
This PDF is available to Subscribers Only