The Neural Substrates of Visual Implicit Memory: Do the Two Hemispheres Play Different Roles?

The Neural Substrates of Visual Implicit Memory: Do the Two Hemispheres Play Different Roles? N. E. A. Kroll 1, A. P. Yonelinas 1, M. M. Kishiyama 1, K. Baynes 1, R. T. Knight 2, and M. S. Gazzaniga 3 Abstract & Identification of visually presented words is facilitated by implicit memory, or visual priming, for past visual experiences with those words. There is disagreement over the neuroanatomical substrates of this form of implicit memory. Several studies have suggested that this form of priming relies on a visual word-form system localized in the right occipital lobe, whereas other studies have indicated that both hemispheres are equally involved. The discrepancies may be related to the types of priming tasks that have been used because the former studies have relied primarily on word-stem completion tasks and the latter on tasks like word-fragment completion. The present experiments compared word-fragment and word-stem measurements of visual implicit memory in patients with right occipital lobe lesions and patients with complete callosotomies. The patients showed normal visual implicit memory on fragment completion tests, but essentially no visual priming on standard stem completion tests. However, when we used a set of word stems that had only one correct solution for each test item, as was true of the items in the fragment completion tests, the patients showed normal priming effects. The results indicate that visual implicit memory for words is not solely dependent upon the right hemisphere, rather it reflects changes in processing efficiency in bilateral visual regions involved in the initial processing of the items. However, under conditions of high lexical competition (i.e., multiple completion word stems), the lexical processes, which are dominant in the left hemisphere, overshadow the visual priming supported by the left hemisphere. & INTRODUCTION 1 University of California Davis, 2 University of California Berkeley, 3 Dartmouth College In tests of explicit memory, such as recognition and free recall, subjects are requested to actively remember previous events. In contrast, the instructions for tests of implicit memory either make no reference to previous events or even tell the subject to ignore them. For example, in a word-stem completion test, subjects are first exposed to a series of words (e.g., salmon) and are later shown word stems (e.g., sal ) and are instructed to complete the stem with the first word that comes to mind. Subjects are more likely to complete a stem with a word if it had been previously seen (i.e., they exhibit priming effects), even if they do not consciously recollect having seen that word previously. Another frequently used implicit memory task is the word-fragment completion test, which is similar to the stem completion task except that subjects are presented with fragmented words at test (e.g., e_e_an_e for the word elegance). What makes the explicit/implicit distinction particularly interesting is that patients with damage to the medial temporal lobes perform more poorly than normal subjects on tests of explicit memory, but perform normally on tests of implicit memory (e.g., Schacter, Chiu, & Ochsner, 1993). This indicates that implicit and explicit forms of memory rely on partially distinct neuroanatomical substrates. These two forms of memory also exhibit distinct functional properties. For example, changes in the perceptual format of words between study and test, such as from uppercase letters to lowercase letters or from auditory to visual presentation, greatly reduces the priming effects observed in implicit tests, such as stem completion (for reviews, see Roediger & McDermott, 1993; Schacter et al., 1993). In contrast, similar changes in perceptual format often do not affect performance on explicit tests. The sensitivity of implicit memory performance to changes in perceptual format has been taken as evidence that these tests rely on form-specific sensory memory processes or systems (Schacter, 1992; Roediger & Blaxton, 1987; Jacoby, 1983). However, significant priming effects are observed even when the study and test format changes, indicating that implicit memory tasks can also rely in part on abstract form-independent memory processes, such as phonological, lexical, or semantic processes (e.g., Curran, Schacter, & Galluccio, 1999; Rueckl & Mathew, 1999). Several studies using a divided visual field technique in healthy subjects have D 2003 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 15:6, pp. 833 842

suggested that the form-specific component of visual implicit memory is supported primarily by the right hemisphere, whereas the abstract component is supported equally by both hemispheres (e.g., Marsolek & Hudson, 1999; Marsolek, Kosslyn, & Squire, 1992). Another set of studies that support the hypothesis that structures in the right hemisphere are required for the form-specific component of visual implicit memory are based on the study of a patient with a right hemisphere occipital lobe lesion. This patient has been studied quite extensively and consistently exhibits a severe perceptual priming deficit in word-stem completion tests, despite normal performance on a variety of explicit memory tests (e.g., Vaidya, Gabrieli, Verfaellie, Fleischman, & Askari, 1998; Fleischman et al., 1995; Gabrieli, Fleischman, Keane, Reminger, & Morrell, 1995; Keane, Gabrieli, Mapstone, Johnson, & Corkin, 1995), suggesting that the right occipital lobe plays a necessary role in implicit visual priming but not explicit memory. However, other studies suggest that both