Interhemispheric visual interaction in a patient with posterior callosectomy

Transcription

1 Neuropsychologia 41 (2003) Interhemispheric visual interaction in a patient with posterior callosectomy S.R. Afraz, L. Montaser-Kouhsari, M. Vaziri-Pashkam, F. Moradi School of Intelligent Systems, Institute for Studies in Theoretical Physics and Mathematics, Niavaran, P.O. Box , Tehran, Iran Received 1 February 2001; received in revised form and accepted 19 September 2002 Abstract The role of anterior commissure (AC) and anterior parts of corpus callosum in visual interactions was investigated in a partial split-brain patient whose posterior and middle parts of the corpus callosum were resected surgically leaving intact only a thin portion of anterior corpus callosum. Although the primary visual areas of the two hemispheres are disconnected in the patient, we found that visual distracters presented to one hemisphere (in a crowding paradigm) impaired recognition of the target stimulus presented to the other hemisphere. The normal control group showed the same result. To rule out the possible contribution of subcortical areas to this interaction, we repeated the same crowding task with texture-defined stimuli. The patient again showed an interhemispheric interaction, even though subcortical structures respond poorly or do not respond at all to texture defined shapes. Despite the evidence for interhemispheric interaction, a classic match-to-sample task confirmed that the patient was unable to explicitly report when stimuli in left and right hemifields were the same or different. Similarly, in a search task, the patient s reaction time was unaffected by distracters in the hemifield opposite the target whereas normals response time was affected. Considering the dissociation between these two tasks, we conclude that the anterior commissure and/or the anterior corpus callosum contribute to interhemispheric interactions in the attentional selection of location Elsevier Science Ltd. All rights reserved. Keywords: Split-brain; Attentional selection; Crowding effect; Visual search 1. Introduction The corpus callosum is a thick neural bundle interconnecting cerebral cortices and is divided anatomically into different parts, named rostrum, genu, trunk and splenium [23]. The rostrum and the genu interconnect prefrontal cortices, the trunk interconnects motor, somatosensory, superior temporal and parietal areas, and the splenium links occipital and inferior temporal areas [2,16]. In addition to the corpus callosum, the anterior commissure (AC) interconnects inferotemporal and occipital areas [4,15]. Psychophysical studies on partial split-brain patients have revealed some of the functions of the different parts of the corpus callosum. It has been shown that when the splenium that interconnects primary visual areas is cut, the patient cannot perform match-to-sample tasks in which he or she has to compare two visual items presented to opposite visual fields [5,19]. Based on these now classic experiments, it was hypothesized that the splenium is the route for interhemispheric transfer of visual information, and that resection of the splenium blocks visual information transfer between the two hemi- Corresponding author. Tel.: ; fax: address: srafraz@ipm.ir (S.R. Afraz). spheres. However, areas such as the temporal, parietal and prefrontal cortices are involved in higher visual functions so there is the possibility that in some higher-level visual tasks the anterior commissure or the anterior corpus callosum be involved in the integration of information across hemifields (in this article, rostrum and a tiny portion of genu are referred to as anterior part of the corpus callosum or ACC). The anterior commissure and the anterior corpus callosum have been considered as pathways for transmitting semantic information [19], and recently as a channel for visual memory signals [21], but there is no psychophysical evidence about the possible role of these cortical commissures (AC and ACC) in interhemispheric visual interactions. Do AC and/or ACC contribute to any kind of visual interaction? To answer this question, we investigated the role of these cortical commissures in different attentional tasks in a partial callosectomy patient in which rostrum, and a tiny anterior portion of the genu were intact. 2. Experiment 1 In the first experiment, we tested a classic crowding paradigm in our patient (MD) and normal controls /02/$ see front matter 2002 Elsevier Science Ltd. All rights reserved. PII: S (02)

