Purchase this article with an account.
V.J. Chen, K. Tarczy–Hornoch; Interpolating Light And Shadow In V1 . Invest. Ophthalmol. Vis. Sci. 2006;47(13):5876.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The human visual system can perceive contours and forms that are not fully specified in the physical stimulus. Interpolation across time and space from relatively sparse sample signals is an essential property of such a robust system. A moving pattern of luminance modulation applied to a fixed sparse, random array of dots on a homogeneous background is perceived as clearly demarcated bands of light or shadow defined by the spatiotemporal pattern of dot luminance modulation, despite no actual luminance variation in the background. FMRI studies of illusory contours suggest they are processed by V3A or lateral occipital areas, but we find activation in primary visual cortex associated with spatiotemporal patterns of stimulation that give rise to interpolated percepts.
Baseline stimuli were designed to match test stimuli in magnitude/frequency of dot luminance changes, and retinal motion of the dot array and the pattern of average luminance zones. A separate experiment controlled for any effect of reversal of retinal sweep direction for luminance zones in test vs. baseline conditions in the main experiment. The V1 region of interest (ROI) of each subject was located using a 10%/100% contrast–modulated checkerboard, and exported to other experiments for ROI analysis (SPM MarsBaR add–on).
In V1, Subject 1 showed positive BOLD contrast (t=2.35, p=0.01) in the main experiment, and negative BOLD contrast in the control. Subject 2 had borderline significant activation (t=1.53, p=0.06) in the main experiment, and no significant contrast in the control. With an alternative control design, Subject 3 had activation (t=2.11, p=0.02) in the main experiment, and none in the control.
Increase of BOLD signals associated with stimuli eliciting spatiotemporally interpolated visual percepts can be seen in V1. Further work is needed to determine whether higher visual areas are essential for this, with feed–back to V1, or whether long–range connections in V1 itself can support integration over space and time to recover spatiotemporal patterns based on a series of fragmented views of a visual scene.
This PDF is available to Subscribers Only