hemispheres are involved in supporting visual implicit memory. For example, divided visual field studies using the word-fragment completion task (Kroll, Rocha, Yonelinas, Baynes, & Frederick, 2001) and the perceptual identification task (e.g., Burgund & Marsolek, 1997; Koivisto, 1995) have indicated that both hemispheres exhibit equivalent visual priming effects. Moreover, in a previous study ( Yonelinas et al., 2001), we evaluated the contribution of the two hemispheres to visual implicit memory by examining patients who suffered from extensive right occipital lobe lesions and patients whose left (verbal) hemisphere had been surgically separated from the right by means of a callosotomy. If the right occipital lobe is necessary for visual implicit memory, then, we reasoned, both of these types of patients should exhibit a pronounced deficit on tests of visual implicit memory. 1 We measured implicit memory using four different visual implicit tests to determine if the results generalized across a variety of test procedures (i.e., lexical decision, mirror reading, picture fragment completion, and visual word-fragment completion). In all of these priming tests, both types of patients presented with normal visual priming, suggesting that the right hemisphere was not necessary for visual priming. Why are the previous studies in disagreement about the role of the right hemisphere in visual implicit memory? One possible explanation for the discrepancies observed in the patient studies is that only certain types of occipital lobe lesions might lead to visual priming deficits. Given that the different patients have not been tested on the same test procedures it is not possible to directly rule out this possibility. However, differences in lesions could not explain the differences observed across different implicit tests in the divided visual field studies of healthy subjects. A more promising explanation is that the discrepancies arose because of the different retrieval tasks that have been used in previous studies. The studies that have indicated a dominant role for the right hemisphere have relied almost exclusively on the word-stem completion task (e.g., Marsolek & Hudson, 1999; Vaidya et al., 1998; Fleischman et al., 1995; Gabrieli et al., 1995; Keane et al., 1995; Marsolek et al., 1992), whereas the studies that have indicated no special role for the right hemisphere have used a variety of other visual implicit memory tests, such as word-fragment completion and perceptual identification (Kroll et al., 2001; Yonelinas et al., 2001; Burgund & Marsolek, 1997; Koivisto, 1995). Although the stem completion task is similar to other visual implicit tests there is evidence that it relies less on perceptual processes and more on phonological or semantic processes than do tasks like word-fragment completion (Blum & Yonelinas, 2001; Rueckl & Mathew, 1999; Bassili, Smith, & MacLeod, 1989). Thus, differences in the processing demands of the retrieval tasks might lead to differences in how the two hemispheres contribute to performance. In the current study, we examined stem and fragment completion priming in patients with extensive right occipital lobe lesions and callosotomy patients in whom the left hemisphere had been surgically separated from the right. We first report results on standard wordfragment completion (Experiment 1) and word-stem completion tasks (Experiment 2) to determine whether the previous discrepancies were due to differences in patient populations, or to differences in retrieval tasks. The results indicated that the patients exhibited normal fragment priming but were impaired on stem priming, showing that there is a critical difference in the manner in which these two tasks rely on the two hemispheres. Experiments 3 and 4 examined fragment and stem completion priming, respectively, using the same sets of words, and again showed reduced stem priming and normal fragment priming in the patients, indicating that subtle differences in materials were not responsible for the observed dissociation. Experiment 5 was designed to determine if the reduced priming in the stem task was due to the fact that stems typically have many competing solutions, whereas fragments typically have only one, by using stems that had only one correct solution. The same patients who exhibited stem completion priming deficits on multiple completion stems were now found to exhibit normal priming on the single completion stems, indicating that the role of the right hemisphere in priming is determined by the extent to which there are multiple competing items for a given retrieval cue. RESULTS The scores on the implicit memory tests are presented in Table 1, and the visual priming scores are presented in Figures 1 and 2. For each of the experiments, a subject s visual priming score was measured as the 834 Journal of Cognitive Neuroscience Volume 15, Number 6