2 598 S.R. Afraz et al. / Neuropsychologia 41 (2003) Crowding is a phenomenon in which recognition of a visual target is impaired by other similar visual stimuli (distracters) presented near the target [20]. In the crowding paradigm, the viewer has to report the identity of a visual target placed in a predetermined location known to the subject. Adding distracters to the display near the location of the target makes the identification task harder, but adding distracters far from the target does not have this effect. We test whether distracters placed across the vertical midline degrade identification in our split-brain patient. He et al. [6] have shown that the presence of distracters in a crowded scene blocks conscious access to the target, but does not affect neural responses to the target in the primary visual areas. They suggested that higher cortical areas in dorsal stream that underlie selection of the location by attention are involved in the crowding effect. We assume that in our partial split-brain patient, any interaction across the vertical midline in the crowding paradigm must be mediated by the anterior commissure and/or the remaining anterior corpus callosum. There are two other plausible mechanisms for these interhemispheric interactions in our patient. The first is interaction along subcortical pathways. This possibility will be examined in a subsequent experiment with specially designed stimuli. The second is the potential for competition between the two hemispheres for attentional resources. Specifically the distracters in one hemifield may draw attention from the target in the other hemifield. To rule out this possibility, we varied the distance between the distracters and the target. The relative distance of the stimuli should not affect the competition between hemispheres for attentional resources (given the same number of distracters) Methods The patient (MD) was a 22 years old, right handed female. She had an old scarring near the posterior part of her corpus callosum in the right hemisphere causing intractable generalized seizures. The scarred tissue and posterior parts of the corpus callosum were resected surgically to prevent seizures (in 1996). The anterior commissure and anterior parts (containing rostrum and a tiny portion of the genu) of the corpus callosum (ACC) were spared. Post-operational MRI in 2002 confirmed the anatomic location of the surgical cut and the remaining parts of the corpus callosum (Fig. 1). Her generalized seizures are cured now but she is still dependent to low doses of anti-epileptic drugs. Clinical examinations did not show any neurological deficit in MD, except for an occasional minor clumsiness in her left hand. She did not cooperate to use her left hand for experiments, so she responded with her right hand in all experiments. Her visual acuity was normal. Threshold static automated perimetry did not show any deficit in her visual fields. We conducted identical experiments on two male and two female right handed normal subjects, aged 20 25, with similar education level as patient MD. Fig. 1. Sagittal T2W/TSE MRI section of MD s brain. The splenium and most of the corpus callosum is resected (filled with cerebrospinal fluid which is white for this MRI technique). Stimuli were programmed on a Pentium 233 MHz PC using DMDX package under windows ( arizona.edu/ kforster/dmdx). Images were displayed on a CRT monitor, 800 H 600 V pixel resolution at 60 Hz frame rate (Studioworks 585E, LG, Korea). Subjects were placed in a dimly lit room and their heads were fixed on a chin and forehead rest and viewed the displays binocularly. The distance between eyes and the screen was 40 cm. The target and distracter stimuli were light gray T shapes (oriented in four different directions: left, right, up, or down) on a dark gray background. The size of each stimulus was 2 2 of visual angle, placed at 13.2 eccentricity. Experiment blocks consisted of 64 trials. In half of the trials, called non-crowded trials (Fig. 2a), the display contained the target and only two flanker stimuli in the right visual field (RVF), arranged in a circular array. In the remaining trials, called crowded trials (Fig. 2b), the stimulus array continued into the left visual field (LVF), with seven more flankers. Crowded and non-crowded trials were randomly ordered in each