Effects of Background Color on the Global and Local Processing of Hierarchically Organized Stimuli Chikashi Michimata Sophia University
Matia Okubo and Yosuke Mugishima University of Tokyo
Abstract ■ Recent studies have shown that (1) the global precedence effects in processing the hierarchically organized stimulus can be attenuated by eliminating the low spatial frequencies contained in the stimulus and (2) the human magnocellular pathway is responsible for processing low spatial frequencies and the pathway can be attenuated by imposing a red background on the stimulus. In the present study, a reaction-time experiment was conducted to examine the effect of background
color of the stimulus to the processing of the hierarchically organized stimulus. The result showed that although the control condition (a green background) produced a prototypical asymmetric global interference, a red background that was equiluminant to the green background produced a symmetrical interference. It was concluded that the human magnocellular pathway is at least partially responsible in producing the global precedence effects. ■
INTRODUCTION
frequencies from the hierarchical geometric patterns, by using “contrast balanced dots,” attenuated the asymmetric global interference. They concluded that the GPE is “heavily dependent on the low spatial frequency content of the patterns” (p. 272). On the other hand, Lamb and Yund (1993), again using contrast balanced dots, found that eliminating the low spatial frequencies from the hierarchical letter stimuli attenuated the global reactiontime advantage, but it did not affect the asymmetric global interference. Thus, although there is a discrepancy in speciªcally what aspect of the GPE is inºuenced by the spatial frequency manipulation (probably the discrepancy is due to different stimulus patterns employed), it is relatively well-established that the low spatial frequency plays a crucial role in producing the GPE (see Hughes, Nozawa, & Kitterle, 1996, for an extensive review). The present study further examines the effect of stimulus input factors on the GPE, based on the recent developments in modularity theories of visual perception. Speciªcally, the present experiment is designed on the basis of ªndings about the function and characteristics of the magnocellular pathway in the visual system. It is now widely acknowledged that in both primates and humans the lateral geniculate body has a six-layer structure, which can be divided into two major portions. Four layers contain cells with relatively small receptive ªelds and have sustained response charac-
Since Navon’s (1977) inºuential work, much research effort has been devoted to understanding the processing characteristics of global and local elements of the hierarchically organized visual patterns (see Kimchi, 1992, for a review). A hierarchically organized pattern refers to a large, global pattern composed of small, local patterns, a classic example being a large letter composed of small letters. It has been repeatedly demonstrated that processing of the global level is faster than processing of the local level (a global reaction-time advantage), and the global level interferes with local processing more than the local level interferes with global processing (an asymmetric global interference). These global precedence effects (hereafter referred to as GPE) have been intensively studied in the context of multistructural feature processing and are frequently cited as a representative case that perceiving the whole precedes its parts. Although it seems apparent that the GPE is inºuenced by multiple factors, there is a growing awareness that the physical characteristics of stimulus play an important role in producing the GPE. Speciªcally, it has been suggested that the spatial frequency components contained in the stimulus have direct inºuence on the GPE. For example, Hughes, Fendrich, and Reuter-Lorenz (1990) have demonstrated that eliminating the lower spatial © 1999 Massachusetts Institute of Technology
Journal of Cognitive Neuroscience 11:1, pp. 1–8
teristics. They are called the parvocellular layers. The other two layers contain cells with large receptive ªelds and have transient response characteristics. These two layers are called magnocellular layers. The division is preserved in the primary visual cortex (V1 area), and their functional differences are fundamental in understanding human visual information processing (Livingstone & Hubel, 1988). Particularly important to the present study is the fact that the magnocellular pathway contains cells (so-called Type IV cells ) that have receptive ªelds with a tonic red surround mechanism. That is, animal studies have shown that imposing diffuse red light causes the tonic suppression of these cells’ activity (Livingstone & Hubel, 1984; Marrocco, McClurkin, & Young, 1988). Taking advantage of this characteristic, Breitmeyer and his colleagues conducted experiments using human subjects to examine the effects of