Dissociation of the N2pc and Sustained Posterior Contralateral Negativity in a Choice Response Task

BRAIN RESEARCH 1215 (2008) 160– 172

available at www.sciencedirect.com

www.elsevier.com/locate/brainres

Research Report

Dissociation of the N2pc and sustained posterior contralateral negativity in a choice response task

Pierre Jolicœur⁎, Benoit Brisson, Nicolas Robitaille

Centre de Recherche en Neuropsychologie et Cognition, Département de Psychologie, Université de Montréal, Montréal, Québec, Canada

ARTICLE INFO ABSTRACT

Article history: The N2pc, a greater negativity at posterior electrodes on the side contralateral to an attended Accepted 15 March 2008 visual stimulus, usually between 180 and 280 ms, is thought to reflect the moment-to- Available online 7 April 2008 moment deployment of visual–spatial attention. In tasks that require the retention of information in visual short-term memory, there is also a sustained posterior contralateral Keywords: negativity (SPCN) that often begins about 300–400 ms after stimulus onset and that persists Spatial attention for the duration of the retention interval. A positive-going deflection at around 300 ms often N2pc separates the N2pc and the SPCN. An SPCN is also observed in tasks that are not defined as Visual Short-Term Memory memory tasks, but that presumably engage visual short-term memory as an intermediate SPCN processing buffer (e.g., in order to make a choice response to a briefly-presented visual PCN stimulus). The SPCN in memory tasks has been shown to increase in amplitude as the memory load is increased. We used this property of the SPCN to verify that the SPCN observed during the performance of a choice task with a response following each stimulus display is related to the SPCN observed in tasks that are structured as memory tasks. Using identical physical stimuli, we manipulated the hypothesized memory load across trial blocks by instructions either to encode only one stimulus or two stimuli. We observed an increase of the amplitude of the SPCN as memory load increased, with no concomitant increase in the amplitude of the N2pc that immediately preceded it. The results provide a clear dissociation between the N2pc (spatial attention, not affected by memory load) and the SPCN (visual short-term memory, sharply sensitive to memory load). © 2008 Elsevier B.V. All rights reserved.

1. Introduction immediate empirical goals have important theoretical conse- quences for the interpretation of the N2pc and the SPCN. Our We had two main immediate empirical goals and a more long- longer-term theoretical goal was to provide support for the view term theoretical goal. Firstly we wanted to dissociate functionally that the SPCN observed in tasks that are not designed as memory two event-related potential (ERP) components, the N2pc and the tasks nonetheless reflects neural activity specifically associated sustained posterior contralateral negativity (SPCN). Secondly we with the process of retaining visual information in visual short- wanted to demonstrate that the SPCN observed in a choice task term memory. Our work, in combination with previous work in exhibits many of the same properties as those observed in tasks the literature, makes a strong link between the initial spatially- that are defined as memory qua memory tasks. In turn, these specific processing of an attended lateralized visual target with

⁎ Corresponding author. Département de Psychologie, Université de Montréal, C.P. 6128 Succursale Centre-ville, Montréal, Québec, Canada H3C 3J7. Fax: +1 514 343 5787. E-mail address: [email protected] ( Jolicœur).