block. Subjects were instructed to fixate at a small red square. The fixation spot was always present during the block. In each trial (crowded or non-crowded) the stimulus array appeared for 180 ms. Subjects had to respond within 2 s after offset of the stimuli. The next trial started 500 ms after response. No feedback was given to the subjects. The target stimulus was always placed in the upper RVF, 1 away from the midline (Fig. 2), 13.2 above the fixation point and it was randomly oriented either to the left or the right (the subjects learned its place before the experiment). Subjects had to report the direction of the target by pressing one of the two buttons on the computer keyboard with the

3 S.R. Afraz et al. / Neuropsychologia 41 (2003) Fig. 2. Crowding effect in the partial split-brain patient. (a and b) Schematic diagrams of non-crowded and crowded trials. A gray vertical line is drawn to show the visual midline (not shown to the subject). The small black square shows the location of the fixation point. (c) Results of experiment 1 showing higher percent correct for non-crowded trials compared to crowded trials in partial split-brain patient whose splenium was resected. index or the middle finger of their right hand. The index finger indicated that the target T is tilted to the left and the middle finger response was required when the target was tilted to the right. There were four types of blocks differed in the distance between the target and the distracters in left visual field. The distance of the right border of the nearest (rightmost) distracter of the left visual field to the target in each condition was 3.8, 5.4, 9.2 and 15 of visual angle, respectively. We will refer to this distance as distracter target distance (DTD). The distance between the target and the nearest distracter in the right visual field and also the spacing of the distracters in left and right visual fields was 2, and did not change across the four types of blocks with different DTDs. All subjects were trained before the main experimental blocks with several blocks until their overall error rate (crowded and non-crowded) became lower than 30%. After training, four experimental blocks for each condition were tested. The experimental blocks for the different conditions were presented randomly. Subjects rested for about 15 min after each two or three blocks. The results were analyzed using Statistica Software (version 98) Discussion These results suggest that some form of interhemispheric transfer of information about distracters occurs. The only difference between crowded and non-crowded trials in this experiment is the presence of seven flankers in the LVF in the crowded trials. These flankers should be registered only in the right hemisphere whereas the target is registered only in the left hemisphere, at least in the early visual areas. How do these LVF distracters affect the responses of the left hemisphere? Competition between hemispheres for attentional resources cannot explain this effect: with the same number of distracters, competition between hemispheres should not be affected by proximity of distracters to the target. The other explanation is that visual information related to the distracters is transferred between the hemispheres. In the next experiment, we investigated whether subcortical mechanisms are involved in this phenomenon Results Subjects had little difficulty in responding to the non-crowded trials. For normal subjects the average error rate was 5.8%. Normal subjects made significantly more errors when distracters were presented to the right hemisphere (all four DTD conditions together: error rate = 18.4%, ratio difference test, P<0.0001). MD also showed an increase in error rates when the distracters were presented (7% versus 17.6%, ratio difference test, P<0.0001, Fig. 2c). Overall performance of the split-brain patient was similar to normal controls for both non-crowded and crowded conditions (P = 0.43, P = 0.74, respectively). Data from all four conditions were pooled. For both MD and normal controls, error rates decreased significantly as the distance between target and distracters increased from 3.8 to 15 (comparing 3.8 and 15, Bonferoni corrected t-test, P<0.05) (Fig. 3). Fig. 3. The effect of distance between distracters and targets on percent correct for MD (a) and average of four normal controls (b). Vertical bars show the standard error of mean; nc: non-crowded.