background color on psychophysical tasks (Breitmeyer & Breier, 1994; Breitmeyer & Williams, 1990). They found that imposing a red background over the stimulus reduced the performance of tasks that required the processing of low spatial frequency (Breitmeyer & Breier, 1994) and of the tasks that required the processing of high temporal frequency (Breitmeyer & Williams, 1990). These studies are important because they provide empirical evidence that the magnocellular pathway is the physiological basis of the transient, low spatial frequency channel that has long been hypothesized in psychophysics literature (cf. Tolhurst, 1975). Furthermore, they provide an effective experimental tool (i.e., background color manipulation) to directly affect the functioning of the magnocellular pathway. Based on these studies, it is very interesting to consider the possibility that the human magnocellular pathway is responsible, at least partially, for mediating the GPE. This possibility implies that the inºuence of functional characteristics of the magnocellular pathway goes beyond the psychophysical levels and that a presumably higher cognitive effect such as GPE is at last partially related to the function of a physiologically identiªed neuronal mechanism. To investigate this possibility, the present experiment examines the effects of background color of stimulus on the processing of hierarchically organized patterns. We simply hypothesize that imposing a red background over the stimulus will reduce the amount of the GPE. The geometric patterns consisting of global and local elements, which closely approximate what Hughes et al. (1990) employed (see Figure 1), will be presented in red or in green background. These geometric patterns are chosen because they are particularly sensitive to spatial frequency manipulation. Furthermore, special attention will be paid to making the red and green backgrounds have the same subjective luminance to avoid the confounding effect of contrast difference.1
2
Journal of Cognitive Neuroscience
Figure 1. Hierarchically organized stimuli used for the experiment. Top left: Congruent stimulus with Square as Global and Local elements. Top right: Incongruent stimulus with Square as Global element and Diamond as Local elements. Bottom left: Incongruent stimulus with Diamond as Global element and Square as Local elements. Bottom right: Congruent stimulus with Diamond as Global and Local elements.
RESULTS For each subject, the median reaction time (RT) for correct responses and the percentage of errors were computed for each of the 16 experimental conditions and subjected to a 2 × 2 × 2 × 2 analysis of variance (ANOVA) with repeated-measurements, deªned by an orthogonal combination of Task (Global/Local), Background Color (Red/Green), Congruency (Congruent/ Incongruent), and Judgment (Square/Diamond). The results of the error analysis will not be reported because the error level was too low (overall average = 4.0%) to permit meaningful analysis. Correlation between the RT and the error rates was high and positive (r = +0.68, p = 0.004), so there was no indication of the speed-accuracy trade-off. To clarify the presentation of complicated fourway ANOVA results, in the following the analysis will be divided into four parts.
Main Effects and Interactions of Task, Congruency, and Judgment This section provides an overview by mostly discussing main effects and two-way interactions. Consistent with previous studies, the Global task was performed faster than the Local task, producing a signiªcant Task main effect (F(1, 21) = 6.37, MSe = 3123, p = 0.020: Global =
Volume 11, Number 1
421 msec, Local = 435 msec). A Congruency main effect was also signiªcant, F(1, 21) = 97.40, MSe = 282, p < 0.001, indicating a reliable interference between the Global and Local levels (Congruent = 418 msec, Incongruent = 436 msec). Square judgments were quicker than Diamond judgments, producing a Judgment main effect (F(1, 21) = 5.88, MSe = 3408, p = 0.024: Square = 420 msec, Diamond = 435 msec). There was also a Task × Judgment interaction, F(1, 21) = 7.65, MSe = 557, p = 0.012, showing that the difference in the mean RT between the Global and the Local Task was larger for the Square judgments (mean of Local minus Global = +22 msec) than for the Diamond judgments (+8 msec). In fact, when the analysis was restricted to the Diamond judgments, the Task main effect was no longer signiªcant (p > 0.20). Thus, for unknown reasons the global reaction-time advantage was restricted to the Square judgments. Advantage of the Square judgments over the Diamond judgments was also observed in the pattern of interference: There was a signiªcant Congruency × Judgment interaction, F(1, 21) = 5.69, MSe = 446, p = 0.027, indicating that the overall interference was larger for the Square judgments (Incongruent minus Congruent = +23 msec) than for the Diamond judgments (+13 msec). Interactions of Task, Color, and Congruency This section considers interactions involving Task, Background