0006-8993/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.brainres.2008.03.059 BRAIN RESEARCH 1215 (2008) 160– 172 161 the N2pc. In contrast, activity supporting the retention of the et al., 2006a,b; Luck et al., 1993; Robitaille and Jolicoeur, 2006). In information in visual short-term memory is reflected by the these studies there was a likely functional role of visual short- SPCN. term memory because of the nature of the task, which required The N2pc (N2, posterior contralateral) is a greater negativity the report of a shape attribute of a visual target defined by at posterior electrodes on the side contralateral to the visual a colour difference relative to one or more distractors. We field of an attended visual stimulus relative to the voltage at believe that extracting shape information from the target in corresponding ipsilateral electrodes in the event-related poten- such situations requires passage through visual short-term tial (ERP) response. This lateralized component is usually ob- memory. Despite the plausibility of this argument, it is by no served between about 180 and 280 ms post-stimulus onset means certain, and it would be desirable to provide a more when the feature that defines the stimulus to be processed, the direct linkage between visual short-term memory and the target,isveryeasytolocate(thatis,whenitisdefinedbyaso- SPCN observed in tasks that were not specifically designed to called pop-out feature; Girelli and Luck, 1997). The N2pc has a test visual short-term memory. clearly posterior scalp distribution with a maximum often in A strong link between the SPCN and visual short-term the neighbourhood of electrode PO7 for targets in the right memory comes from experiments that were specifically de- visual field and PO8 for targets in the left visual field. Source signed to investigate visual short-term memory. Klaver et al. localization analyses of magnetoencephalographic recordings (1999) presented two ten-sided random polygons, one in the left suggest that the neural generators of the N2pc are in extra- visual field and one in the right visual field. A cue indicated striate visual cortex with a possible early parietal contribution whether the polygon on the left, the polygon on the right, or (Hopf et al., 2000, 2002). both, had to be remembered for a later (1500 ms) test of visual Luck and his colleagues proposed that the N2pc reflects a memory. A sustained posterior contralateral negativity was process of distractor suppression needed when visual distrac- observed during the encoding and retention interval when tors make it more difficult to process a visual target (Luck et al., subjects encoded either the left or the right polygon. They called 1997a,b; Luck and Hillyard, 1994). A key point in support of this this response CNSW, for contralateral negative slow wave. We hypothesis is that the size of the receptive field of neurons in the prefer SPCN because this acronym specifies better one aspect of visual system increases as processing progresses from earlier to the scalp distribution, namely the strongly posterior location of later visual areas. Consequently, neurons in later areas would the effect (in addition to the strong lateralization). often receive input from both the target and distractors. A Vogel and Machizawa (2004) studied visual short-term mechanism that would suppress the input of the distractors and memory using briefly-presented coloured squares, half in leave only input from the target would thus be desirable (Luck et left visual field and half in right visual field. Prior to their al., 1997a,b), and could be reflected in the N2pc. presentation, an arrow pointing left or right indicated which However, it is possible to observe large N2pc ERPs even when squares were to be encoded for a later test of visual memory. the target to be processed is the only visual stimulus in a given After a delay of 900 ms, a test display containing coloured visual field, accompanied by a single distractor in the opposite squares was presented and the task was to determine whether visual field (and sometimes a fixation point; e.g., Brisson and the test display was the same as, or different from, the original Jolicoeur, 2007a,b,d; Eimer, 1996; Jolicoeur et al., 2006a,b; display. As in the experiment of Klaver et al. (1999), an SPCN was Dell'Acqua et al., 2006; Robitaille and Jolicoeur, 2006). Eimer observed during the entire duration of the retention interval. In (1996) suggested that such results are more consistent with the later work Vogel et al. referred to this response as the CDA, for view that the N2pc reflects processes related to target proces- contralateral delay activity (McCollough et al., 2007). sing, per se, rather than to distractor suppression. Of course, it is not certain that the SPCN, CNSW, and/or CDA It will not be necessary for us to distinguish between these are the same component and reflect the same underlying neural two alternative interpretations of the underlying functional activity. However, we suspect that they are indeed one and the cause of the N2pc. For our purposes, it is sufficient to note that same, with variations in detail owing to differences in low-level all researchers agree that the N2pc reflects differential proces- stimulation, and higher-level task differences, which would sing of stimuli in one visual field relative to those in the opposite affect the scalp distribution, amplitude, and onset and offset visual field, and that such processing reflects the operation of latency of the component. To ease the exposition, we will refer visual–spatial attention (e.g., Eimer, 1996; Jolicoeur et al., 2006a, to all of these manifestations of visual short-term memory as b; Woodman and Luck, 2003). Indeed, proper experimental de- SPCN. Indeed, one of the main goals of the present work is to signs ensure that the N2pc can only arise as a result of dif- provide evidence that these measurements do reflect the same ferential allocation of visual–spatial attention to the target (e.g., underlying visual short-term memory mechanisms. Luck and Hillyard, 1994). The N2pc is sometimes referred to as Perhaps the most interesting result reported by Vogel and the posterior contralateral negativity (PCN) to indicate that the Machizawa (2004) was that the amplitude of the SPCN increased component is not related functionally to the typical N2 com- as the number of coloured squares in their memory array was ponent (e.g., Van der Lubbe et al., 2001; Wascher, 1995; Wascher increased, but only up to the estimated capacity of individual and Wauschkuhn, 1996; Wauschkuhn et al., 1998). subject's visual short-term memory. McCollough et al. (2007) The SPCN, or sustained posterior contralateral negativity, is provided further evidence in support of the notion that the often observed immediately