4 600 S.R. Afraz et al. / Neuropsychologia 41 (2003) Experiment 2 Subcortical regions might mediate the interhemispheric interactions observed in the crowding results of experiment 1. For example, there is evidence showing the involvement of the pulvinar in selective attention [10]. To rule out this possibility, we used the same crowding paradigm but this time, we used texture defined T like shapes as the stimuli (Fig. 4b). We assume that these shapes are poorly represented in subcortical structures. The individual lines of the texture may be well represented but the shape they define may only emerge explicitly at higher cortical levels [3,9,12,13] Methods Subjects were the same as in experiment 1. Each stimulus in this experiment was a T region, filled with an oblique sinusoidal grating with the spatial frequency of 1.5 cycles/ tilted to the right (45 ) in a background of gratings with the same contrast and spatial frequency tilted to the left (135 ) (Fig. 4b). The second experiment was identical to the first experiment except that the size of texture defined T like shapes was The rightmost border of the nearest distracter in the LVF was 3 away from the vertical midline, and the target in RVF was 1 away from the vertical midline. There were two stimuli in non-crowded trials (one target and one flanker in RVF) and five stimuli in crowded trials (three more distracters in LVF) (Fig. 4b). The eccentricity of the stimulus array was 16 from the fixation point. The distance between the target and the nearest distracter in the right visual field and also the spacing of the distracters in left and right visual fields was 4. Subjects were trained before the main experiment, which consisted of four blocks, each containing 64 trials Results and discussion The results again show better performance in non-crowded trials (Fig. 4a) (ratio difference test, P < 0.05) with 92.9% correct for non-crowded trials and 83.6% correct for crowded trials for MD. For normal controls percent corrects were 95.3 and 84.3% for non-crowded and crowded trials, respectively (P <0.05). In this experiment, low-level visual information like brightness, contrast and spatial frequency, which can be mediated by subcortical structures, was not different in the crowded and non-crowded trials. Subcortical structures cannot discriminate texture-defined shapes [3,9,12,13] thus it is unlikely that they can mediate interhemispheric interaction in this experiment. 4. Experiment 3 In order to control results from previous experiments and examine alternative strategies we repeated the classic match-to-sample design with contrast defined and texture defined stimuli used in experiments 1 and 2. This classic experiment ensures that visual feature information do not transfer through the anterior commissure and spared part of the corpus callosum. We presented our stimuli at the same eccentricity used in the first two experiments. According to Iacoboni and Zaidel [8], we expect that motor commands of one hemisphere transmit through subcortical structures and anterior parts of the corpus callosum when required Methods Same subjects and apparatus as the previous experiments were used. Stimuli were T shapes in four orientations (up, down, left, and right). Contrast defined figures with identical shape, contrast and size of the first experiment, or texture defined shapes identical to the experiment 2 were tested in two separate sets of experiments. The fixation point was a small red square in the middle of the screen. In each trial two stimuli with the eccentricity of 13.2 (16 for texture defined stimuli) from the fixation point, in the upper visual field were presented. Two stimuli were shown for 180 ms. The orientations of stimuli were random; in half of the trials they had the same and in the other half they had different orientations. Subjects had to report if the two stimuli were Fig. 4. Texture defined crowding effect in the partial split-brain patient. (a) Results of experiment 2 showing higher percent correct for non-crowded trials compared to crowded trials. (b) Schematic diagram of a crowded trial in texture defined crowding effect experiment. A gray vertical line is drawn to show the visual midline. The small black square shows the location of the fixation point.