Table 1. (continued ) Table 1. Percentage of Correct Completions for Each of the Patients Tested and Median Percentage of Correct Completions for Each of the Control Groups Tested in Each of the Conditions (New, Auditorily Primed, Visually Primed) for Each of the Experiments Patient/Group New Condition Auditorily Primed Visually Primed Experiment 1: Eight-letter word fragments Student controls 0.45 0.60 0.80 Aged-matched controls 0.48 0.53 0.75 Right occipital lobe patients C. B. 0.38 0.45 0.78 F. N. 0.38 0.48 0.70 J. C. 0.08 0.23 0.40 N. A. 0.30 0.53 0.80 Callosotomy patients D. R. 0.15 0.35 0.58 J. W. 0.23 0.45 0.60 V. J. 0.13 0.40 0.58 Experiment 2: Standard word stems Student controls 0.10 0.18 0.33 Right occipital lobe patients C. B. 0.00 0.25 0.30 F. N. 0.10 0.15 0.18 J. C. 0.15 0.15 0.20 N. A. 0.10 0.30 0.33 Experiment 3: Five-letter word fragments Student controls 0.28 0.33 0.44 Right occipital lobe patients C. B. 0.19 0.32 0.57 F. N. 0.22 0.32 0.50 N. A. 0.28 0.33 0.52 Experiment 4: Five-letter word stems with multiple solutions Student controls 0.19 0.32 0.50 Aged-matched controls 0.24 0.30 0.46 Right occipital lobe patients C. B. 0.22 0.32 0.33 F. N. 0.13 0.35 0.35 N. A. 0.22 0.41 0.43 Patient/Group New Condition Auditorily Primed Visually Primed Experiment 4: Five-letter word stems with multiple solutions Callosotomy patient J. W. 0.37 0.44 0.48 Experiment 5: Five-letter word stems with unique solutions Student control 0.44 0.54 0.67 Aged-matched control 0.48 0.56 0.69 Right occipital lobe patients C. B. 0.67 0.74 0.83 F. N. 0.28 0.41 0.54 N. A. 0.33 0.39 0.56 Callosotomy patient J. W. 0.48 0.67 0.74 difference between the percentage of correct completions of words previously seen and the percentage of correct completions of words previously heard (i.e., the heard words, which should have also primed lexical processing components, served as a baseline). This difference score was used to obtain a measure of priming that was based solely on the visual component of the stimulus. A completed fragment or stem was judged as correct if it matched the corresponding word from the original word list (i.e., the word that had been seen or heard). Not all patients were available for testing on every experiment, but if they were tested on an experiment, those results are reported. Note that most of the patients were older than the student control subjects. Although visual priming is not generally influenced by aging (e.g., Rybash, 1996), to confirm that aging did not play a critical role in the current studies, an older control group was included in Experiments 1, 4, and 5. Experiment 1 Experiment 1 is an extension of a word-fragment completion experiment that we reported previously ( Yonelinas et al., 2001). The left panel of Figure 1 presents the visual priming performance for fragment completion that we reported in that earlier experiment on 2 occipital lobe patients (F. N. and J. C.), 3 callosotomy patients (D. R., V. J., and J. W.), and the median of the 16 normal older control subjects. The priming performance of a third occipital lobe patient (C. B.) and 27 student controls added in the present experiment is Kroll et al. 835

Figure 1. (A) Visual priming scores for the fragment completion task of Experiment 1. (B) Visual priming scores for the stem completion task of Experiment 2. The data for the 27 student control subjects in Experiment 1 and the 16 student control subjects in Experiment 2 are presented in a box-and-whisker format, with the median shown as a horizontal line, the upper and lower quartiles shown as the upper and lower edges of the box, and the most extreme scores shown as the ends of the whiskers. The median of the 16 older control subjects is represented by a circle. The scores of patients with right occipital lobe lesions ( J. C., F. N., N. A., and C. B.) are represented by square-shaped symbols; those of the callosotomy patients ( J. W., D. R., and V. J.) are represented by diamond-shaped symbols. also included in that figure. Before we discuss the priming scores, it is appropriate to compare the baseline (priming for words heard) performance. The older normal subjects had a baseline of M =.61, SD =.19, and the students had a baseline of M =.62, SD =.11. The baseline for the five patients, however, was considerably lower, M =.42, SD =.13. However, differences in baseline did not appear to influence the priming effects on the fragment completion test. The eight older control subjects with the lowest baseline completion rates had baselines that were comparable to the patients (M =.38, SD =.10) and the priming scores for these control subjects were not appreciably different from the subjects with higher baseline performance. Note that no baseline differences were observed in the subsequent experiments. All of the right occipital lobe lesion and callosotomy patients exhibited robust priming effects on the fragment completion test (Figure 1, left panel). The patients were all within the range of the student controls, with the exception of one, C. B., who exhibited slightly greater priming than the controls. The results are clear in showing that right occipital lobe damage does not disrupt visual priming on the word-fragment completion task and that the isolated left hemisphere is capable of supporting normal visual priming on this task. The results are consistent with a previous report showing that these patients exhibited normal visual priming on perceptual identification, mirror reading and picture fragment completion tests ( Yonelinas et al., 2001). Experiment 2 Experiment 2 was similar to Experiment 1 except that priming was measured using a stem completion, rather than a fragment completion task. The right panel of Figure 1 presents the visual priming performance for stem priming for 4 occipital lobe patients (C. B., J. C., F. N., and N. A.) and for 16 student controls. In contrast to the results from the fragment completion test in which the patients exhibited normal priming scores, all four of the patients tested had extremely low stem completion priming scores scores matched by only the lower quartile of the control subjects. The deficit in stem completion priming is consistent with the stem completion deficits reported for a patient with right occipital lobe damage (e.g., Vaidya et al., 1998; Fleischman et al., 1995; Gabrieli et al., 1995; Keane et al., 1995) and confirms that the deficit observed in that patient is also observed in other similar patients. Based on Experiments 1 and 2, it is difficult to determine why the word-stem and word-fragment tests make different demands on the two hemispheres because the two tests differ in a number of ways. For example in most fragment completion tests, including the one used in Experiment 1, the items are low-frequency words. In 836 Journal of Cognitive Neuroscience Volume 15, Number 6