after the N2pc in experiments that variation in the amplitude of the SPCN as a function of the require processing of, and/or memory for, information found number of representations held in visual short-term memory at the location of a lateralized visual target, in experiments could be taken as a hallmark of neural activity specifically requiring an initial visual search for the target (e.g., Brisson and related to maintenance in visual short-term memory. In Jolicoeur, 2007c; Dell'Acqua et al., 2006; Eimer, 1996; Jolicoeur particular, they showed that the increase in the amplitude of 162 BRAIN RESEARCH 1215 (2008) 160– 172 the SPCN was not due do an increase in the visual area side), and to vary how much was encoded (by the instruction to subtended by the visual stimuli (presentation area and set size encode only one digit, or both digits). were confounded in the experiments of Vogel and Machizawa, The task was to respond by pressing one of nine response 2004). keys, on the numeric keypad of a computer keyboard, depend- In dual-task experiments (Brisson and Jolicoeur, 2007d, ing on the identity of the encoded digit(s). When two digits had 2007c), an attenuation of the N2pc mean amplitude was ob- to be encoded, the subject made two responses in succession, served as task overlap increased between a speeded auditory one for each digit. task and a speeded visual task. In contrast, no such effect on the We recorded the EEG while subjects performed the task. The ultimate amplitude of the SPCN was observed. Rather, the onset logic of the experiment was straightforward. First, we expected latency of the SPCN was progressively delayed as task overlap to observe an N2pc because we assumed that the task would increased. These different patterns of N2pc and SPCN modula- require the deployment of visual–spatial attention to the digit(s) tions provided a first dissociation between the N2pc and SPCN on the side cued by the target colour. Given the brief stimulus components. presentation and the fact that digits can be encoded very rapidly In the present experiment we made use of the effect of (Sperling et al., 1971), there was no reason to expect an increase memory set size on the amplitude of the SPCN as an empirical in the magnitude of the N2pc as a function of the number tool to achieve two goals: 1) to demonstrate that the SPCN of digits to be encoded. If anything, we might expect a slight measured, incidentally, in a choice task, also varies in amplitude decrease in N2pc amplitude in encode-2 blocks relative to as memory set size is increased, just as in experiments spe- encode-1 blocks, because the yellow digit could be considered as cifically designed to engage visual short-term memory; and 2) to a proximal distractor in encode-1 blocks but as a target to be provide the complementary dissociation to the one observed in processed in encode-2 blocks. On the distractor-suppression the above mentioned dual-task studies, by showing that the hypothesis proposed by Luck and his colleagues, this difference N2pc does not increase in amplitude with memory set size in could lead to a larger N2pc in the encode-1 blocks because of the contrast to what is observed for the SPCN. need to suppress the yellow digit distractor. Fig. 1 shows the type of visual display that we used. In We hypothesized that the task would require the engage- every trial four digits were displayed for 100 ms, half in the left ment of visual short-term memory as an intermediate step visual field and half in the right visual field. Half of the digits between the initial encoding of the digit (reflected by N2pc) were yellow, one in each hemifield. One of the remaining and the processes of identification and response selection digits was red and the other one was green. Each subject was required to make a response based on digit identity (Brisson tested in two blocks of trials (counterbalanced for order across and Jolicoeur, 2007a; Robitaille and Jolicoeur, 2006). If this subjects). In one block the task was to encode just the red digit assumption is correct, then the results should also reveal a (for half of the subjects) or just the green digit (for the other clear SPCN. Furthermore, if this SPCN reflects the involve- half of the subjects). In the other block, both digits on the side ment of visual short-term memory, then two further predic- of the red digit (half of the subjects) or on the side of the green tions can be made. First, the amplitude of the SPCN should be digit (other half) had to be encoded. larger when the task required encoding and responding to This manipulation allowed us to keep the physical stimula- two digits relative to the SPCN observed when only one digit tion constant across all trials, but to vary which side was en- was required for the task. Furthermore, we would expect that coded (by the instruction to encode either on the red or green visual short-term memory would be engaged for a longer period of time in the encode-2 condition compared with the encode-1 condition. This prediction follows in a straightforward way from the supposition that subjects would only maintain information in visual short-term memory for as long as required to perform the task, and that their response times would be significantly longer in the encode-2 condition (because here they had to encode two stimuli and make two responses). Consequently, the SPCN should return to baseline sooner in the encode-1 condition than in the encode-2 condition. An alternative account of the SPCN is that it could reflect the continued deployment of visual–spatial attention at the location(s) of the target(s). That is, the SPCN could be a continuation of the N2pc. Given that the N2pc and SPCN are often Fig. 1 – Example of the type of visual display that was used in separated by a return of the contralateral minus ipsilateral every trial. Two digits were yellow (shown in uniform grey difference waveform towards zero, one would need to pos- here). One digit was green and the other digit was red (texture tulate another neural mechanism to account for this ap- was used in this figure to indicate the difference in colour; parent interruption of the N2pc. In the present experiment, a the actual displays had no texture). All colours were longer sustained N2pc would be predicted on the assumption equiluminant and digits were shown on a black background. that attention would continue to be deployed to the location Subjects encoded the red or green digit, in encode-1 trials, or formerly occupied by target digits when processing two digits the red and yellow, or the green and yellow, digits (on the red rather than one. However, it is not clear what functional or green side) in encode-2 trials. See text for further details. role would be served by maintaining spatial attention at a BRAIN RESEARCH 