5 S.R. Afraz et al. / Neuropsychologia 41 (2003) Fig. 5. Percent correct performance in the match-to-sample task as a function of hemisphere presentation for MD and normal controls. The partial split-brain patient had difficulty comparing two stimuli when they were presented in different visual fields (bilateral condition) and her performance was not significantly different from chance. However, when both stimuli were presented in the same hemisphere, either hemisphere could compare the stimuli and initiate motor response in the left hemisphere (uncrossed and crossed unilateral conditions). the same or not by pressing two buttons with the index and the middle finger of their right hand. Three different types of blocks were tested in random order, in which the two stimuli appeared in three different situations. (1) Bilateral condition: one stimulus appeared in the right visual field 1 away from the vertical midline (exactly at the place of the target in the first and second experiments) and another stimulus in the left visual field exactly at the place of the nearest distracter to the midline in the first and second experiments (2.8 and 3 ). (2) Unilateral uncrossed condition: the two stimuli appeared in the right visual field with the same distance between them and the same eccentricity. The nearer stimulus to the vertical midline was 3 away from it. As the left hemisphere sees the stimuli and responds (with a right handed button push) no information has to pass through the cortical commissures (uncrossed condition). (3) Unilateral crossed condition: both stimuli were presented in the left visual field, with the same distance between them and the same eccentricity. The nearer stimulus to the vertical midline was 3 away from it. So the left hemisphere cannot respond to the stimuli and it needs to rely on the right hemisphere to acquire the visual patterns but the left hemisphere to direct the motor responses (crossed condition). Two blocks of training were used for each condition of the experiment before the main experimental blocks. The experiment for either of the texture defined and the contrast defined stimuli consisted of two blocks of 64 trials for each of the three conditions mentioned above Results For the contrast defined stimuli, the performance of MD in the bilateral blocks where the two stimuli were in separate hemifields was 53.1% (not significantly different from chance). In contrast, the percent correct when both stimuli were in the same hemifield was 81.2% for the uncrossed unilateral condition (significantly above chance level, P< ) and 77.3% for the crossed unilateral condition (also significantly above chance level, P<0.0001). The percent correct of crossed and uncrossed conditions was not significantly different (P = 0.4). For texture-defined stimuli, similar results were obtained. Percent correct was 51.5% (not significant compared with chance level), 76.1% (P <0.0001), and 73.4% (P <0.001) for bilateral, uncrossed, and crossed unilateral conditions, respectively. Results of MD are compared with the results for the normal controls in Fig Discussion These results show that only if we present both stimuli to one hemifield the patient can perform the comparison task. Interestingly, in the crossed condition, even though the left hemisphere had no access to the visual stimuli, the comparison in the right hemisphere could trigger a motor command that, as expected, crossed over to control the response of the right hand. In the bilateral condition the patient could not perform the task because neither hemisphere has access to visual descriptions of both stimuli. These results confirm that visual information about orientation or shape is not transferred between hemispheres when the posterior part of the corpus callosum is cut. 5. Experiment 4 The fourth experiment was a visual search paradigm. When an observer is asked to find a target defined by a conjunction of simple visual features like shape and color in an array of distracters, the reaction time often increases with the number of distracters [22]. Luck et al. [11] have shown that in normal subjects adding a distracter in the field either contralateral or ipsilateral to the target increases the reaction time of the responding hemisphere equally but in split-brain patients, only adding distracters in the visual field ipsilateral to the target increases the reaction time. Experiment 4

6 602 S.R. Afraz et al. / Neuropsychologia 41 (2003) compared these cross-midline distracter effects for normal subjects and the partial split-brain patient Methods The partial split-brain patient MD, and five normal subjects were tested in this experiment. Normal subjects were three females and two males who participated on a voluntary basis. The ages ranged from 21 to 27, all right handed. The apparatus was the same as the previous experiments. The stimuli were red or yellow T like shapes (oriented up or down) in a dark blue background. The size of each stimulus was 2 2 visual angle. The stimuli were arranged in a circular array with eccentricity of 7 around the fixation point, which was a red square in the middle of the screen. In half of the trials (3S trials) only three stimuli were displayed either in the RVF or LVF. In the remaining trials (6S trials) six stimuli were displayed bilaterally, three in RVF and three in LVF. The sequence of trials was completely randomized. The nearest border of the nearest stimulus to the vertical midline was 4 away from it. The display time was 180 ms. The target was an inverted red T and was presented in half of the trials. Subjects had to report