Color, and Congruency, which are the main concerns of the present experiment. The signiªcant main effects of Task and Congruency were qualiªed by a signiªcant Task × Congruency interaction, F(1, 21) = 12.20, MSe = 519, p = 0.002. This interaction indicated that the Local task produced more interference (+27 msec) than the Global task (+9 msec). Thus, interference from Global level to Local level was more pronounced than vice versa, showing a characteristic asymmetric global interference effect. There was another signiªcant interaction involving Congruency variable, a Color × Congruency interaction (F(1, 21) = 13.75, MSe = 300, p = 0.001). This effect indicated that the overall interference effect was larger for a green background (+24 msec) than for a red one (+11 msec). These interactions were further qualiªed by a signiªcant Task × Color × Congruency interaction, F(1, 21) = 18.27, MSe = 233, p < 0.001. This effect indicated that a larger interference for the Local task than for the Global task was mainly attributable to difference in the green condition. To further clarify this effect, the green and the red conditions were separately analyzed. Colorwise Analysis When the analysis was restricted to the green condition, the Task and the Congruency main effects were signiªcant (F(1, 21) = 6.68, MSe = 1935, p = 0.0173 and F(1,
21) = 96.27, MSe = 275, p < 0.001, respectively), the patterns being equivalent to the overall analysis. The Task × Congruency interaction also was signiªcant, F(1, 21) = 27.75, MSe = 378, p < 0.001, indicating that the interference was asymmetrical (Global task = +9 msec; Local task = +40 msec). On the other hand, when the analysis was restricted to the red condition, the Task and Congruency main effects were again signiªcant (F(1, 21) = 4.55, MSe = 1619, p = 0.045 and F(1, 21) = 16.78, MSe = 308, p < 0.001, respectively). However, the Task × Congruency interaction was not signiªcant (F < 1.0), indicating that there was a comparable amount of interference effect both in the Global task (+9 msec) and in the Local task (+13 msec). Thus, separate analysis for each color condition revealed that although both colors showed the global reaction-time advantage and produced a signiªcant interference between the Global and Local levels, the asymmetrical global interference effect was restricted to the green condition. The pattern of this three-way interaction is illustrated in Figure 2. As can been seen in the ªgure, the asymmetric global interference was absent in the red condition. Square versus Diamond In addition to these ªndings, there were some complicated interactions involving the Judgment variable. For example, there was a signiªcant Task × Congruency × Judgment interaction, F(1, 21) = 11.75, MSe = 371, p =
Figure 2. The signiªcant Task × Color × Congruency interaction. Left and right panels indicate the red and green background conditions, respectively. In the green background condition, the amount of interference (i.e., RT for the Incongruent condition minus RT for the Congruent condition) was larger for the Local task than for the Global task. In the red background condition, the amounts of interference were equivalent for both tasks.
Michimata et al.
3
0.0025. This effect indicated that the asymmetrical global interference was more pronounced in the Square judgments than in the Diamond judgments, as illustrated in Figure 3. As can be seen in the ªgure, for the Square judgments (left panel) there was a larger interference in the Global task than in the Local task (in both colors), whereas for the Diamond judgments (right panel) the interference was more symmetrical when two colors were collapsed. Figure 3 also shows that effects of Judgment further interact with Color. In fact, a Color × Congruency × Judgment interaction was signiªcant (F(1, 21) = 13.94, MSe = 221, p = 0.0012). As can be seen in Figure 3, although for the Square judgments (left panel) there were comparable amounts of interference in the green and the red conditions, for the Diamond judgments (right panel) the interference was almost restricted to the green condition and was virtually absent in the red condition. Taken together, the analyses of the Judgment variable indicated that the overall asymmetrical global interference was more pronounced in the Square judgments than in the Diamond judgments, and the overall effect of the color over the interference was due to the difference in the Diamond judgments. However, it should be stressed that the Judgment variable did not affect the Task × Color × Congruency interaction, the most crucial ªnding of the present result. That is, the four-way interaction was not signiªcant (F < 1.0). Thus, the effects of Judgment do not seem to be very relevant to the major ªnding that the background color affected the asymmetrical global interference. The present results clearly indicate that the red color background had the effect of making the characteristic asymmetric global interference more symmetrical.