1215 (2008) 160– 172 163 location where it is known that no further relevant events will occur. If the SPCN is simply a continuation of the N2pc, 2. Results then the effect of processing load (encode-1 vs. encode-2) on N2pc amplitude should be the same throughout the duration 2.1. Behavior of the N2pc, including the postulated continued N2pc that we are calling SPCN. Overall, the accuracy was very high (95.7% of all the targets Although our aim was to manipulate only the memory load, presented were accurately reported). In the single target block, across encode-1 and encode-2 trials, keeping the physical dis- accuracy was 97.8%. It was slightly lower in the dual-target plays constant, the conditions could differ also in the number block (89.95%; F(1, 9)=34.2, MSE=.000902, p<.0003, when we of shifts of spatial attention required to encode 2 digits rather consider only the trials with both responses correct). The mean than only 1 digit. In this view, subjects would likely first encode response time was 942 ms in encode-1 trials and 1922 ms in the digit in the colour specified for selection of the relevant side encode-2 trials (RT of the second key press), F(1, 9)=40.0, (i.e., green or red). This would require a shift of attention the MSE=239952.54, p<.0016. These results show that subjects green (or red) item. Some time later, a second shift of attention performed the task with very high accuracy, and took longer might be required to encode the yellow digit on that side. This when there were two responses than when there was only one additional shift of attention could produce a second N2pc, such response, as expected. The mean response time of 942 ms for that there would now be two N2pc responses on encode-2 trials the encode-1 trials is reasonable given that the task required a (or a prolonged N2pc) but only one N2pc (or a shorter N2pc) on choice response with nine alternatives and given that the encode-1 trials. On this model, however, there should be no instructions did not stress response speed. increase in the amplitude of the posterior contralateral negativity on encode-2 trials versus that found during the first 2.2. Electrophysiology N2pc time window. There is no reason to expect that the amplitude of the second N2pc would be greater than the first. If The EEG was averaged starting 200 ms prior to the onset of the anything, component smear, produced by variations in the digits and ending 600 ms post-digits onset, and baseline onset of the second attention shift to the yellow digit, would corrected based on the 200 ms pre-digits period. To increase lead us to predict a reduction of the observed second N2pc the signal-to-noise ratio by including as many trials as possible, wave because different timings of the component across trials we did not exclude incorrect trials from the grand averages and would average to a lower peak amplitude. In past research with analyses reported below. If errors were caused by late processes a single lateralized target to be processed, the SPCN is often as taking place after visual short-term memory, such as response large as the initial N2pc. If the SPCN reflects activity in visual selection or response execution, then including incorrect trials short-term memory and if encoding 2 visual objects produces a should not affect N2pc or SPCN amplitude. On the other hand, if larger SPCN, the memory account of the SPCN predicts that the the errors occurred because attention was not deployed to the SPCN in encode-2 trials should be larger than in encode-1 trials. correct location, or because the targets were not encoded into If, the amplitude of the SPCN in encode-2 trials was larger than visual short-term memory, then including incorrect trials could the amplitude of the N2pc, that would provide evidence that only decrease the amplitude of the N2pc and/or SPCN. Given would be difficult to reconcile with a view that the SPCN does that accuracy was slightly higher in the encode-1 condition than not reflect visual short-term memory but rather an extended or in the encode-2 condition, including incorrect trials could only, additional N2pc. if anything, reduce the effect of our manipulation on the am- In summary, the experiment allowed us to distinguish plitude of the SPCN in the encode-2 condition (an effect that between two rather different interpretations of the SPCN. One goes in the opposite direction to what we hypothesized and account is that the SPCN reflects neural activity mediating found). In any case, in order to ensure that our conclusions did the maintenance of representations in visual short-term not hinge on whether error trials were included or excluded, all memory — SPCN as an index of visual short-term memory the analyses reported below were also performed on correct- (Brisson and Jolicoeur, 2007c; Dell'Acqua et al., 2006; Jolicoeur trials only (excluding error trials), and the patterns of results et al., 2006a,b; Klaver et al., 1999; Vogel and Machizawa, 2004). were essentially the same as those reported below (which The other account is that the SPCN reflects continued de- included error trials). ployment of visual–spatial attention — SPCN as extended The ipsilateral waveform (average of voltage at the left-sided N2pc. Generally speaking, the SPCN as extended N2pc account electrode for a left visual field target and voltage at the right- predicts that factors that affect the classic N2pc (in the 180– sided electrode for a right visual field target) and contralateral 280 ms window) should have the same effect on the later, waveform (average of left electrode for right visual field target extended N2pc. That is, it should be difficult to find functional and right electrode for left visual field target) time-locked to the dissociations between the earlier and later portions of the visual display for both processing load conditions at O1/O2, PO7/ N2pc because the later portion would simply be a continuation PO8, and P7/P8 electrode sites were computed separately. These of the earlier neural activity. In contrast, the SPCN as an index waveforms are shown in Fig. 2. The N2pc and SPCN were quan- of visual short-term memory account postulates that the N2pc tified following the subtraction of the ipsilateral waveforms and SPCN reflect distinct functional mechanisms. The N2pc from the contralateral waveforms. These subtraction wave- would reflect the initial deployment of visual–spatial attention forms are shown in Fig. 3. and the engagement of attention on the target. The SPCN As can be seen in Fig. 2, we obtained both of the anticipated would reflect neural activity associated with retention in vi- N2pc and SPCN responses. First, we found a contralateral / sual short-term memory. ipsilateral difference starting before 200 ms post digit onset, 164 BRAIN RESEARCH 1215 (2008) 160– 172

