[1] 14w NEURAL CORRELATES OF VISUO-SPATIAL ATTENTION DURING AN ANTISACCADE TASK IN SCHIZOPHRENIA: AN ERP STUDY

[1] 103w The authors investigated the role of visuo-spatial attention in an antisaccade task using event-related potentials (ERPs) in schizophrenia patients compared to healthy controls. ERPs between 80-130 ms (P100) after stimulus onset showed differences between pro-and antisaccades only for controls and can be related to the suppression of irrelevant stimulus features. Between 150-180 ms (N100), a larger amplitude for anti-compared to prosaccades over centroparietal electrodes showed that processes of visuo-spatial attention seem to be engaged in performance of the antisaccade task. Left temporo-occipitally, this activity was only evident in schizophrenia patients, possibly reflecting additional neuronal recruitment in order to perform the antisaccade task successfully.

[1] 45w First of all, behavioral data from right and left hemifield stimulation were compared and revealed no differences in performance for prosaccades or antisaccades so that behavioral results from both hemifield stimulations were collapsed together to increase the statistical power of the data. In the prosaccade

[2] 184w Int J Neurosci 8.7 ± 1.8% of the trials with a mean latency of 194.9 ± 33.1 ms. Patients had 96.2 ± 5.1% correct prosaccades with a mean latency of 201.2 ± 24.7. As expected, there was no significant difference in prosaccades between patients and controls, neither for percentage of correctly performed saccades (p > .5, z = -l.29), nor for latency (p > .06, z = -1.87). However, schizophrenia patients had 61.2 ± 13.9% correct (mean latency of 331.5 ± 75.5 ms) and 33.1 ± 14.3% incorrect (mean latency 197.7 ± 38.5 ms) antisaccades, whereas controls performed correct in 79.1 ± 13.8% (mean latency 263.5 ± 61.1 ms) and incorrect in 18.9 ± 13.8% (mean latency 194.3 ± 45.4 ms). Thus, the patients yielded a statistically significantly lower percentage of correctly performed antisaccades (p < .002, z = -3.48) and a longer latency for correctly performed antisaccades (p < .005, z = -2.47). Moreover, after a wrongly performed antisaccade, patients (85.5 ± 15.7%) compared to controls (92.3 ± 2 1.7%) corrected reliably fewer wrongly performed antisaccades (p < .05, z = -2. 16).

[3] 26w The ERPs associated with the aforementioned conditions are shown in Figures 1 through 4. Note that attentional ERP effects due to hemifield stimulation Int J Neurosci

[4] 125w will cause a maximal amplitude on the contralateral hemisphere (Hillyard et al., 1995;Mangun, 1995). One needs to compare electrodes from left and right hemisphere when looking at the factor of hemifield stimulation because over one hemisphere one will automatically get a difference simply as a result of a larger amplitude for the contralateral stimulation and a smaller amplitude for the ipsilateral stimulation. So, for example, ERPs associated with a right hemifield stimulation over the left hemisphere should be compared with ERPs associated with a left hemifield stimulation over the right hemisphere. This can best be done for corresponding electrode pairs as mentioned earlier. Because multiple comparisons had to be conducted, the authors corrected for Type 1 errors by adjusting the alpha with the Bonferoni method.

[5] 84w A 2-way ANOVA for the P100 and N100 associated with correct prosaccades was performed for each of the following pairs C3/C4, P3/P4, O1/O2, and P7/ P8, employing the factor of stimulation (right/left-henceforth called r/l) and group (schizophrenia patients/controls) as between-subject factor as follows: lC4/rC3, rC4/lC3 and so forth. None of these ANOVAs revealed any significant difference in the P100 and N100, neither for the factor of stimulation nor for the factor group. Thus, right and left hemifield stimulation caused similar amplitudes in both groups.

