Purchase this article with an account.
Zoey Isherwood, Colin Clifford, Mark Schira, Branka Spehar; BOLD tuning of human visual cortex to natural statistical properties in space and time. Journal of Vision 2018;18(10):1245. doi: 10.1167/18.10.1245.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Natural scenes vary considerably – forests and canyons for instance do not share obvious perceptual similarities. However, in the statistical properties of visual input that the human brain receives across natural scenes, there exists a common distribution of amplitude variations across both spatial and temporal scales. This is known as the 1/fα amplitude spectrum, where across natural scenes α ≈ 0.6 – 1.6. Perceptual tuning to this spectral distribution has been well documented in the spatial domain using psychophysics and fMRI (Isherwood et al., 2017). It is yet to be examined, however, whether the preferential tuning of the visual system to the statistical properties of natural scenes extends to the temporal domain (i.e., the specific kind of fluctuations in visual input that tend to occur over time in natural scenes). Here, we used psychophysics and fMRI across the same subjects (N = 10) to measure perceptual sensitivity (4AFC "odd one out" task) and BOLD responses in visual areas V1-V4 to synthetic noise movies that varied in their spatial (α = 0.25, 1.25, 2.25) and temporal (α = 0.25, 0.75, 1.25, 1.75, 2.25) amplitude spectra. Just noticeable difference thresholds were smallest for natural distributions (α = 1.25) in both spatial and temporal domains. Thresholds were also lowest for stimuli with similar 1/f spectra in space and time, i.e. thresholds were lower for low spatial α values when temporal α values were also low. BOLD responses across visual areas and eccentricities revealed a similar pattern of results, where responses peaked for natural 1/f spatiotemporal distributions. This suggests, measured through both behaviour and physiology, that the visual system has evolved to the typical spatial and temporal properties observed in natural environments, evident by its preferential tuning toward natural scene statistics.
Meeting abstract presented at VSS 2018
This PDF is available to Subscribers Only