Fig. 2 – Grand average event-related potential (ERP) waveforms, ipsilateral and contralateral relative to the target digit(s), time-locked to the onset of the visual display at P7/P8, PO7/PO8, and O1/O2 electrode sites, for encode-1 (load 1) and encode-2 (load 2) conditions. All artefact-free trials, including correct and incorrect behavioral performance (see text for detail) were included in the grand-average ERPs.

corresponding to the N2pc. The curves then converged, track- We quantified the N2pc by computing the mean amplitude ing the end of the N2pc, which was followed by another con- of the contralateral minus ipsilateral difference waveform in a tralateral negativity with an onset after 300 ms, corresponding 180–280 ms post digit-onset time window, separately for each to the onset of the SPCN. Although the N2pc effect appears condition and each subject. We submitted the mean ampli- similar across processing load, the SPCN effect was clearly tude of the N2pc to a repeated-measures ANOVA with elec- larger and longer lasting in the encode-2 condition relative to trode site (O1/O2, PO7/PO8, and P7/P8) and processing load the encode-1 condition. These relationships are more easily (encode-1 vs. encode-2) as within-subjects factors. There was seen by examining the N2pc/SPCN difference waveforms dis- no main effect of load, F(1, 9)=.96, MSE=.47, p>.35, or of played in Fig. 3, which also displays the pooled response electrode position, F(2, 18)=1.68, MSE=.27, p>.21. And, there (average) for three posterior electrode pairs (P7/P8, PO7/PO8, was no interaction between these two factors, F(2, 18)=.28, and O1/O2), in the bottom panel. MSE=.16, p>.7. We also performed a t-test against zero on the BRAIN RESEARCH 1215 (2008) 160– 172 165