[6] 109w Although the same logic as mentioned earlier holds true for analyses of ERPs to antisaccades, one has to bear in mind that if there is a stimulation on the right side, the subject will have to shift attention and look at the left side. Thus, ERP amplitudes for correctly performed antisaccades need to be compared contralaterally to where the subject looks (which is then in fact ipsilateral to the side of stimulation). Thus, one will need to compare electrodes over both hemispheres that are ipsilateral to the hemifield stimulation as, for example, lC3/rC4 (left hemifield stimulation for electrode C3 with right hemifield stimulation for electrode C4) and so forth.

[7] 32w Again, none of these 2-way ANOVAs employing the factor of stimulation (right/left-henceforth called r/l) and group (schizophrenia patients/controls) over corresponding electrode pairs as mentioned above gave rise to For personal use only.

[8] 23w significant differences for the P100 and N 100. Thus, also for antisaccades, right and left hemifield stimulation caused similar amplitudes in both groups.

[9] 71w Next, analyses were performed in a similar manner to detect differences between anti-and prosaccades across groups. To account for the effect of hemifield stimulation, the authors compared P100 and N100 to contralateral hemifield stimulation prosaccades (left hemifield stimulation for prosaccades called lp and right rp) with ipsilateral hemifield stimulation for antisaccades (left hemifield stimulation for antisaccades henceforth called la and right ra) between groups as follows: lpC4/raC4, rpC3/laC3 and so forth.

[10] 126w For the P100, overall there was no significant difference between ERPs to correctly performed pro-and antisaccades or between both groups. Only over the left occipital electrode O1, an interaction between the factors of saccade and group occurred for the comparison between the left hemifield stimulation for prosaccades and the right hemifield stimulation for antisaccades (F(1,38) = 9.82, p < .003). Subsidiary analyses for each group revealed that the interaction occurred because a main effect for the factor of saccade was only evident in controls (F(1,19) = 11.13, p < .001). Comparison of the means revealed that this effect reflected a larger amplitude of the P100 for Antisaccades (mean = l .04 µV) as compared to Prosaccades (mean = 0.31 µV) in controls (see also Figure 5).

[11] 119w Only for the N100, across both groups, there was a significant difference between pro-and antisaccades over central electrodes C3 (for left hemifield stimulation in prosaccades and right hemifield simulation in antisaccades F[1,38] = 31.48, p < .001) and C4 (for right hemifield stimulation in prosaccades and left hemifield stimulation in antisaccades F [1,38] = 19.18, p < .001). A similar effect occurred over the parietal electrodes, with left hemifield stimulation in prosaccades and right hemifield stimulation in antisaccades over P3 (F[1,38] = 33.49, p < .001) and vice versa over P4 (F[1,38] = l1.83, p < .003). As is evident from Table 2, these effects occurred, because across both groups, the N100 to prosaccades was smaller compared to antisaccades.

[12] 183w For the N100, a significant difference between both groups for pro-and antisaccades was found over left occipital-parietal electrodes for a left hemifield stimulation in prosaccades and a right hemifield stimulation in antisaccades (O1: F(1,38) = 10.31, p < .001 ; P7: F (l,38) = 14.47, p < .005). Subsequent analyses performed separately for each group revealed a reliable difference between pro-and antisaccades only for the patients (O1: F(1,19) = 9.24, p < .001; P7: F(1,19) = 13.78, p < .003). So the interaction reflected the fact that over the given electrodes only in schizophrenia patients, the N100 to prosaccades for the P100, the difference between both groups became significant over the electrode 01, as only in the control group, the P100 amplitude for the antisaccade larger compared to that for the prosaccade. (B) For the N100, the difference between both groups became significant over the electrode O1, as only in the patient group, the N100 amplitude for the antisaccade was larger compared to the P100 for the prosaccade. (C) Same effect as described in (B), now over electrode P7. For personal use only.

[13] 26w was smaller (O1: mean = -1.35 µV, P7: mean = -1.03 µV) as compared to antisaccades (O1: mean = -3.79 µV, P7: mean = 3.39 µV).