presence or absence of the target, by pressing on the left or the right button of the computer mouse, with the index and the middle fingers of their right hand. The main experiment contained four blocks of 64 trials. All subjects in this experiment were trained with two training blocks before the main experimental blocks. To test the visual search ability in the right hemisphere of the patient, we also repeated the same experiment in a unilateral condition in which all the stimuli were presented in the RVF for both 3S and 6S trials. The experiment was identical to the previous experiment, except that the arrangement of the stimuli was not on a circular array. Because the circular array of the main experiment did not have enough space for the three stimuli all on one side, the stimuli were placed randomly in a 7 7 rectangular space 3 away from the vertical midline Results and discussion The performance of MD was quite high (>97%) when the target was presented in the RVF. In contrast, when the target was presented in the LVF the patient always (>97%) reported that the target was absent. We therefore limited the analysis to target present trials with the target in the RVF, correct responses only, for the patient and the normal subjects. Mean reaction time for 3S trials for MD was ms. Mean reaction time for her 6S trials was ms. There was no significant difference between the 3S and 6S trials for MD (t-test, P = 0.2). The mean difference between 3S and 6S trials for the five normal subjects was 67.2 ms, a significant increase (t-test, P<0.05 for all of them). In fact, each normal subject individually showed a significant increase in reaction times for 6S trials compared to the 3S trials (Fig. 6). Fig. 6. Visual search reaction time for correct responses to targets in the RVF (experiment 4) for patient (MD) compared to a representative normal subject (results for the other five normal subjects were similar). White bars show mean reaction time and standard errors for 3S trials with only three stimuli in RVF, black bars show the same for trials with three more distracters in LVF and the target in RVF. Distracters presented in the LVF increase the reaction times to targets in the RVF for all normal subjects. For MD, there was no significant difference between the two kinds of trials. The texture-filled bars show reaction times of the target present trials of the complementary experiment with all the stimuli in RVF. The patient and normal controls all showed increased reaction times for the trials with six stimuli in RVF. The results of the experiment with unilateral stimuli showed significant increase in the reaction time of 6S trials compared to 3S trials in both MD (716.4 ms versus ms) and normal subjects (709.2 ms versus ms) (both P < ) (Fig. 6). This shows that the visual search ability of the left (responding) hemisphere in the patient is the same as normal subjects. These results show that in normal subjects, distracters from the right hemisphere interfere with target processing in the left hemisphere but not in our partial split-brain patient. Luck et al. have shown the same effect in total split-brain patients and suggest that the attentional systems used for visual search are independent in the two hemispheres [11]. Highly salient distracters may interfere with the other hemisphere even in totally split patients [18], but in the case of the shape-color conjunction search our results failed to show interhemispheric interference. Low salient distracters in a pop-out task may not interfere with the contralateral hemisphere even in normal subjects [17,18], but in our case all of the stimuli (target and distracters) had the same saliency. Why was our partial split-brain patient unable to detect the target when it was in her LVF? Recall that the patient was able to perform the match-to-sample when both stimuli were in the LVF where, as here, she responded with the right hand. We are not certain why there was this difference between the two tasks but the complexity of the task is a possibility. Specifically, unlike the match-to-sample, the visual search task requires a match to memory, and the right hemisphere may not understand the more complex search task or have access to the appropriate memory. On the other hand, in the visual search task, the left hemisphere is instructed to say whether an inverted red T is present or absent, thus, when the target T is not presented directly to the left hemisphere, the answer is clearly no and it do not need to wait for