Fig. 3 – Grand average contralateral minus ipsilateral difference waves time-locked to the onset of the digit display at electrode sites P7/P8, PO7/PO8, and O1/O2, for the encode-1 (load 1) and encode-2 (load 2) conditions. The bottom panel shows the grand average waveform created by pooling over the three electrode pairs in the upper portion of the figure.

pooled response over the three electrode-pair sites for each These results corroborate what can be seen in Fig. 3, namely condition. The mean amplitude of the N2pc was −.98 µV in the that we found an N2pc in both conditions and that the load encode-1 condition, t(9)=−4.44, p<.002. The mean amplitude manipulation had little effect in the N2pc time window (and, if of the N2pc was −.80 µV in the encode-2 condition, t(9)=−3.77, anything, the N2pc was numerically smaller when load was p<.004. To ensure that the results did not depend on the de- increased, although not significantly so). tails of the chosen analysis window (e.g., some researchers The results were quite different for the SPCN. As can be might argue that the chosen N2pc measurement window of seen in Fig. 3, the amplitude of the SPCN was clearly larger 180–280 ms is too broad for a fast, phasic, ERP component such in the encode-2 condition than in the encode-1 condition. as the N2pc), we also reanalyzed all of the data using a shorter Furthermore, the SPCN lasted much longer in the encode-2 window of 210–250 ms. All of the patterns of results found condition than in the encode-1 condition. These observations with the broader window were replicated with the shorter were corroborated by results of within-subjects ANOVAs. We window. quantified the SPCN by computing the mean amplitude of 166 BRAIN RESEARCH 1215 (2008) 160– 172 the contralateral minus ipsilateral difference waveform in a dicated an amplitude that was not significantly different from 340–420 ms and in a 450–600 ms post digit-onset time window, zero), while the SPCN was still clearly present in the encode-2 separately for each condition and each subject. For the 340– condition, and greater than that in the encode-1 condition. 420 ms post digit-onset time window, the SPCN was larger in In a subsequent step, we lengthened our segments up to the encode-2 condition than in the encode-1 condition, F(1, 9)= 800 ms post-visual display to evaluate whether the SPCN am- 7.73, MSE=.93, p <.022. There was no main effect of electrode plitude difference was still present in the 600–800 ms post-visual position, F(2, 18)=1.69, MSE=.48, p >.21, nor a load×electrode display time window. In this analysis, we had to exclude four position interaction, F(2, 18)=.18, MSE=.31, p >.83. A pair of more subjects because of excessive residual eye movement in t-tests against zero on the pooled response showed that the 600–800 ms time window. Nevertheless, the analyses con- the SPCN was significant in both processing load conditions. firmed that the SPCN was still larger in the encode-2 condition The mean amplitude of the SPCN was −.81 µV in the encode- than in the encode-1 condition, F(1, 5)=7.02, MSE=1.52, p<.045. 1 condition, t(9)=−2.30, p <.047. The mean amplitude of the There was no main effect of electrode position, F(2, 10) =1.10, SPCN was −1.50 µV in the encode-2 condition, t(9)=−3.99, MSE=.63, p>.35, nor a load×electrode position interaction, F(2, p <.003. These results show that the SPCN was significantly 10)=.33, MSE=.23, p>.68. This analysis is important because it greater than zero in both conditions, and larger in the en- demonstrates a very long-lasting SPCN for the encode-2 code-2 condition than in the encode-1 condition. condition that is inconsistent with a simple account of the Interestingly, in the encode-2 condition, the SPCN was SPCN in terms of an extended N2pc or a second N2pc. On either significantly larger than the N2pc, F(1, 9)=7.26, MSE=.34, of these views, the SPCN was both too large and too long lasting p<.025, providing further evidence that the two components (given that the N2pc typically lasts about 100 ms). are functionally distinct. Indeed, if the SPCN and N2pc in- The dissociation of the N2pc and SPCN by processing load dexed the same process, than similar amplitudes would have was also confirmed by a significant processing load×time- been predicted. window interaction in subsequent analyses on the pooled Given that only one response was required in the encode-1 O1/O2, PO7/PO8, P7/P8 waveforms, with factors of load condition and two responses were required in the encode-2 (encode-1 vs. encode-2) and time window (180–280 ms vs. condition, it was not surprising to find a large difference in 450–600 ms), F(1, 9)=15.6, MSE=.45, p <.004. The load×time- reaction time between conditions. In order to verify that the window interaction was also significant when comparing the difference in SPCN amplitude in the 340–420 ms time window 180–280 ms and 340–420 ms time windows, F(1, 9)=8.97, was not caused by differences in reaction times, we conducted MSE =.21, p <.016. two additional analyses. We performed a median split based on We also computed scalp distribution of the electrical the mean reaction time of each participant in each condition, potentials during the N2pc time window and the two SPCN creating two groups (fast participants and slow participants) for time windows as well as current source density maps derived each condition. In the encode-1 condition, the fast group from the voltage maps. These distributions were calculated on had an SPCN of −.82 µV vs. an SPCN of −.8 µV for the slow group the basis of the contralateral minus ipsilateral difference wave- (t(8)=−.2, p>.98). The difference in reaction time across groups forms used to estimate the N2pc and SPCN, and are thus sym- was 398 ms but lead to a non-significant difference of .02 µV. For metric about the midline. We show the distribution maps for the the encode-2 condition, the fast group had an SPCN of −1.58 µV encode-2 condition, which produced the strongest SPCN. Fig. 4 vs. an SPCN of −1.42 µV for the slow group (t(8)=−.2, p>.84). The shows the scalp distributions of the average voltages in each difference in reaction time across group was 1044 ms, and lead time window, projected onto the side and back of the head. In all to a non-significant reduction of .16 µV in amplitude. Conse- cases, we observed a strong negativity posterior and contral- quently, it is unlikely that the .69 µV difference between the ateral to the memorized items. These maps are very similar to encode-1 and encode-2 conditions reported above was caused those reported by McCollough et al. (2007). Although it might by the 980 ms difference in reaction time. be tempting to interpret small differences among these maps, In the 450–600 ms post digit-onset time window, the SPCN perhaps the most interesting result is that they are quite sim- was larger in the encode-2 condition than in the encode-1 ilar. Fig. 5 shows the current source density transforms of condition, F(1, 9)=18.6, MSE=1.81, p<.002. There was no main these maps. Although neural generators of ERPs are not neces- effect of electrode position, F(2, 18)=2.21, MSE=.59, p>.14, nor sarily located directly beneath the locus of maximal voltage, the a load×electrode position interaction, F(2, 18)=1.38, MSE=.58, transformation of voltage maps into current source density in p>.28. The mean amplitude of the SPCN was −.17 µV in the the maps shown in Fig. 5 emphasizes nearby, superficial, radial encode-1 condition, t(9)=−.44, p>.66. The mean amplitude of neural generators (Pernier et al., 1988). Thus, the sharp foci in the the SPCN was −1.67 µV in the encode-2 condition, t(9)=−3.37, maps in Fig. 5 are consistent with a neural generator for the p<.008. These results show that the SPCN had returned to N2pc and SPCN components in posterior cortical areas, most baseline in the encode-1 condition (given that the t-test in- likely in extra-striate visual cortex.