[1] 81w The present study investigated the neural correlates of visuo-spatial attention as measured by changes of P100 and N100 during an antisaccade task in schizophrenia patients compared to healthy controls. First, the present analyses both on P100 and N100 suggested that any given difference between ERPs to pro-and antisaccades in both groups could not be simply related to the fact that there was a difference between right and left hemifield stimulation. The P100 was less sensitive to the difference between pro-and antisaccades.

[2] 87w Only, over the left occipital electrode a difference between both groups occurred because a larger P100 for antisaccades as compared to prosaccades was only evident in the control group. In contrast for the N100 in both groups, there was a larger amplitude to antisaccades over central and parietal electrodes. Although a similar effect occurred right occipitally for both groups, a Only mean values from electrodes are shown where statistical analyses gave rise to a significant difference between amplitudes to pro-and antisaccades which did not differ across groups.

[3] 25w b rp = right hemifield stimulation prosaccades, lp = left hemifield stimulation prosaccades, ra = right hemifield stimulation antisaccades, la = left hemifield stimulation antisaccades.

[4] 17w Int J Neurosci prosaccades was only found in schizophrenia patients over left occipital and inferior parietal electrodes.

[5] 171w The present behavioral results fit within the now well-established finding that schizophrenia patients perform poorly on the antisaccade task but reveal no differences with controls when generating prosaccades (Clementz et al., 1994;Fukushima et al., 1994;McDowell & Clementz, 1997;Curtis et al., 2001). In particular, the authors were able to support the finding that the poor antisaccade performance in the patient group was tied to an increased number of reflexive errors, that is, performing a prosaccade toward the stimulus (Clementz et al., 1994;McDowell & Clementz, 1997;Curtis et al., 2001). Moreover, in the patient group, fewer corrected antisaccades after a wrongly performed antisaccade occurred and they were associated with a longer latency in comparison to controls. The lack of correcting antisaccades shows that schizophrenia patients' disability may not be simply related to a lack of fixation activity but because of a lack of generating voluntary saccades (Everling & Fischer, 1998). One reason may be that schizophrenia patients may have problems in shifting their visuo-spatial attention toward a new location as outlined in the introduction.

[6] 318w The authors' ERP results support previous studies, which have shown that schizophrenia patients are impaired on a visuo-spatial location task (Cadenhead et al., 1998;O'Donnell et al., 1996). It has been suggested that the P100 seems to be invoked by suppression of unattended stimuli (Hillyard et al., 1995;Coull, 1998 for review). As correct performance of the antisaccade demands a stronger suppression of the stimulus occurring on the contralateral hemifield (thereby inhibiting a reflexive saccade toward this stimulus), one would have expected that the P100 to antisaccades is larger than to prosaccades. However, such an effect was only evident for controls over the left occipital electrode. This result is difficult to interpret as, restricted to one electrode, it may simply be a problem of power. Future studies should therefore replicate this finding. Interestingly, recent research suggests that the N100 may be more reflecting discriminative processes than the P100 in visual attention (Hopf et al., 2002). Although the spatial resolution of the EEG is much poorer than its temporal resolution, one may speculate that the electrode overlying the left occipital cortex that gave rise to the significant difference is possibly covering primary visual cortex as well as extrastriate areas. Note that the sensory association areas have also been shown to be modulated by attention possibly as a result of attending to different stimulus attributes (Corbetta et al., 1990;Haxby et al., 1994;Heinze et al., 1994). Thus, although the functional significance of the larger P100 effect in controls remains to be further investi- gated, one explanation could therefore be that controls succeeded better in suppressing the irrelevant stimulus attributes in the antisaccade task. As a consequence they have a better performance in the antisaccade task compared to schizophrenia patients. Thus, the data are well in line with the more general finding that impairment in processing irrelevant information is one important aspect when defining the cognitive deficit in schizophrenia (Braff et al., 1999;Gray, 1998).