contrast, in stem completion tests, including Experiment 2, higher frequency words are used. Other item differences, such as the linguistic class of words used and word length, also differ. Thus, it is possible that differences in materials were responsible for the different outcomes in these two tasks. To address this issue, we examined fragment and stem completion priming in Experiments 3 and 4, respectively, using the same set of words in both experiments. Experiment 3 Experiment 3 tested for fragment priming with five-letter words. The leftmost panel of Figure 2 presents the visual priming performance for fragment completion for 3 occipital lobe patients (C. B., F. N., and N. A.) and 15 student controls. All three of the patients measured had priming scores that fell within the normal range. The results are consistent with those of Experiment 1 in showing that damage to the right occipital lobe does not reduce visual priming effects seen in the fragment completion test. Experiment 4 Experiment 4 tested for stem priming with the same fiveletter words used in the fragment test in Experiment 3. The center panel of Figure 2 presents the visual priming performance for stem completion for 3 occipital lobe patients (C. B., F. N., and N. A.), 1 callosotomy patient ( J. W.), the median for 5 normal older control subjects, and 15 student controls. While the median performance of the older controls is very close to that of the students, the priming scores of all four of the patients fall below that of all of the control subjects. The results are consistent with previous studies, including Experiment 2, indicating that the damage to the right occipital lobe or isolation of the left hemisphere leads to a deficit in visual priming in stem completion. Moreover, because the words were the same as those used in the fragment completion test in Experiment 3, the results indicate that the differences between fragment and stem completion priming cannot be attributed to differences in the type of words used in those tasks. One other potentially critical difference between the stem, and fragment completion tests is in the number of competing completions for each test cue. For most word fragments, there is only a single correct completion (e.g., the only English word that can complete e_e_an_e is elegance), whereas most word stems have numerous correct completions (e.g., sal can be completed with salmon, sales, salute, etc). It is possible that differences in the number of competitors may have led to the dissociation between the two tasks. To test this hypothesis, we developed a set of words that had only a single obvious five-letter completion for each word s stem. Figure 2. (A) Visual priming scores for the five-letter word fragment completion task of Experiment 3. (B) Visual priming scores for the fiveletter word-stem completion task of Experiment 4. (C) Visual priming scores for the fiveletter word unique stem completion task of Experiment 5. The data for the 15 student control subjects in each of Experiments 3 5 are presented in a box-and-whisker format. The medians of the five older control subjects who were tested in both Experiments 4 and 5 are represented by circles. The scores of patients with right occipital lobe lesions (F. N., N. A., and C. B.) are represented by square-shaped symbols; those of the callosotomy patient ( J. W.) are represented by diamond-shaped symbols. Kroll et al. 837

Experiment 5 Experiment 5 was similar to Experiment 4, but used word stems that could be completed with only one fiveletter word. The right panel of Figure 2 presents the visual priming performance for stem completion for 3 occipital lobe patients (C. B., F. N., and N. A.), 1 callosotomy patient ( J. W.), the median for 5 normal older control subjects, and 15 student controls. With the unique-solution stem completion task, the priming scores of all four of the patients were normal, indicating that stem completion priming is not impaired by right occipital lobe damage or by isolation of the right hemisphere as long as there are no competing completions to the word stems. DISCUSSION Previous studies have led to conflicting results regarding the importance of the right hemisphere for visual implicit memory for verbal materials. The current study indicates that the discrepancies were due to the different retrieval tasks that were used. Namely, the right hemisphere is not necessary for visual priming as measured on fragment completion or on stem completion tasks in which there is a single possible completion. However, for word stems with multiple completions, damage to the right hemisphere does disrupt visual priming. The finding that normal visual priming on multiple completion stems requires the right hemisphere is consistent with previous studies of a patient with severe right occipital lobe damage who did not exhibit visual priming on the stem completion tests (Vaidya et al., 1998; Fleischman et al., 1995; Gabrieli et al., 1995; Keane et al., 1995) and with divided visual field studies of healthy subjects that indicate that the left hemisphere does not exhibit visual priming effects as measured on the stem completion task (Marsolek et al., 1992; Marsolek & Hudson, 1999). Importantly, all of these studies utilized multiple