Fig. 4 – Scalp distribution maps of the electrical potentials measured during the N2pc (180–280 ms, top panel), the early SPCN (340–420 ms, middle panel), and the latter portion of the SPCN (450–600 ms, bottom panel), for the encode-2 condition. A side projection is shown on the left, and a back projection on the right. The voltage scale ranges from −1.0 µV to .0 µV for the top panel, −1.7 µV to .0 µV for the middle panel, and −2.0 µV to .0 µV for the bottom panel, with a darker shade corresponding to a greater negativity. Different scales were used to allow direct comparisons of the distribution of relative potential intensities. BRAIN RESEARCH 1215 (2008) 160– 172 167 168 BRAIN RESEARCH 1215 (2008) 160– 172

tained in previous dual-task experiments (i.e., an attenuation 3. Discussion of the N2pc and a delay of the SPCN onset latency without any modulation of the SPCN ultimate amplitude; Brisson and The processing load manipulation — encoding and processing Jolicoeur, 2007c,d; Jolicoeur et al., 2006a,b), the results show one digit versus two digits — had a substantial effect on the SPCN and essentially no effect on the N2pc. The SPCN was larger when two digits were processed than when one digit was processed, but no such effect was observed on the N2pc. The increased amplitude of the SPCN was predicted if the SPCN reflects neural activity associated with retention in visual short-term memory, because previous work suggests that this effect is an electrophysiological hallmark of retention activity in visual short-term memory (McCollough et al., 2007; Vogel and Machizawa, 2004). In both the encode-1 and the encode-2 conditions, subjects needed to localize the red (or green) item to determine which side of the display should be processed. In the encode-2 condition, however, it is possible that processing required two shifts of attention, one for the red (or green) item, and another one for the yellow item. Consequently, there may have been two sequential attentional shifts in the encode-2 condition and only one in the encode-1 condition (to the red or green item). A second shift of attention could have resulted in a second posterior contralateral negativity of similar amplitude and duration as the first N2pc, with perhaps some component smearing due to a greater variability in the onset of the second N2pc (which would lengthen the component and reduce its amplitude). Could this sequential attention account explain the difference in amplitude between the encode-1 and encode- 2 conditions in the earlier portion of the SPCN (340–420 ms)? We believe that the results allow us to rule out this possibility because of the large amplitude difference between the SPCN waves in the encode-2 condition relative to the amplitude of the N2pc. Take, for example, the amplitude of the SPCN in the 450–600 ms time window. Here the SPCN in the encode-2 condition is clearly larger in amplitude relative to the amplitude of the N2pc for the encode-2 condition. There is no reason to expect that a second N2pc would be any larger than the first N2pc, and component smear during the averaging process when computing the ERP would preserve the total area of the component, but produce, if anything, a lower amplitude. In contrast, we found a substantial increase in SPCN amplitude in the encode-2 condition and this amplitude effect was quite long-lasting, extending into the 600–800 ms post-stimulus window. Although it is possible that a second shift of attention may have occurred in the encode-2 condition, this alone could not explain the significant increase in the amplitude of the SPCN relative to the measured N2pc. Consequently, the present experiment provides a dissociation of the N2pc and the SPCN in a task that was not specifically designed to test visual short-term memory. When this dissociation is combined with a complementary dissociation ob-