[7] 217w For the N100, there was a larger waveform to correctly performed antisaccades over central and parietal electrodes in both groups. Moreover, a larger N100 to anti-as compared to prosaccades was only found in schizophrenia patients but not in controls over left occipital and inferior parietal electrodes. Thus, the patients seem to have recruited additional brain regions on the left hemisphere in order to solve the antisaccade task. Although in the present study, the exact neural generators are difficult to determine, a recent study combined ERP and magnetoencephalographic (MEG) distributions in order to estimate the neural generators sources of an N100 (Hopf et al., 2002), which resembled the authors' temporo-occipital ERP effect. Hopf and colleagues (2002) found that a maximum of their N100 was located in inferior occipital and occipito-temporal cortex. Moreover, it has been suggested that during visuo-spatial attention activity occurs both in the parietal regions, which direct attentional signaling, and in ventral occipital regions, which are directed by attentional signals (Posner & Petersen, 1990). Indeed, recent brain imaging studies have identified a network covering frontal and posterior activations that seem to be engaged in both covert attention and sensorimotor systems that control the related overt behaviors (Nobre et al., 2000). Future studies combining EEG with MEG should therefore further elucidate the neural generators of the N100.

[8] 202w One has to bear in mind that amplitude increases of the N100 are elicited by the enhancement of attended stimuli, and attentional modulation of posterior (extrastriate) areas seem to be associated with processing of specific stimulus features (Corbetta et al., 1990;for review Coull, 1998). Thus, the authors supposed that schizophrenia patients are processing additional information as reflected by the N100, possibly because they are not able to inhibit irrelevant features. This argument is also supported by the larger P100 in controls over the electrode overlying the left occipital cortex that may reflect the suppression of irrelevant information as outlined before. As a consequence, and turning to the initial question, schizophrenia patients seem to make use of at least other posterior neural circuits, possibly not sufficient enough and thereby explaining the larger number of errors. Given that schizophrenia patients are known to have gray matter deficits in the occipital lobes (Mitelman et al., 2003), this may at least partially account for the functional impairment observed here. This hypothesis is further supported by a functional magnetic resonance imaging study of smooth pursuit eye movement deficits in schizophrenia patients that revealed subtle activity deficits in occipital regions compared to healthy controls (Tregellas et al., 2004).

[9] 176w An effect of medication cannot be ruled out as the subjects were taking atypical antipsychotics at the time of testing. It has been proposed that increased error rates and latencies of correct antisaccades in schizophrenia may be due to the increased dopaminergic neurotransmission in the mesofrontocortical dopamine pathways (Dursun et al., 1999). In fact. several studies point to a key role for the dopaminergic system in attentional processes requiring a greater degree of executive control (Brozoski et al., 1979;Berger et al., 1989;Roberts et al., 1994b). Given that the patients were all under a stable medication with atypical antipsychotics, this may have actually normalized their performance. However, they still revealed significant deficits in antisaccade performance compared to controls. These findings are in line with previous studies showing that antisaccade performance does not basically change under medication with neuroleptics (Mueller et al., 1999) or are not due to an acute effect of antipsychotic medication (Green & King, 1998). Further, the authors were able to demonstrate differences in neural activity between patients and controls even for correctly performed antisaccades.

[10] 80w To conclude, the data support the view that posteriorely-mediated processes of visuo-spatial attention are engaged in correct performance of the antisaccade task. During an antisaccade, processes that seem to reflect suppression of irrelevant stimulus features seem to be altered in schizophrenia patients compared to healthy controls. Moreover, patients exhibit additional neural activity, in particular occipitally, when trying to perform the antisaccades task correctly. Future research should aim at disentangling what the nature of this additional (possibly compensating) recruitment may be.

[1] 96w Twenty patients with schizophrenia and 20 healthy controls matched for age, gender, and years of education were investigated (see Table 1 for further Int J Neurosci izophrenia was diagnosed according to DSM-IV criteria by two independent raters (senior psychiatrists from the University of Cologne). Exclusion criteria for patients were any anticholinergic co-medication such as biperiden, comorbidity with relevant somatic illness, other psychiatric illness and abuse of alcohol or illegal drugs. Exclusion criteria for healthy controls were mental illness, relevant somatic illness, or substance abuse in personal history and psychiatric illness in 1st and 2nd degree relatives.