completion word stems. The finding that fragment completion priming was not disrupted by right occipital lobe damage or by isolation of the right hemisphere is consistent with previous studies of these patients that showed that visual priming on a variety of different tests was not impaired ( Yonelinas et al., 2001) and with divided visual field studies of fragment completion in healthy subjects who exhibit comparable visual priming in the left and right hemispheres (Kroll et al., 2001). The current results are important in showing that visual implicit memory is not rigidly localized in the right hemisphere. That is, the left hemisphere can support normal visual priming when the right occipital lobe has been severely damaged and when the right hemisphere has been isolated from the left. Thus, the results do not support claims that implicit visual priming effects are based on a visual word-form system that is localized in the right hemisphere (Gabrieli, 1998; Schacter, 1994; Schacter & Tulving, 1994; Squire, 1994). Marsolek (1999) proposes that visual implicit memory depends upon multiple components. Form-specific visual priming and abstract visual priming are two of these components that he assumes operate in parallel (rather than in sequence), with the form-specific system operating most efficiently in the right hemisphere and the abstract system operating most efficiently in the left. In this way, the right hemisphere would generally produce the most robust visual priming effects, but task demands might be altered in such a way that visual and abstract priming could be observed in both hemispheres. How, though could the task demands of the different implicit memory tests have produced the pattern of results seen in the current study? In particular, why does the left hemisphere support visual priming on most priming tasks but not with multiple completion word stems? One possibility is that with multiple completion stems, the visual priming effects supported by the isolated left occipital lobe are masked by the contribution of lexical processes that are dominant in that hemisphere. That is, the left hemisphere, like the right, supports visual priming of words, in the sense that the visual networks initially involved in processing an item are primed and thus facilitate reprocessing of the item at time of test. The left hemisphere differs from the right, however, in the sense that lexical processes dominate over visual processes in the left hemisphere. Thus, in the case of multiple completion stems, the lexical system will typically be capable of producing a valid word completion. The lexical system is dominant in the left in the sense that if the lexical system produces a completion, then that completion will be selected for output. Although the left hemisphere visual system may have been primed for the studied item, its contribution will be masked or preempted by the lexical process. In contrast, with single completion word stems, it is not necessary to select between alternative responses and, thus, the contribution of the visual system should be unobscured. Other priming tasks like fragment completion are similar to single completion stems in the sense that the effect of visual (form) processes is observed because the lexical processes cannot produce a valid completion for every test item and thus the less dominant visual processes will contribute to performance. This account is also consistent with the pattern of results observed in divided visual field studies. That is, visual priming effects will be seen in the left hemisphere with single completions because both visual and lexical processes contribute to performance, whereas with multiple completion stems, visual priming effects will be masked by the dominant lexical processes in that hemisphere. In contrast, in the right hemisphere, the lexical process is less likely to be effective, resulting in 838 Journal of Cognitive Neuroscience Volume 15, Number 6

visual priming effects in that hemisphere for single and multiple completion tasks. The current results show that the standard stem completion task is somewhat unusual, compared to other visual priming tasks, in that it is disrupted by right occipital damage or when the right hemisphere is isolated. Thus, the study joins a growing body of research indicating that word-stem completion is distinct from other visual implicit memory tests. For example, as previously discussed, earlier studies indicated that stem completion is less perceptual and more sensitive to semantic variables than are the other visual priming tasks (e.g., Blum & Yonelinas, 2001; Rueckl & Mathew, 1999; Bassili et al., 1989). The current study focused on visual priming for verbal materials and indicated that it was not localized specifically to the right hemisphere. However, whether visual priming for nonverbal materials exhibit laterality effects is not yet clear. There is some evidence to suggest that the right hemisphere does play a more important role in nonverbal priming effects. For example, larger visual priming effects have been observed in the right than left hemisphere using divided visual field methods (Marsolek, 1999), and neural deactivations associated with priming of visual objects have been reported to be more pronounced over the right than left hemispheres (e.g., Henson, Shallice, Gorno-Tempini, & Dolan, 2002; Vuilleumier, Henson, Driver, & Dolan, 2002; although also see Elliott & Dolan, 1998). Further studies examining visual priming in patients with left and right occipital lobe damage will be useful in determining the role of the two hemispheres in nonverbal visual priming. An explanation of implicit