DISCUSSION The results of the green background condition closely resembled previous ªndings: The judgments on the global level were faster than the judgments on the local level (i.e., a global reaction-time advantage) and interference of the global level toward the processing of the local level was larger than the interference of the local level toward the processing of the global level (i.e., an asymmetric global interference). Thus the result of the green condition conformed to the classic pattern of the GPE. On the other hand, results of the red background condition were qualitatively different from the results of the green condition. Although there was a global reaction-time advantage, the interference between the global and the local levels was symmetrical: Small and comparable amounts of the global-to-local and the local-toglobal interference were observed. In the green and red conditions the stimulus patterns were identical and the background colors were manipulated to be equiluminant. Thus, the difference between the green and the red conditions cannot be attributed 4
Journal of Cognitive Neuroscience
Figure 3. Effects of the Judgment variable. Vertical axis represents the amount of interference (i.e., RT for the Incongruent condition minus RT for the Congruent condition). Left and right panels indicate the Square and the Diamond conditions, respectively. The ªgure shows that (1) the asymmetrical global interference was more pronounced in the Square judgments than in the Diamond judgments and (2) effects of the background color over the amount of interference were restricted to the Diamond judgments.
to factors such as a difference in the general stimulus visibility. If our experimental procedure has failed in achieving an equiluminance between the two background colors, it is possible that one background condition has become more perceptible than the other because of the different contrast between the stimulus pattern and the background. However, it would be difªcult to account for the present ªnding that there was no main Color effect and the effect of color emerged as a Task × Color × Congruency interaction. That is, effect of color was not uniform but was speciªc to the direction and magnitude of the interference. Such a selective effect of color could hardly be accounted for by the difference in the general perceptivity of the stimulus. Thus, the present result should be accounted for by difference in color per se. As stated in the introduction, the stimulus used in the present study closely approximated the one used by Hughes et al. (1990). It is important that the effects of the red background in the present study are strikingly similar to the effects of eliminating the low spatial frequency in Hughes et al. (1990): Introducing a red background and eliminating the low spatial frequencies both resulted in a symmetrical interference between the global and local levels and neither affected the difference of the average reaction times between the global and local tasks. Such a ªnding is consistent with the argument by Breitmeyer and his colleagues that a red Volume 11, Number 1
background attenuates the function of the human magnocellular pathway that mediates the processing of the low spatial frequencies (Breitmeyer & Breier, 1994; Breitmeyer & Williams, 1990). To explain the fact that introducing a red background and eliminating low spatial frequencies produce a similar effect, it is necessary to consider the characteristics (i.e., attenuated by a red background) and function (i.e., mediating the processing of low spatial frequency) of the magnocellular pathway. In fact, the speciªc effects of the red background in the present results are difªcult to account for without considering the characteristics and the function of the human magnocellular pathway. Thus the present results suggest the possibility that the asymmetric global Interference observed in the processing a hierarchically organized stimuli is at least partially subserved by the human magnocellular pathway. As stated earlier, the present study showed that although the asymmetrical global interference was affected by the red background, the global reaction-time advantage was not. A similar ªnding was observed by Hughes et al. (1990). Such ªndings suggest that the two effects may be subserved by different mechanisms. According to Navon (1977), both effects are explained by assuming that information required to make the response becomes available earlier for the global level than for the local level. Such a view is consistent with the spatial frequency account of the GPE because the low spatial frequency channel is assumed to be faster acting than the high spatial frequency channel (Breitmeyer, 1975; Harwerth & Levi, 1978). According to such a temporal advantage model, there should always be the covariation of the two effects in a variety of experimental situations. The present results apparently do not support this view. An alternative view of this temporal advantage explanation is the inhibitory-interaction hypothesis (cf. Lamb & Yund, 1993), which assumes that the asymmetric global interference occurs because the low spatial frequency channel actively inhibits the function of the high frequency channel (De Valois, 1977; Tolhurst, 1972). Because this hypothesis does not require the temporal precedence of global processing to explain the asymmetric global interference, it does not necessitate a universal covariation of the global reaction-time advantage and the asymmetric global interference. In fact, consistent with the present results, there is growing evidence that the two effects do not necessarily covary (Hughes et al., 1990; Lamb & Robertson, 1989; Paquet, 1992). However, a close examination of the present results does not readily support the inhibitory-interaction hypothesis. Note that the hypothesis predicts that elimination of the low spatial frequency will increase the local-to-global interference by releasing the inhibition of the high-frequency channel. On the other hand, in the present study, introducing a red background selectively decreased the global-to-local interference but did not affect the local-
