Purchase this article with an account.
Joshua A. Solomon, Michael J. Morgan; Contextual effects on orientation identification and contrast discrimination in the fovea. Journal of Vision 2005;5(8):180. doi: 10.1167/5.8.180.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
We wanted to know whether contextual effects on orientation identification and contrast discrimination were consistent with contemporary models of orientation-selective lateral interaction. To measure contextual effects, we used phase-locked and phase-randomized Gabor patterns for target and flanks. Wavelength and spread were λ=0.25° and σ=0.18°. When present, eight flanks surrounded the central target. Center-to-center separation between the target and each flank was 1.06°. Contrast discrimination was assessed using an adaptive, 2IFC procedure. Orientation identification was assessed by having observers decide whether the target was clockwise or counter-clockwise of some reference orientation. Implicit and explicit references were used. The former was horizontal; the latter was formed by the junction of black and white semi-annuli (inner radius: 2.4°; outer radius: 2.5°), which appeared in a different orientation prior to each trial. Flanks had little effect on contrast discrimination (or detection). Orientation judgments can be described by the standard normal distribution χ[(t-μ)/σ], where t is target tilt and the parameters reflect two different aspects of performance: tilt bias -μ and tilt sensitivity 1/σ. Tilted flanks produced biases of opposite sign. This is the tilt illusion. Largest biases were obtained when flank tilts were θ±22.5°, where θ is the reference orientation. The tilt illusion is usually attributed to lateral inhibition between neurons with similar orientation preferences. However, despite a sizeable reduction of tilt sensitivity induced by grossly oblique flanks (θ±22.5° and θ±45°), flanks aligned with the reference orientation induced no such reduction. This latter result seems to be incompatible with an iso-orientation preference for divisive inhibition. Model fits were only slightly improved when a) other features of the contrast-response function or b) the strength of excitatory interactions were allowed to vary monotonically with target/flank orientation difference.
This PDF is available to Subscribers Only