7 S.R. Afraz et al. / Neuropsychologia 41 (2003) the other hemisphere. But in the match-to-sample task, the question is not about presence or absence of an stimulus and the left hemisphere has to judge about the similarity of the stimuli which are not visible for it! So the left hemisphere may rely on motor commands from the other hemisphere only in such ambiguous conditions. Whatever the cause of the loss of target detection in the LVF, the critical result is that distracters in the LVF did not disrupt processing of the target in the RVF for the patient but did for the normals. 6. General discussion In our first experiment, we found that although primary visual areas of the two hemispheres were disconnected in our patient (see Section 2.1) the visual interactions underlying crowding still operated across the midline. In texture defined shapes (experiment 2), the patterns presented inside and outside of the T like regions are the same in luminance, contrast, and spatial frequency. Therefore, subcortical structures are likely to be activated similarly for both the pattern and the background so the subcortical projections would not convey information about texture boundaries. Moreover, the textures we used required cells with orientation selectivity to discriminate them and these cells are not found earlier than V1 cortex in the visual pathway. There are single unit recording studies [13], fmri [9], ablation studies [3], and also some theoretical analyses [12] confirming that such texture-defined shapes (different only in orientation) are not discriminated by subcortical structures. So crowding interactions in this experiment, which were similar for contrast-defined and texture-defined stimuli, are unlikely to be mediated by subcortical structures, leaving only AC and ACC as reasonable candidates. The first two experiments reveal an interaction between visual information in the two hemispheres, probably mediated through AC and/or ACC. Is this evidence in favor of the transmission of visual features (like shape, texture and color) by the AC and/or ACC, or of some other high-level visual interactions? The third experiment revealed that MD could not compare the shapes of two stimuli presented bilaterally to different visual fields. Thus, we conclude that at least simple visual properties like shape are not transmitted through the AC and ACC. An important question is the nature of the interaction that mediated the crowding across the midline for the partial split-brain patient. Both crowding and visual search tasks require the identification of the target and yet distracter effects crossed the midline for the patient for the crowding task but not for the visual search task. The difference between crowding experiments (experiments 1 and 2) and the visual search experiment (experiment 4) is that during a crowding paradigm an unknown target must be identified at a specific, known location. The presence of crowding distracters near the target disturbs the attentional selection of the target location [20]. Reducing the distance of the distracters from the target increases the crowding effect, and this phenomenon is considered evidence for location-based attentional selection [1,7]. In contrast with the crowding paradigm, conjunction visual search requires that the attentional system search for a known target at an unknown location [14]. In our patient MD, LVF distracters affected responses of the left hemisphere in crowding task but not in visual search. We think that the dissociation found here is the result of the difference in the nature of the two tasks. The crowding effect is the result of the interference of nearby distracters on the selection of information from a known location. Only a single selection is attempted. The data of our patient shows that interference on the selection of a location in one hemisphere can arise from distracters in the other hemisphere despite the partial split of the corpus callosum. In the visual search task, in contrast, the distracter interference arises because each distracter must be selected in turn until the target is identified. Several locations are selected. However, in the split-brain patient, the location selection is limited to items within each hemifield (and proceeding independently in both according to Luck et al. [11]). While attentional processes in the left hemisphere are selecting distracters and eventually the target from among items in the RVF, distracters from the other hemifield are never selected. Our interpretation leads to two further points. In the visual search task, stimuli were fairly widely spaced so that there was little crowding between the search items. If the stimuli were more densely packed, crowding may slow the visual search speed and reduce its accuracy. In this case, we would expect some effect of distracters from across the midline but only because we had introduced crowding into the search task. Furthermore, in the crowding task, as in the visual search and match-to-sample tasks, we expect that for patient MD, the selection mechanisms in one hemisphere have no primary access to stimuli in the other. This means that, in the crowding task, the patient could not have responded to the target in the LVF when still using her right hand to respond. We say primary access because she would not be able to report or match the