to-global interference. The present results seem to show that imposing a red background simply attenuated the processing of low spatial frequency, and it does not seem to readily support the notion of the release from inhibition of the high spatial frequency channel. We are not ready to discuss these speciªc issues in more detail because the present experiment was not designed to make a direct inference about spatial frequency. Further study is apparently necessary, in which the patterns of reaction-time advantage and interference are compared between spatial frequency manipulation and color manipulation in a single study. If the present results can be interpreted as showing that the human magnocellular pathway is at least partially responsible for producing the GPE, it will be valuable to reconsider the theoretical accounts of the GPE from the viewpoint of recent modularity theories of visual perception. For example, it is now widely accepted that the separation of the parvocellular and the magnocellular pathways is preserved further in the cortical areas beyond the V1 area. Although the parvocellular pathway is further divided into two major tracts in the subsequent cortical areas, the magnocellular pathway preserves its identity in the higher cortical areas. Furthermore, the parvocellular pathways are, mediated by areas such as V4 area, ultimately connected to the temporal-occipital lobe (the ventral system). On the other hand, the magnocellular pathway goes through areas such as the middle temporal gyrus (MT) area and is ultimately connected to the parieto-occipital lobe (the dorsal system. See DeYoe and Van Essen, 1988, and Schiller, 1994, for reviews). It is also widely accepted that the ventral and dorsal systems mainly mediate the processing of the stimulus features and the stimulus location in a visual space, respectively (Mishkin, Ungerleider, & Macko, 1983). Based on such physiological ªndings, it is interesting to consider the possibility that the visual subsystem that processes the stimulus location in a visual space plays a part in producing the GPE. Traditional views in theorizing the GPE were mainly concerned with the mechanism of stimulus feature processing, and the involvement of a stimulus location processing mechanism has rarely been considered. However, there is a possibility that a global task may not necessarily require subjects to process global “features” because a coarse detection of the general alignment of local elements can often be enough to respond correctly. This is particularly true in the present experiment because it only required two-choice responses, and the Global task did not necessarily require subjects to detect subtle feature differences of the global patterns. This possibility, that both the feature and the location processing mechanisms are involved in processing the hierarchically organized patterns, could have a potential to develop a different theoretical account of the global precedence effects. It is suggested that the future studies should consider the involvement of both the feature and Michimata et al.
5
the location analysis to build a coherent theoretical framework.
CONCLUDING COMMENTS The present results indicate that in processing the hierarchically organized stimulus, imposing a red background over the stimulus pattern shifted a prototypical asymmetric global interference toward a symmetrical interference. Such a ªnding raises the possibility that the asymmetric global interference is at least partially mediated by the human magnocellular pathway. This possibility implies that functional differences of physiologically identiªable visual information processing modules, originating in a level as peripheral as the lateral geniculate body, can directly inºuence speciªc cognitive effects such as the GPE. Furthermore, neuronal connection between the magnocellular pathway and the parieto-occipital lobe suggests the possibility that processing the stimulus location information can play a crucial role in processing the global aspects of the hierarchically organized visual pattern. In future studies, it seems to be particularly important to reexamine the spatial-frequency-related topics in cognitive psychology, ranging from ªgure-ground segregation (Klymenko & Weisstein, 1986 ) to the processing of human faces (Sergent, 1985), by carefully considering the functional differences between the parvocellular and the magnocellular pathways. Such an orientation of research would contribute signiªcantly to the understanding of the neurological basis for classic ªndings in cognitive psychology.
METHOD Subjects Twenty-two student volunteers from Meiji Gakuin University participated in the experiment. All the subjects were naive to experimental psychology and were unaware of the hypothesis under investigation. They all had normal or corrected-to-normal vision in both eyes and reported no color vision deªcit. Apparatus An Apple Power Macintosh 6100/60AV with an Apple Audio Vision 14-in. color display was used for the presentation of the stimuli and for recording subjects’ responses. The experiment was controlled by a program written by a MacProbe software package (cf. Hunt, 1994). A 10-key pad (Elecom Notemini Ex) was connected to the Macintosh and served as a two-key response console. Another 14-in. color display was connected to the Macintosh to monitor the subject’s responses.
