Interactions of chromaticity and luminance in edge identification
depend on chromaticity
VIVIANNE C. SMITH and JOEL POKORNY
Visual Science Laboratories, University of Chicago, Chicago
(Received September 7, 2003; Accepted February 17, 2004)
Abstract
The goal of this work was to study interactions of chromaticity and luminance in edge identification. Two horizontal
spatial sawtooth patterns, one with positive and the other with negative harmonics, were compared in a
two-alternative forced-choice (2-AFC) procedure. The observer identified which pattern had sharp upper or lower
edges. The fundamental frequency was 2 cycles 0deg (cpd), with 5 cycles presented in a 2.5-deg square field. The
pattern was presented as a 1-s raised temporal cosine, replacing part of an 8-deg background. Stimuli were specified
in a cone troland (l, s, Y) chromaticity space, with correction for individual equiluminance at a nominal 115 td, and
individual tritan direction. A preliminary set of interleaved staircases established edge identification for the six
directions of the ~ l, s, Y) space. Three compound stimuli combining two orthogonal directions were chosen and
included with the end-points in five randomly interleaved staircases. For combinations of Y with l-chromaticity,
or l- with s-chromaticity, probability summation was observed. Combinations of Y with s-chromaticity revealed
opponency. Data for +s, +Y and -s, -Y were subadditive; data for +s, -Y and -s, +Y were additive. Control
studies using detection rather than edge identification revealed probability summation for all combinations.
Luminance edges did not enhance stimuli with l-chromaticities. There was an interaction of luminance edges with
s-chromaticities. Dim “blues” and bright “yellows” showed linear summation. Bright “blues” and dim “yellows”
showed opponency.
Keywords: Edge identification, Chromaticity, Luminance
Introduction
Early studies of color vision used optical systems with simple
spatial displays, usually circular fields in a surround. The major
stimulus variables were narrowband spectral stimuli (Wright, 1946),
reviewed in Pokorny and Smith (1986). A few studies found ways
to study chromaticity-luminance interactions using optical systems
(Cole et al., 1993; Chaparro et al., 1994). The advent of monitor
systems allowed not only the use of complex spatial stimuli but
also stimuli that varied only in chromaticity at equiluminance,
varied only in luminance at a chromaticity metameric to equal
energy white, or provided a combination of the two. While many
studies concentrated on discrimination of equiluminant stimuli
(Mullen, 1985; Krauskopf & Gegenfurtner, 1992; Zaidi et al.,
1992; Mullen & Losada, 1994), interest turned also to chromaticity-
luminance interaction (Mullen et al., 1997; Mullen & Sankeralli,
1999) and chromaticity interactions between color pathways (Mullen
& Sankeralli, 1999). The term “interaction” refers to the threshold
sensitivity to stimuli that contain two components, for example,
luminance plus chromaticity of one color pathway or two chroma-
ticities from two color pathways.
The results of studies of chromaticity-luminance interactions
are equivocal, some found stochastic independence in detection of
stimuli exciting long (L-) and middle (M-) wavelength-sensitive
cones (Mullen et al., 1997; Mullen & Sankeralli, 1999) while
others (Gur & Akri, 1992; Gur & Syrkin, 1993; Syrkin & Gur,
1997) reported linear summation. The discrepancy is not resolved,
but Syrkin and Gur (1997) suggested that linear threshold sum-
mation of chromatic and luminant components might be a feature
of suprathreshold tasks. A study of chromaticity-luminance inter-
actions for short (S-) wavelength-sensitive cone stimuli and a
study of equiluminant chromatic axis interaction similarly revealed
stochastic independence (Mullen et al., 1997; Mullen & Sanker-
alli, 1999).
Modern electrophysiology has revealed three primary process-
ing streams from retina to the lateral geniculate nucleus (LGN).
These are the magnocellular (MC-) pathway, summing L- and
M-cone signals, and processing achromatic contrast information,
the parvocellular pathway (PC-), a spectral opponent system dif-
ferencing L- and M-cone signals, and a portion of the koniocellular
(KC-) pathway dominated by S-cone signals. The MC-pathway is
often considered to be the substrate of a psychophysical (L + M)
“luminance” pathway (Lee et al., 1988); the PC-pathway is con-
Address correspondence and reprint requests to: Vivianne C. Smith,
Visual Science Laboratories, University of Chicago, Chicago, IL 60637,
USA. E-mail: vcsmith@uchicago.edu
Visual Neuroscience (2004), 21, 377–382. Printed in the USA.
Copyright © 2004 Cambridge University Press 0952-5238004 $16.00
DOI: 10.10170S0952523804213220
377