Surface characterizations of color thresholds

Abstract

We evaluate how well three different parametric shapes, ellipsoids, rectangles, and parallelograms, serve as models of three-dimensional detection contours. We describe how the procedures for deriving the best-fitting shapes constrain inferences about the theoretical visual detection mechanisms. The ellipsoidal shape, commonly assumed by vector-length theories, is related to a class of visual mechanisms that are unique only up to orthogonal transformations. The rectangle shape is related to a unique set of visual mechanisms, but since the rectangle is not invariant with respect to linear transformations the estimated visual mechanisms are dependent on the stimulus coordinate frame. The parallelogram is related to a unique set of visual mechanisms and can be derived by methods that are independent of the stimulus coordinate frame. We evaluate how well these shapes approximate detection contours, using 2-deg test fields with a long (1-sec) Gaussian time course. Two statistical tests suggest that the parallelogram model is too strong. First, we find that the ellipsoid and rectangle shapes fit the data with the same precision as the variance in repeated threshold measurements. The parallelogram model, which has more free parameters, fits the data with more precision than the variance in repeated threshold measurements. Second, although the parallelogram model provides a slightly better fit of our data than the other two shapes, it does not serve as a better guide than the ellipsoidal model for interpolating from the measurements to thresholds in novel color directions.

Colour adaptation modifies the long‐wave versus middle‐wave cone weights and temporal phases in human luminance (but not red‐green) mechanism.

“…The ellipse was fitted using the parameters a/b = 2-2 (specifying the relative L' and M' weights) and q = 80 deg (specifying the intrinsic M' signal lag). An elliptical detection contour alone cannot provide information about the chromatic properties of visual mechanisms (Poirson, Wandell, Varner & Brainard, 1990). However, the good simultaneous fit of the ellipse and straight contours with the same two parameters demonstrates that the motion is detected by a single luminance mechanism which linearly sums L' and M. LUM thus detects all the moving gratings that lie on the ellipse (Fig.…”

Textures as Probes of Visual Processing

“…2A). In this case, it is impossible to determine the mechanism sensitivity functions(Poirson et al 1990). Thus, although unbiased-salience experiments can determine a lower bound on the number of basic mechanisms, usually they cannot tell us what those mechanisms sense.…”

Equidiscrimination Contours around an Achromatic Colour Are Convex but Not Balanced

“…As it was stressed by Poirson and collaborators (Poirson et al, 1990) when the EDS is ellipsoidal there are many triads of linear colour mechanisms which can account for the same EDS equally well. That is, when in Eq.…”

“…(1) fits the data best. While some researchers have claimed the EDS to have an ellipsoidal shape that corresponds to an exponent of q=2 (Knoblauch & Maloney, 1996;Noorlander & Koenderink, 1983;Poirson & Wandell, 1996;Poirson, Wandell, Varner, & Brainard, 1990), others have reported that their data were better fitted with a parallelepiped with rounded corners as predicted by model (1) when the exponent q > 2 (Cole, Hine, & McIlhaga, 1993, 1994Cole, Stromeyer, & Kronauer, 1990;Sankeralli & Mullen, 1996). So, the issue of what exponent in Eq.…”