stimulus. Nevertheless, the crowding results show that this primary access to the target identity somehow involves polling of a larger area and that larger area surprisingly includes regions of space across the midline despite the absence of most of the corpus callosum. He et al. [6] proposed that the dorsal visual stream, which is involved in visual spatial information processing, plays a critical role in crowding effect. We think that the transfer of crowding in our patient is the result of interhemispheric interactions during location selection in the high level areas of the dorsal stream mediated through AC and/or ACC. The results of the visual search and match-to-sample experiments show that the retrieval of the selected information from the other hemisphere cannot be mediated through AC and/or ACC. Nevertheless, the access to the location

8 604 S.R. Afraz et al. / Neuropsychologia 41 (2003) itself appears to involve a larger area within which nearby distracters, even those across the midline, create interference in the retrieval. In a normal subject, performance in a typical attention selection task is determined by both uncertainties of the location to be selected and interference from items at nearby locations. In tasks with distracters and targets in separate hemifields, the partially split-brain of patient MD gives us a unique opportunity to study the nature and underlying anatomy of interference from nearby items in isolation from uncertainty in the location to be selected. Acknowledgements We thank Prof. Patrick Cavanagh for his useful comments, Dr. Khosro Parsa for his help, and especially MD for her cooperation. We also would like to thank the editor for his suggestion about experiment 1. References [1] Bahcall DO, Kowler E. Attentional interference at small spatial separations. Vision Research 1999;39: [2] De Lacoste MC, Kirkpatrick JB, Ross ED. Topography of the human corpus callosum. Journal of Neuropathology and Experimental Neurology 1985;44: [3] De Weerd P, Desimone R, Ungerleider LG. Cue-dependent deficits in grating orientation discrimination after V4 lesions in macaques. Visual Neuroscience 1996;13: [4] Di Virgilio G, Clarke S, Pizzolato G, Schaffner T. Cortical regions contributing to the anterior commissure in man. Experimental Brain Research 1999;124(1):1 7. [5] Gazzaniga MS. Principles of human brain organization derived from split-brain studies. Neuron 1995;14: [6] He S, Cavanagh P, Intriligator J. Attentional resolution and the locus of visual awareness. Nature 1996;383: [7] He S, Cavanagh P, Intriligator J. Attentional resolution. Trends in Cognitive Science 1997;1: [8] Iacoboni M, Zaidel E. Crossed uncrossed difference in simple reaction times to lateralized flashes: between- and within-subjects variability. Neuropsychologia 2000;38: [9] Kastner S, De Weerd P, Ungerleider LG. Texture segregation in the human visual cortex: a functional MRI study. Journal of Neurophysiology 2000;83: [10] LaBerge D, Buchsbaum MS. Positron emission tomographic measurements of pulvinar activity during an attention task. Journal of Neuroscience 1990;10: [11] Luck SJ, Hillyard SA, Mangun GR, Gazzaniga MS. Independent hemispheric attentional systems mediate visual search in split-brain patients. Nature 1989;342: [12] Malik J, Perona P. Preattentive texture discrimination with early vision mechanisms. Journal of the Optical Society of America A 1990;7: [13] Nothdurft HC. Texture discrimination by cells in the cat lateral geniculate nucleus. Experimental Brain Research 1990;82: [14] Palmer J, Verghese P, Pavel M. The psychophysics of visual search. Vision Research 2000;40: [15] Pandya DN, Rosene DL. Some observations on trajectories and topography of commissural fibers. In: Jeeves AG, editor. Epilepsy and the corpus callosum. New York: Plenum Press; p [16] Pandya DN, Seltzer B. The topography of commissural fibers. In: Lepore F, Ptito M, Jasper HH, editors. Two hemispheres one brain: functions of the corpus callosum. New York: Liss; p [17] Pollmann S. Extinction-like effects in normals: independence of localization and response selection. Brain Cognition 2000;44(3): [18] Pollmann S, Zaidel E. The role of the corpus callosum in visual orienting: importance of interhemispheric visual transfer. Neuropsychologia 1998;36(8): [19] Sidtis JJ, Volpe BT, Holtzman JD, Wilson DH, Gazzaniga MS. Cognitive interaction after staged callosal section: evidence for transfer of semantic activation. Science 1981;212: [20] Toet A, Levi DM. The two-dimensional shape of spatial interaction zones in the parafovea. Vision Research 1992;32: [21] Tomita H, Ohbayashi M, Nakahara K, Isoa H, Miyashita Y. Top-down signal from prefrontal cortex in executive control of memory retrieval. Nature 1999;401: [22] Treisman A, Sato S. Conjunction search revisited. Journal of Experimental Psychology: Human Perception and Performance 1990;16: [23] Williams PL, et al. Gray s anatomy. Edinburgh: Churchill Livingstone; p