visual memory that is consistent with the existing results is that visual priming is not the product of a single localized memory system that is rigidly localized in one hemisphere or the other, but that it rather reflects the general by-products of previous visual processing. Thus, when both hemispheres are involved in processing visual stimuli, as in healthy subjects, then both hemispheres will exhibit evidence of priming. In contrast, when the left hemisphere is responsible for processing visual stimuli, as with the lesion patients in the current study, then the left hemisphere will exhibit visual priming effects when these priming effects are not masked by other, more compelling left hemisphere processes. METHODS Subjects Four patients with right occipital lobe damage were tested. Three of these patients had incurred right posterior cerebral artery infarcts with right occipital and right inferior temporal occipital lesions (patients J. C., F. N., and N. A.). Patient J. C. suffered a perinatal right posterior cerebral artery infarct resulting in extensive damage to the right occipital temporal cortex. Magnetic resonance imaging (MRI) revealed complete destruction of the right occipital lobe with damage involving V1, V2, V3, V4, significant amounts of the fusiform and lingual gyri as well as the posterior hippocampal and parahippocampal regions. There was a thin segment of residual tissue extending through the region of extensive cystic encephalomalacia likely representing scar tissue. Patient F. N. suffered bilateral posterior cerebral artery infarcts. There was minimal damage (<1 cm) in the V2 region of the left occipital pole. No visual deficits were detected in the right visual field. Damage in the right hemisphere included all of V1 and medial portions of V2 and V3. The lesion included posterior portions of the fusiform and lingual gyri extending to the margin of the hippocampal and parahippocampal gyri. Damage did not extend to the lateral convexity. Both patients had dense left hemianopsias and neither patient had any evidence of residual form discrimination in their left visual fields (these two patients are discussed in detail in Wessinger, Fendrich, & Gazzaniga, 1997). Patient N. A. sustained an infarct of the right posterior cerebral artery. Damage included extensive loss of V1 and infarction of medial portions of V2 and V3. There was additional extensive damage to fusiform, lingual, hippocampal, and parahippocampal cortices. Patient N. A. had a dense left hemianopsia and difficulty with spatial memory. The fourth patient, C. B., suffered a closed head injury during an equestrian accident requiring surgery to remove a hematoma in the right occipital region. Her CT scan revealed extensive damage to the right occipital and parietal lobes. There was complete loss of V1, V2, V3 as well as extensive damage to the fusiform lingual, hippocampal, and parahippocampal cortices. Damage extended superiorly to include the right inferior and parietal cortices. Patient C. B. had a dense left hemianopsia. In addition, three patients with complete callosotomies (patients V. J., J. W., and D. R.) were tested. For these patients, the callosotomies were carried out in two stages as treatment for intractable epilepsy. No extracallosal damage was noted in V. J. or J. W., but D. R. s MRI revealed spared fibers in the rostrum and an area of hyperintense signal in the region of the quadrigeminal cistern. Neither D. R. nor V. J. was able to make verbal responses to stimuli presented to the left visual field (Baynes, Tramo, & Gazzaniga, 1992; Baynes, Eliassen, Lutsep, & Gazzaniga, 1998), and J. W. s ability to make verbal responses to left visual field stimuli is limited (Gazzaniga et al., 1996; Baynes, Wessinger, Fendrich, & Gazzaniga, 1995). The demographic and neuropsychological data for the lesion patients is presented in Table 2 and Figure 3. Standard test scores were obtained after trauma. The patients performed well on the IQ and memory tests with the following exceptions. D. R. and V. J. demonstrated borderline performance IQs (about 2 SD below the mean) that were significantly below the level of their Kroll et al. 839

Table 2. Demographic and Neuropsychological Data for the Lesion Patients Group Right Occipital Lesion Callosotomy Participant J. C. F. N. N. A. C. B. J. W. D. R. V. J. Sex female male male female male female female Birth 1939 1938 1948 1974 1953 1944 1953 Trauma year perinatal 1988 1995 1988 1979 1983 1995 Verbal IQ 87 119 110 135 97 105 94 Performance IQ 81 131 106 122 95 72 73 Delayed Memory Index 96 125 75 120 80 83 79 For right occipital lobe patients, the trauma year is the year of the infarct or, in the case of C. B., the year of her equestrian accident. For callosotomy patients, the trauma year is the year of the surgery. respective verbal IQs. J. C. exhibited low-average verbal and performance IQ (about 1 SD below the mean). All three callosotomy patients performed below normal on the delayed memory index. The patients were tested over a period between 1994 and 2001 during the course of other behavioral testing. Unfortunately, some patients became unavailable during this period and it was not possible to test all of the Figure 3. Computer reconstructions of magnetic resonance brain scans for each of the four patients with right occipital lesions. Red areas represent site of lesion on transverse sections. The lateral view indicates the level and orientation of each section from the most ventral section (1) to the most dorsal (12). The bottom row represents a summing over all patients. 840 Journal of Cognitive Neuroscience Volume 15, Number 6