6
Journal of Cognitive Neuroscience
Stimulus Stimuli consisted of eight small black squares or diamonds having diameters of 0.5 × 0.5° of visual angle (Local elements). A diamond was created by rotating a square 45°. These elements were spatially arranged to form larger squares or diamonds with diameters of 3.5 × 3.5° of visual angle (Global elements). As presented in the Figure 1, combining the global and local elements resulted in four different patterns. In the two patterns the global and local elements are Congruent (i.e., both levels should elicit the same response), examples being a large square made of small squares and a large diamond made of small diamonds. In another two patterns the two levels are Incongruent (i.e., the two levels should elicit different responses), examples being a large square made of small diamonds and a large diamond made of small squares. These stimulus patterns were placed at the center of either a red or green background, which measured 5.0 × 5.0° of visual angle. The luminance of the small black squares and diamonds were 0.1 cd/m2. The CIE x,y-chromaticity coordinates for the red and green phosphors were (0.63, 0.35) and (0.28, 0.59), respectively, as measured by a chromameter (Minolta CS-100). The luminance of the red patch was ªxed at 4.0 cd/m2. The equiluminance of the two background colors was individually determined by a ºicker photometry method, as explained in the Procedure section.
Procedure Subjects were seated in a dimly lit room approximately 57 cm away from the display with their head positioned by a chin rest. In the ªrst phase of the experiment (the Flicker Session), red and green patches (5 × 5° of visual angle) were alternately presented at the center of the screen for 1500 msec, in a temporal square wave manner at a rate of 15 Hz. After a 10-min dark-adaptation period, subjects were asked to adjust the luminance intensity of the green patch and to ªnd the particular point at which heterochromatic ºicker was eliminated or minimized. Subjects were instructed that when they pushed one of the two response buttons, the luminance of the green patch would be increased in the next trial, and when they pushed another button, the luminance would be decreased. They were also told that the luminance of the red patch was always constant. Subjects could go to the next session after they found the least ºickering point and reported it to the experimenter. There were six sessions consisting of three ascending sequences and three descending sequences. Each session started at a different luminance value so that subjects could not rely on the number of button pushes to guide their judgments. Each subject performed ascending and descending sequences alternatively. Two practice sessions, one
Volume 11, Number 1
for ascending sequences and another for descending ones, were given prior to the task. After a subject completed the ºicker task, the six equiluminance values were averaged and the luminance of the green background of the stimulus in the reaction-time experiment that followed was adjusted to the average value. In the second phase of the experiment (the ReactionTime Session), subjects were instructed to place their index ªnger of each hand on one of the response buttons. On different tasks, subjects were told that the global or local elements of a stimulus were the target level to which they were to make judgments. A tone signal warned the subject to ªxate on the central ªxation dot, which appeared for 750 msec. After the offset of the ªxation dot, a stimulus appeared on the center of the display for 150 msec. The subjects’ task was to press one of the two response buttons, as accurately and as quickly as possible, to indicate that a ªgure at the designated level was a square and to press another button when it was a diamond. The ªnger-response mapping was counterbalanced across subjects. Half of the subjects performed the Global task ªrst, followed by the Local task. The remaining half of the subjects performed the tasks in the opposite order. Before each task, the subjects were shown a sheet of paper that indicated a sample of the experimental stimuli at about the same size as they appeared on the display, to facilitate their understanding of the target level for each task. They then were given 12 practice trials before starting the task. The entire experiment consisted of 384 trials. Twenty-four trials were assigned to each of 16 experimental conditions deªned by an orthogonal combination of Task (Global/Local), Background Color (Red/Green), Congruency (Congruent/ Incongruent) and Judgment (Square/Diamond). Each task was divided into four trial blocks and each block consisted of 48 experimental trials. The subjects were allowed to take a short break after each block. Acknowledgments The third author, Mr. Yosuke Mugishima, died February 12, 1998, and this paper is dedicated to his memory. The experiment reported in this article was conducted while the ªrst author was on the faculty of Meiji Gakuin University, and we thank to the university for providing a research grant. We also thank to our undergraduate assistants, Fumihiko Igarashi, Syuichi Nakayama, and Mizuko Suzuki, for their help in conducting the experiment. We are also grateful to Ken McCallum for English proofreading and for two anonymous reviewers for their helpful comments in preparing the manuscript. Parts of the experiment were presented at the Sixtieth Convention of the Japanese Psychological Association, September 10–12, 1996, Rikkyo University, Japan. Reprint Requests should to be sent to Chikashi Michimata, Psychology Department, Sophia University, 7–1 Kioi-cho, Chiyoda-ku, Tokyo, 102, Japan, or via e-mail: c-michim@hoffman. cc.sophia.ac.jp.
