Visual Neuroscience ~2008!, 25, 355–360. Printed in the USA. Copyright © 2008 Cambridge University Press 0952-5238008 $25.00 doi:10.10170S0952523808080085 Misbinding of color to form in afterimages STEVEN K. SHEVELL,1,2 REBECCA ST.CLAIR,1,3 and SANG WOOK HONG1,3 1Department of Psychology, University of Chicago, Chicago, Illinois 2Department of Ophthalmology & Visual Science, University of Chicago, Chicago, Illinois 3Department of Psychology, Vanderbilt University, Nashville, Tennessee (Received October 1, 2007; Accepted December 3, 2007! Abstract Under dichoptic viewing conditions, rivalrous gratings that differ in both color and form can give the percept of the color from one eye in part of the form in the other eye. This study examined the afterimage following such misbinding of color to form. The first experiment established that afterimages of the misbound percept were seen. Two possible mechanisms for the misbound afterimage are ~1! persisting retinal representations that are rivalrous and subsequently resolved to give misbinding, as during rivalrous viewing, and ~2! a persisting response from a central neural representation of the misbound percept with the form from one eye and color from the other eye. The results support afterimage formation from a central representation of the misbound percept, not from resolution of rivalrous monocular representations. Keywords: Afterimage, Binocular rivalry, Color rivalry, Feature binding Introduction work shows the afterimage is from a neural representation of a surface, not from representations of physical stimuli that cause An afterimage is a percept of a visual stimulus after the physical filling-in. The neural mechanism is very likely to be cortical but is stimulus has been removed. A basic question ~Loomis, 1972; thought to be monocular ~Kamitani & Shimojo, 2004!. Sakitt & Long, 1979; Virsu & Laurinen, 1977; Wilson, 1997! is These previous studies examine afterimages whose appear- whether the experience of an afterimage depends on a retinal ances depend on stimulation of only one eye, not on afterimage representation of the extinguished stimulus or, at least in part, on a appearance that arises from binocular combination. The present cortical representation of it. The experiments presented here inves- study uses dichoptic presentation of two orthogonal chromatically tigated whether persisting signals from a retinal ~or other monoc- different equiluminant gratings; rivalry between these stimuli is ular! neural mechanism can fully account for afterimages from known to cause misbinding of color to form ~Hong & Shevell, percepts that depend on combining signals from the two eyes. If 2006!. For example, if a vertical orange0gray grating is presented afterimages are purely retinal in origin, then percepts that depend to one eye and a horizontal blue0gray grating to the other eye, an on binocular combination should not influence the afterimage. If, orange0blue grating is often perceived. The orientation alternates however, afterimages are influenced by a neural representation that between vertical and horizontal but the color from the suppressed depends on binocular combination, then afterimages, at least in part, form is expressed in non-retinotopic regions of the dominant form are due to a cortical neural representation that incorporates signals ~thus misbinding of color to form!. An unexplored question, in- from both eyes. vestigated here, is the origin of the afterimage following such Behavioral studies show that cortical processes influence after- misbound percepts. images. Interocular suppression of a binocularly rivalrous stimulus The measurements showed that a misbound afterimage was per- alters the strength of its subsequent afterimage ~Tsuchiya & Koch, ceived but only if misbinding was seen during viewing of rivalrous 2005; Gilroy & Blake, 2005!. Afterimages, however, still might stimuli. This finding is consistent with a contribution to afterimages arise from a retinal representation that is regulated in strength by from a cortical mechanism encoding the observer’s percept while a cortical neural response driven by the percepts during rivalrous viewing rivalrous stimuli, even when the percept results from mis- viewing. Another important result concerns the afterimage from binding the color in one eye to the form in the other eye. viewing a perceptually filled-in surface: the afterimage includes the opponent color of the filled region without adaptation of the corresponding portion of the retina ~Shimojo et al., 2001!. This Materials and methods Apparatus Address correspondence and reprint requests to: Steven K. Shevell, Two rivalrous gratings were presented using a Macintosh G4 Visual Science Laboratories, The University of Chicago, 940 E 57th Street, computer driving a Sony color display ~GDM-F520!. The pixel Chicago, IL 60637. E-mail: [email protected] resolution of the cathode ray tube ~CRT! was 1360 ϫ 1024 with a 355 356 Shevell et al. refresh rate of 75 Hz noninterlaced. Ten-bit lookup tables were of Fig. 1a!. The