Asymmetries in oriented-line detection indicate two orthogonal filters in early vision

Abstract

Visual detection of a line target differing in orientation from a background of lines may be achieved speedily and effortlessly. Such performance is assumed to occur early in vision and to involve filter mechanisms acting in parallel over the visual field. This study establishes orientational limits on this performance and analytically derives some generic properties of the underlying filters. It was found that, in brief displays, target orientation detection thresholds increased approximately linearly with background orientation, from minima at 0 degree (vertical) and 90 degrees, whereas background orientation detection thresholds decreased approximately linearly with target orientation, from maxima at 0 degree and 90 degrees. Target and background threshold functions were exactly antisymmetric. These data are shown to indicate a model of early line processing dominated by two classes of orientation-sensitive filter with axes close to the vertical and horizontal and orientation-tuning half-widths each of approximately 30 degrees at half-height.

Response profiles to texture border patterns in area V1

“…Neurons in V1 generally respond less strongly to texels surrounded by similar texels than to the same texels when surrounded by texels at the orthogonal orientation (Knierim & Van Essen, 1992;Kastner et al, 1997Kastner et al, , 1999Nothdurft et al, 1999). Because texels in orthogonal surrounds are highly salient and "pop out" (Treisman, 1985;Treisman & Gormican, 1988;Foster & Ward, 1991;Nothdurft, 1991Nothdurft, , 1992, it has been proposed (Knierim & Van Essen, 1992;Nothdurft, 1991Nothdurft, , 1994aKastner et al, 1997) that perceptual popout could be based on the surround modulation observed in area V1. Similar surround enhancement effects might account for the detection of texture boundaries (Nothdurft, 1994b).…”

Multiple groups of orientation–selective visual mechanisms underlying rapid orientated–line detection

“…The mask consisted of 20 patches of four randomly orientated line elements, with the arrangement of the elements di¡ering from patch to patch and each patch covering one of the previously displayed line elements. The stimulus timecourse was chosen on the basis of previous experiments (Foster & Ward 1991;Doherty & Foster 1996b). …”

“…The choice of criterion level of d H was constrained by two requirements: that the Á should be stable (i.e. have small standard deviations) and that the shape of the incrementthreshold functions should be invariant under reasonable changes in d H (Foster & Ward 1991). For values of d H greater than about 0.8, one or other of these requirements was not satis¢ed.…”

“…There is a rapid oscillation of Á with , which varies markedly from observer to observer, along with a more general underlying slow variation with a period of about 90 (cf. Foster & Ward 1991).…”

“…As is shown here, there are di¤culties in accounting for this asymmetry in terms of the classical oblique e¡ect, in which orientation acuity and similar measures are poorer along the oblique axes than along the vertical and horizontal axes. It can, however, be explained quantitatively with reference to two classes of broadband anisotropic mechanisms orientated along the vertical and horizontal, with half-widths at half-height of about 30 (Foster & Ward 1991). It has been suggested that the orientation-tuning functions of these mechanisms are closely matched to certain statistical components of realworld images (Baddeley & Hancock 1991).…”