[2] 95w Informed consent was obtained by all participants according to the Declaration of Helsinki (1991) and was approved by the Medical Ethics Committee of the University of Cologne. Patients were on unchanged medication with atypical neuroleptics for a period of at least 4 weeks (8 patients were c Assesses psychopathology with respect to positive and negative symptoms. d Spatial working memory function was measured by a delayed response task (values are Euclidean distances between the indicated position of a point on a computer screen after a delay period of 15 s and the actual target position).

[3] 27w e Values are percentages of correct pattern identifications averaged across four conditions defined by two masking conditions (letters, random dot pattern) and two 151(42 ms, 104 ms).

[4] 9w f Wisconsin Card Sorting Test (percentage of perseverative errors).

[5] 10w g Continuous Performance Test-Identical Pairs version (signal detection parameter dprime).

[6] 51w receiving Olanzapine at a dose of 15 mg/day and 12 patients were treated with risperidone, 4 mg/day). In a neuropsychological test battery (Table 1), patients showed a significant impairment in sustained attention as well as visual attention but no differences in spatial working memory compared to the control group (Table 1).

[7] 95w During EEG recording, a white fixation point was presented for 500 ms in the middle of a computer screen on a black background. Next, a white rectangle (1.5 × 1.5 cm) was displayed randomly 14° to the right or left of the fixation point for a duration of 1000 ms. In the prosaccade task, subjects were instructed to look into the direction of the rectangle whereas in the antisaccade task they were required to look in the opposite direction of the stimulus. For each condition, 90 trials were presented, separated into three blocks of 30.

[8] 169w EEG was recorded with a 0.1 Hz high pass and a 70 Hz low pass filter (SYNAMPS R , Neuroscan; sampling rate: 500 Hz) from 28 scalp electrodes Positioned according to the extended International 10/20 system [Fpl/Fp2, Ft9/Ft10, F7/F8, F3/F4, Fz, Fc5/Fc6, T3/T4, C3/C4, Cz, Cp3/Cp4, Cpl/Cp2, Cpz, T5/T6, P3/P4, Pz, O1/O2 (Jasper, 1958)]. Electro-oculogram (EOG) was recorded bipolarly from Ft9/Ft10 and from Fp1/left infraorbital electrode. All channels were referenced to Cz. Off-line, data were algebraically adjusted to a common average reference, filtered with a 30 Hz low pass filter, and segmented into 1000 ms trials with a prestimulus baseline of 100 ms. EEG was corrected for intrusion of EOG artifact (Gratton et al., 1983). Trials in which base-to-peak EEG amplitude exceeded 70 µV or on which A/ D saturation occurred were rejected. In order to maintain an acceptable signal to noise ratio (Rugg et al., 1998;Tendolkar et al., 1998), only ERPs fonncd from 16 or more artifact free trials for any given response category were accepted for analysis.

[9] 141w Behavioral data were analyzed using a nonparametric test (Mann-Whitney-U). ERPs were formed to correctly performed pro-and antisaccades separately for saccades to either right or left hemifield stimulation in both groups. A comparison between ERPs to incorrectly performed pro-and antisaccades could not be calculated because of too few incorrect trials for the prosaccades in both groups. Although for incorrectly performed antisaccades, there were on average enough trials at least for the patient group (mean of 29 trials), these were largely contaminated with artifacts and had to be rejected so that no sufficient number of trials was left for the statistical analysis. In keeping with previous studies (Hillyard et al., 1995), peak amplitudes were measured with respect to a 100 ms prestimulus baseline for the time range between 80-130 ms (related to P100) and between 150-180 ms (related to N 100). As the

[10] 98w Int J Neurosci d to focus on posterior attentional effects, they a priori chose the parasagittal electrode pairs C3/C4, P3/P4, O1/O2, excluding the frontal electrodes (because of a stronger contamination due to eye artifacts). Additionally, the authors included left and right inferior temporal electrodes P7/ P8 to cover a broader pirate-occipital region, where with respect to previous studies (Hillyard et al., 1995), they would expect their effects of interest. The ERP data were subjected to an analysis of variance (ANOVA) as will be outlined in detail in the Results section and the alpha was corrected for multiple comparisons.