rivalrous stimuli were 2 cpd square-wave gratings used to linearize the red, green, and blue guns of the color CRT. within a circular aperture of diameter 28. Grating luminance was The stimuli on the left ~right! side of the CRT screen were 7.6 cd0m2 throughout the experiments. Three pairs of chromatic- presented to the left ~right! eye with a haploscope. ities were used in the experiments ~Table 1!. Chromaticities are given in cone-stimulus coordinates L0~LϩM! and S0~LϩM! ~MacLeod & Boynton, 1979!. The unit of S0~LϩM! is arbitrary Stimuli and normalized here to 1.0 for equal-energy-spectrum ~EES! Orthogonally oriented equiluminant gratings, one to each eye, “white.” The chromaticities were chosen toward the extremes of were different in chromaticity ~an example is shown in the left half the display’s gamut for L0~LϩM!, and to allow a comparison Fig. 1. Top row: Schematic representation of continuously presented dichoptic stimuli. Middle row: Schematic representation of pulsed dichoptic stimuli. Bottom row: Four percepts observers could report during the experiments. Misbinding of color to form in afterimages 357 Table 1. Three pairs of chromaticities used in the experiments. second condition, the grating to the left eye was presented alone Chromaticities are given in coordinates L/(LϩM) and S/(LϩM) for 500 ms and then the grating to the right eye was presented (see text). The approximate color appearance of each alone for 500 ms. The guidelines and nonius lines were continu- chromaticity is shown in the rightmost column ously presented to each eye. This was continued for a total of 40 s. The stimulation to each eye was the same in both conditions but Color the gratings were not presented simultaneously ~and thus not Pair Chromaticity Appearance rivalrous! in the second condition. Perceptual alternation while viewing the gratings and during 1L0~LϩM! ϭ 0.708 S0~LϩM! ϭ 3.2 L0~LϩM! ϭ 0.621 S0~LϩM! ϭ 0.4 the subsequent period of afterimages was measured in each run. Observers pressed different buttons on a game pad to report their 2L0~LϩM! ϭ 0.621 S0~LϩM! ϭ 3.2 L0~LϩM! ϭ 0.708 S0~LϩM! ϭ 0.4 percept and held the button until the percept changed ~e.g., ob- servers pressed one button when the stimulus from the left eye was 3L0~LϩM! ϭ 0.708 S0~LϩM! ϭ 1.0 L0~LϩM! ϭ 0.621 S0~LϩM! ϭ 1.0 perceived and pressed a different button when the stimulus from the right eye was perceived!. Buttons were assigned for the percept of the left-eye stimulus alone, the right-eye stimulus alone, mis- binding with the orientation of the left-eye stimulus, and misbind- ing with the orientation of the right-eye stimulus ~Fig. 1c!. Observers were told not to press any button if they experienced piecemeal or between a large accompanying difference in S0~LϩM! versus no ambiguous percepts. Afterimage percepts were measured in the S0~LϩM! difference. Following the presentation of rivalrous grat- same manner. In the second experiment, however, in the pulsed ings for 20 s, a uniform achromatic field ~metameric to EES! was simultaneous viewing condition ~only! observers were asked to presented in the circular apertures to evoke an afterimage ~right report only misbound percepts ~to determine if and when misbind- side of Fig. 1a!. The luminance of the achromatic field was also 7.6 ing occurred! because briefly pulsed simultaneous ~and rivalrous! 2 cd0m . In the first experiment, the gratings were presented con- stimuli caused a very rapid change in percept. Pilot work showed tinuously and the background was dark. In the second experiment, the percept changed too rapidly to be accurately tracked with the gratings were pulsed on and off; the luminance of the back- different button-press responses. During the afterimage phase, they 2 ground was fixed at 5.2 cd0m ~Fig. 1b!. The non-zero background were told to report both monocular and misbound percepts, as in luminance was found to enhance the visibility of the afterimages other conditions. Observers repeated every condition five times. following pulsed stimuli. The total duration of each percept, averaged over the replications White circular guidelines surrounded each grating to aid stable ~all on different days!, was used for further analysis. fixation and maintain correct alignment of the two eyes. Top and left nonius lines were presented within the guideline to the left eye, and bottom and right nonius lines were presented within the Observers guideline to the right eye. With each eye properly fixated, the Three observers participated in the study. Each observer was tested observer perceived two horizontally aligned and two vertically for normal color vision using a Neitz anomaloscope and for normal aligned nonius lines. stereoscopic vision using the Titmus Stereo Test. Observer R.S. is one of the authors and had prior experience with binocular rivalry experiments. Observer P.K. had experience with binocular rivalry Procedure but was not knowledgeable about this study. Observer A.B. was Within each session, each chromaticity pair was presented in two naïve.