patients on all the tasks. For those patients tested in more than one experiment, a minimum of 2 weeks separated each pair of tests, with the exception of N. A. whose tests were separated by a single week. Student controls were obtained from undergraduate psychology courses at the University of California Davis and were given class credit for their participation. Twentyseven were tested in Experiment 1, 16 in Experiment 2, and 15 in each of Experiments 3 5. Older controls recruited from the local community were selected to be similar in age to the older patients and were paid for their participation. Sixteen older controls were tested in Experiment 1 (Yonelinas et al., 2001) and five older controls (males, aged 61 64) were tested in both Experiments 4 and 5. Three of these were tested first in Experiment 4 and the other two were tested first in Experiment 5. The two tests were separated by a minimum of 2 weeks. Note that the student controls are approximately the same age as C. B., who was herself a student at the time of testing. Design and Procedure Visual presentation was via a 15-in. computer monitor. The stimuli were white words centered on a black background and using DOS text-mode font. Experiment 1 Forty 8-letter words were presented visually to the participants, and 40 other 8-letter words were read to the participants by the examiner. These were presented in interleaved blocks of 20 items, and subjects were instructed to carefully attend to each word. Then, 160 visual word fragments (e.g., _c_sso_s), 80 studied words plus 80 new words, were presented, and subjects were given 10 sec to complete each fragment with the first word that came to mind. Materials were counterbalanced across subjects to equate the difficulty level of auditory and visual word lists. Experiment 2 Fifty words (5 primacy words, 40 critical words, and 5 recency words) were presented visually to the participants, and 50 other words (5 primacy words, 40 critical words, and 5 recency words) were read to the participants by the examiner. These were presented in interleaved blocks of 25 words, and subjects were instructed to carefully attend to each word. Then, 120 visual word stems (e.g., pea ), 80 of the studied words plus 40 of the new words, were presented, and subjects had 10 sec to complete each stem with the first word that came to mind. The 120 words, graciously supplied by Dr. Larry Squire, had been ranked by frequency, number of possible words beginning with its first three letters (stem), number of pretest subjects who chose that word to complete that stem by chance, and the length of word; then assigned to three lists of 40 each so that each list had roughly the same average and distribution on each of the above measures. The sequencing of study modalities and the assignment of words to lists (i.e., visual study, auditory study, and new lists) were counterbalanced across subjects to equate the difficulty of word lists. Experiment 3 Fifty-four words were presented visually to the participants, and 54 other words were read to the participants by the examiner. The subjects judged the number of syllables in each word. Half of the subjects received 27 visual words, then 54 auditory words, and then 27 visual words. The other subjects received the opposite ordering of the modalities. After finishing the study list, subjects were shown the 162 word fragments (e.g., g_e_d or _w_s_) of the 108 studied words, plus those of 54 new words. Subjects had 10 sec to complete each fragment. Lists were counterbalanced to equate the difficulty of fragments across modalities. Pilot studies, where 20 students were just given the fragments without a preceding study period, were used to rank the fragments for difficulty and words were assigned to lists in such a way as to balance the difficulty of the lists. Experiment 4 The same procedure was used as in Experiment 3, except that now the test stimuli were word stems (e.g., gre ). The words used were the same as in Experiment 3. The average Kucera Francis written frequency value for the words used in both experiments was 38.63. Experiment 5 The same procedure was used as in Experiment 4, except that now the words used were the only fiveletter words that would complete their stems. For example, the stem pau can only be completed by the word pause. In a few cases, alternatives existed but were of such low frequency that subjects were unlikely to know them. For example, the stem lig can be completed by both light and the extremely low frequency liger. The average Kucera Francis written frequency value for the words used in this experiment was 29.72. Acknowledgments We are very grateful to C. B., F. N., J. W., N. A., J. C., D. R., and V. J. for their full and enthusiastic participation in this study. Research in this article was supported by NIMH MH 59352-01 (A. Y.), NIDCD RO1 DC 04442 (K. B.), and NINDS grant NS21135 (R. T. K.). Kroll et al. 841