[1] 209w Schizophrenia patients' disability to inhibit prepotent responses has been intensively investigated in the so-called antisaccade task, whereby subjects upon appearance of a peripheral cue are required to make a saccade not to the target (prosaccade), but to the mirror image location (Hallett, 1978). An initial glance toward the peripheral cue constitutes an antisaccadic error. Although schizophrenia patients can perform simple refixation saccades as good as healthy controls, they exhibit a failure to carry out an initial suppression of a reflexive saccade toward the cue (Fukushima et al., 1990;Crawford et al., 1995;McDowell & Clementz, 1997). These increased antisaccade errors seem to belong to a set of abnormalities during volitional tasks that could best be characterized as a problem with saccadic inhibition (see Clementz, 1998 for review). For example, schizophrenia patients generate more very fast reaction time saccades (express saccades, Clementz et al., 1994), and they make more saccades during the delay period of these ocular motor tasks (McDowell et al., 2001). The requirement to inhibit a response in order to execute an alternative response has been allocated to functions of prefrontal cortex (Roberts et al., 1994a). Consequently schizophrenia patients' reduced ability to do so was taken as evidence for a prefrontal dysfunction (Weinberger & Berman, 1996; for review Selemon, 2001).

[2] 103w A second important element necessary for a correct saccade performance is the visual attention toward and maintenance of the spatial location of a correct response. Indeed, recent brain imaging studies have supported the engagement of visual spatial attention in oculomotor tasks (Buechel et al., 1998;Corbetta, 1998). They have identified a network covering frontal and posterior activations that seem to be engaged in both covert attention and sensorimotor systems that control the related overt behaviors (Nobre et al., 2000). Yet, to the authors' knowledge, no one investigated the role of a possible attentional deficit when trying to understand the antisaccade performance in schizophrenia patients.

[3] 147w Event-related potentials (ERPs) have been extremely inforniative in the study of visuo-spatial attention as they provide direct, high-temporal resolution signatures of neural activity related to a stimulus presentation Here, the P100 effect (80-130 ms post stimulus latency) is the earliest electrical sign of selective attention to be consistently observed in humans in visual tasks (Mangun, 1995), and to be followed by the N100 (150-180 ms). The principle finding is that the amplitude of the P100 and N100 are larger to an attended visual stimulus and thus give evidence for a selection of attention in various visuospatial tasks (Hillyard et al., 1995). However, the P100 and N100 waveforms seem to reflect slightly different attentional mechanisms. Whereas the enhancement of the P100 would appear to be invoked by suppression of unattended stimuli, amplitude increases of the N100 are elicited by the enhancement of attended stimuli (Hillyard et al., 1995).

[4] 207w So far, previous ERP studies on antisaccade performance in schizophrenia have only focused on differences in the contingent negative variation (CNV), which is considered to reflect prefrontal functioning and can be observed in manual and ocular motor responses (Klein et al., 2000). Thus, the present authors set out to investigate the neural processes of visuo-spatial attention related to pro-and antisaccade performance in schizophrenia patients as compared to controls. The authors were interested in particular in neural activity related to the stimulus presentation rather then to the onset of the saccade, that is, changes of the P100 and N100 between pro-and antisaccade as indicators of selective visuo-spatial attention across both groups. The authors focused on the neural activity related to correctly performed pro-and antisaccades to investigate whether patients would use other neural circuits, possibly not sufficient enough and thereby explaining the large number of errors. Further, the authors did not look for later changes as, for example, the P300 because these potentials may be contaminated through the onset of the saccades. Likewise, as in particular EEG over anterior regions is contaminated by eye-artifacts related to the saccades, the authors focused on ERP correlates of posterior attentional processes where eye artifacts are less likely to confound our EEG data.