Note 1. Note that the purpose of equating the subjective luminance of the red and green colors is to equate the stimulus contrast, not to create the equiluminant color stimuli. The equiluminant color stimuli once were frequently used to examine the functional characteristics of the magnocellular pathway because the pathway was believed to be wavelength-insensitive (i.e., color blind). However, it has become widely known that the pathway also contains color-sensitive cells. For this reason the present experiment does not use equiluminant stimuli.
REFERENCES Breitmeyer, B. G. (1975). Simple reaction time as a measure of the temporal response properties of transient and sustained channels. Vision Research, 15, 1411–1412. Breitmeyer, B. G., & Breier, J. I. (1994). Effects of background color on reaction time to stimuli varying in size and contrast: Inferences about human M channels. Vision Research, 34, 1039–1045. Breitmeyer, B. G., & Williams, M. C. (1990). Effects of equiluminant-background color on metacontrast and stroboscopic motion: Interactions between sustained (P) and transient (M) channels. Vision Research, 30, 1069–1075. De Valois, K. K. (1977). Spatial frequency adaptation can enhance contrast sensitivity. Vision Research, 17, 1057–1065. DeYoe, E. A., & Van Essen, D. C. (1988). Concurrent processing streams in monkey visual cortex. Trends in Neuroscience, 11, 219–226. Harwerth, R. S., & Levi, D. M. (1978). Reaction time as a measure of suprathreshould grating detection. Vision Research, 18, 1579–1586. Hughes, H. C., Fendrich, R., & Reuter-Lorenz, P. A. (1990). Global versus local processing in the absence of low spatial frequencies. Journal of Cognitive Neuroscience, 2, 272–282. Hughes, H. C., Nozawa, G., & Kitterle, F. (1996). Global precedence, spatial frequency channels, and the statistics of natural images. Journal of Cognitive Neuroscience, 8, 197– 230. Hunt, S. M. J. (1994). MacProbe: A Macintosh-based experimenter’s workstation for the cognitive sciences. Behavior Research Methods, Instruments and Computers, 26, 345– 351. Kimchi, R. (1992). Primacy of wholistic processing and global/local paradigm: A critical review. Psychological Bulletin, 112, 24–38. Klymenko, V., & Weisstein, N. (1986). Spatial frequency differences can determine ªgure-ground organization. Journal of Experimental Psychology: Human Perception and Performance, 12, 324–330. Lamb, M. R., & Robertson, R. (1989). Do response time advantage and interference effect reºect order of processing of global and local information? Perception and Psychophysics, 46, 254–258. Lamb, M. R., & Yund, E. W. (1993). The role of spatial frequency in the processing of hierarchically organized stimuli. Perception and Psychophysics, 54, 773–784. Livingstone, M. S., & Hubel, D. H. (1984). Anatomy and physiology of a color system in the primate visual cortex. Journal of Neuroscience, 4, 309–356. Livingstone, M. S., & Hubel, D. H. (1988). Segregation of form, color, movement, and depth: Anatomy, physiology, and perception. Science, 240, 740–749. Marrocco, R. T., McClurkin, J. W., & Young, R. A. (1988). Spa-
Michimata et al.
7
tial summation and conduction latency classiªcation of cells in the lateral geniculate nucleus of macaques. Journal of Neuroscience, 2, 1275–1291. Mishkin, M., Ungerleider, L. G., & Macko, K. A. (1983). Object vision and spatial vision: Two cortical pathways. Trends in Neuroscience, 6, 414–417. Navon, D. (1977). Forest before trees: The precedence of global features in visual perception. Cognitive Psychology, 9, 353–383. Paquet, L. (1992). Global and local processing in nonattended objects: A failure to induce local processing dominance. Journal of Experimental Psychology: Human Perception and Performance, 18, 512–529.
8
Journal of Cognitive Neuroscience
Schiller, P. H. (1994). Area V4 of the primate visual cortex. Current Directions in Psychological Science, 3, 89–92. Sergent, J. (1985). Inºuence of task and input factors on hemispheric involvement in face processing. Journal of Experimental Psychology: Human Perception and Performance, 11, 846–861. Tolhurst, D. J. (1972). Adaptation to square-wave gratings: Inhibition between spatial frequency channels in the human visual system. Journal of Physiology, 226, 231–249. Tolhurst, D. J. (1975). Sustained and transient channels in human vision. Vision Research, 15, 1151–1155.
Volume 11, Number 1