Reprint requests should be sent to Neal Kroll, Department of Psychology, University of California, Davis, CA 95616, USA, or via e-mail: neakroll@ucdavis.edu. Note 1. Although the right occipital lobe is present in callosotomy patients, the resection of the corpus callosum prevents communication between the hemispheres and verbal responses are generally considered to reflect processing within the isolated left hemisphere. Thus, facilitation observed in verbal responses of callosotomy patients should reflect only left hemisphere processes. The similarity in the pattern of results across experiments for the two groups of patients supports this hypothesis. REFERENCES Bassili, J. N., Smith, M. C., & MacLeod, C. M. (1989). Presentation modality and type of processing effects on priming in auditory and visual word-stem completion. Quarterly Journal of Experimental Psychology: Human Experimental Psychology, 41A, 439 453. Baynes, K., Eliassen, J. C., Lutsep, H. L., & Gazzaniga, M. S. (1998). Modular organization of cognitive systems masked by interhemispheric integration. Science, 280, 902 905. Baynes, K., Tramo, M. J., & Gazzaniga, M. S. (1992). Reading with a limited lexicon in the right hemisphere of a callosotomy patient. Neuropsychologia, 30, 187 200. Baynes, K., Wessinger, C. M., Fendrich, R., & Gazzaniga, M. S. (1995). The emergence of the capacity to name left visual field stimuli in a callosotomy patient: Implications for functional plasticity. Neuropsychologia, 33, 1225 1242. Blum, D., & Yonelinas, A. P. (2001). Transfer across modality in perceptual implicit memory. Psychonomic Bulletin and Review, 8, 147 154. Burgund, E. D., & Marsolek, C. J. (1997). Letter-case specific priming in the right cerebral hemisphere with a form-specific perceptual identification task. Brain and Cognition, 35, 239 258. Curran, T., Schachter, D. L., & Galluccio, L. (1999). Cross-modal priming and explicit memory in patients with verbal production deficits. Brain and Cognition, 39, 133 146. Elliott, R., & Dolan, R. J. (1998). Neural responses during preference and memory judgments for subliminally presented stimuli: A functional neuroimaging study. Journal of Neuroscience, 18, 4697 4704. Fleischman, D. A., Gabrieli, J. D. E., Reminger, S., Rinaldi, J., Morrell, F., & Wilson, R. (1995). Conceptual priming in perceptual identification for patients with Alzheimer s disease and a patient with right occipital lobectomy. Neuropsychology, 9, 187 197. Gabrieli, J. D. E. (1998). Cognitive neuroscience of human memory. Annual Review of Psychology, 49, 87 115. Gabrieli, J. D. E., Fleischman, D. A., Keane, M. M., Reminger, S. L., & Morrell, F. (1995). Double dissociation between memory systems underlying explicit and implicit memory in the human brain. Psychological Science, 6, 76 82. Gazzaniga, M. S., Eliassen, J. C., Nisenson, L., Wessinger, C. M., Fendrich, R., & Baynes, K. (1996). Collaboration between the hemispheres of a callosotomy patient: Emerging right hemisphere speech and the left hemisphere interpreter. Brain, 119, 1255 1262. Henson, R. N. A., Shallice, T., Gorno-Tempini, M. I., & Dolan, R. J. (2002). Face repetition effects in implicit and explicit memory tests as measured by fmri. Cerebral Cortex, 12, 178 186. Jacoby, L. L. (1983). Perceptual enhancement: Persistent effects of experience. Journal of Experimental Psychology: Learning, Memory, and Cognition, 9, 21 38. Keane, M. M., Gabrieli, J. D. E., Mapstone, H. C., Johnson, K. A., & Corkin, S. (1995). Double dissociation of memory capacities after bilateral occipital-lobe or medial temporal lobe lesions. Brain, 118, 1129 1148. Koivisto, M. (1995). On functional brain asymmetries in perceptual priming. Brain and Cognition, 29, 36 53. Kroll, N. E. A., Rocha, D. A., Yonelinas, A. P., Baynes, K., & Frederick, C. (2001). Form-specific visual priming in the left and right hemispheres. Brain and Cognition, 47, 564 569. Marsolek, C. J. (1999). Dissociable neural subsystems underlie abstract and specific object recognition. Psychological Science, 10, 111 118. Marsolek, C. J., & Hudson, T. E. (1999). Task and stimulus demands influence letter-case specific priming in the right cerebral hemisphere. Laterality, 4, 127 147. Marsolek, C. J., Kosslyn, S. M., & Squire, L. R. (1992). Form specific visual priming in the right cerebral hemisphere. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, 492 508. Roediger, H. L., III, & Blaxton, T. A. (1987). Effects of varying modality, surface features, and retention interval on priming in word fragment completion. Memory and Cognition, 15, 379 388. Roediger, H. L., III, & McDermott, K. B. (1993). Implicit memory in normal human subjects. Handbook of Neuropsychology, 8, 63 131. Rueckl, J. G., & Mathew, S. (1999). Implicit memory for phonological processes in visual stem completion. Memory and Cognition, 27, 1 11. Rybash, J. (1996). Memory aging research: Real life and laboratory relationships. Applied Cognitive Psychology, 10, 187 191. Schacter, D. L. (1992). Priming and multiple memory systems: Perceptual mechanisms of implicit memory. Journal of Cognitive Neuroscience, 4, 244 256. Schacter, D. L. (1994) In D. L. Schacter & E. Tulving (Eds.), Memory systems 1994 (pp. 233 268). Cambridge: MIT. Schacter, D. L., Chiu, C. Y. P., & Ochsner, K. N. (1993). Implicit memory A selective review. Annual Review of Neuroscience, 16. Schacter, D. L., & Tulving, E. (1994). In D. L. Schacter & E. Tulving (Eds.), Memory systems 1994 (pp. 1 38). Cambridge: MIT. Squire, L. R. (1994). In D. L. Schacter & E. Tulving (Eds.), Memory systems 1994 (pp. 203 232). Cambridge: MIT. Vaidya, C. J., Gabrieli, J. D. E., Verfaellie, M., Fleischman, D., & Askari, N. (1998). Font-specific priming following global amnesia and occipital lobe damage. Neuropsychology, 12, 183 192. Vuilleumier, P., Henson, R. N., Driver, J., & Dolan, R. J. (2002). Multiple levels of visual object constancy revealed by event-related fmri of repetition priming. Nature Neuroscience, 5, 491 499. Wessinger, C. M., Fendrich, R., & Gazzaniga, M. S. (1997). Islands of residual vision in hemianopic patients. Journal of Cognitive Neuroscience, 9, 203 221. Yonelinas, A. P., Kroll, N., Baynes, K., Dobbins, I. G., Frederick, C. M., Knight, R. T., & Gazzaniga, M. S. (2001). Visual implicit memory in the left hemisphere: Evidence from callosotomy and right occipital-lobe lesion patients. Psychological Science, 12, 293 298. 842 Journal of Cognitive Neuroscience Volume 15, Number 6