Fig. 5 – Current source density maps during the N2pc (180–280 ms, top panel), the early SPCN (340–420 ms, middle panel), and the latter portion of the SPCN (450–600 ms, bottom panel), for the encode-2 condition. Different scales were used to facilitate comparison of the shapes of the CSD distributions. BRAIN RESEARCH 1215 (2008) 160– 172 169 that we can demonstrate a double dissociation of the N2pc and N2pc and the SPCN. The similarity of the scalp distribution of SPCN, leading us to argue for two functionally distinct observed voltages, and current source density maps, across components. the N2pc time window and the SPCN time windows, could be A common issue when studying visual short-term memory taken as evidence against the proposed functional dissocia- is the possibility that the subjects could recode the information between the N2pc and SPCN. On the other hand, it is tion into a verbal code, thereby bypassing or supplementing possible that activation of the same neural tissue could have visual short-term memory. This could have been the case here different functional roles at different times during the in- given that we used easily-verbalized stimuli (digits) and we formation processing cycle, and it seems plausible that did not have a concurrent task designed to suppress encoding retinotopically mapped mechanisms that engage processing into a verbal system. Although it is possible that stimuli were on a visual target could also participate in a network that recoded verbally, it is not clear how this could have produced maintains this information in an active state (visual short- the patterns of results we observed. The double-subtraction term memory) following the initial encoding operations. In procedure used to compute the SPCN removes any effect that any case, relationships between scalp voltage distributions does not depend on the position of the stimuli in the visual and underlying neural generators can be very complex and display. If we suppose that stimuli were encoded into a verbal such distributions must be interpreted with caution (Urbach memory store (e.g., by storing the names of the digits), the and Kutas, 2002). location of the stimuli at encoding (even if also recoded into The present results are also important because they the verbal store) would no longer maintain the anatomical suggest that the SPCN can be taken as an indicator of the relationship linking visual input to representations in visual involvement of visual short-term memory in tasks that were cortex. Consequently, if verbal memory was implicated in the not specifically designed as memory tasks (e.g., Brisson and process, the electrophysiological activity related to it should Jolicoeur, 2007a,c; Robitaille and Jolicoeur, 2006). Despite the not be lateralized according to the side of the stimuli, and use of a task requiring a choice response to stimuli designed to therefore such a process would not be visible in the SPCN. We elicit an N2pc, we consistently observe both the N2pc and the cannot rule out that the stimuli could have been recoded in SPCN. This suggests to us that passage through visual short- verbal memory, but we can be confident that the observed term memory may be a critical intermediate processing step effects on the SPCN reflect the activity of visual short-term even when the task does not explicitly require a long period of memory rather than of some other, more abstract (verbal or retention, as is usually the case in experiments specifically semantic), memory store. designed to test visual short-term memory. An important aspect of the present experiment, which we Interestingly, the duration of the SPCN in speeded tasks have not emphasized up to now, is that the processing load appears to depend on the amount of time required to arrive at a effect observed on the SPCN could not have been due to dif- response (Robitaille and Jolicoeur, 2006). On the assumption that ferences in physical stimulation. The same physical displays the SPCN reflects neural activity required to maintain a were used for encode-1 and encode-2 trials — two digits were representation active in visual short-term memory, one would presented on each side of fixation in all trials, regardless of expect that the SPCN ERP would return to baseline earlier when processing load. This experimental design eliminated a con- visual short-term memory is required for a shorter period of found present in the experiments of Vogel and his colleagues, time, and return to baseline later when visual short-term given that their tests of the effects of memory load on the SPCN memory is required for a longer period of time. The present always had a larger number of visual stimuli in the visual results confirm this prediction. When subjects had to process a display when the memory load was higher (McCollough et al., single digit, the SPCN returned to baseline much earlier than 2007; Vogel and Machizawa, 2004). The present experiment when two digits had to be processed, as is clearly evident in Fig. 3. eliminated this confound by presenting identical physical dis- Leblanc et al. (2006) used the N2pc ERP component to plays for different memory loads. Given that the memory load determine whether contingent attentional capture entails a manipulation was achieved entirely by instructions, the shift of visual–spatial attention to the location of a visual differences we observed on the amplitude of the SPCN could stimulus believed to capture attention. The paradigm required not be due to differences in physical stimulation. The present subjects to process stimuli in a certain colour (e.g., red) at a results thus converge nicely with those of Vogel and his col- given location (e.g., at fixation), and to ignore stimuli leagues in showing that the amplitude of the SPCN depends on presented at other locations (Folk et al., 2002; Leblanc and what is encoded in visual short-term memory rather than on Jolicoeur, 2005). Note that the goal of the subjects was to resist differences in low-level visual stimulation. attentional capture and to filter out stimuli at locations other The results also rule out an interpretation of the present than the target location. Contingent capture was observed SPCN as an extended N2pc. On that account, the processing when a distractor shared the feature used to select targets at load manipulation should have had an effect on the entire the target location (e.g., a red distractor when the target was duration of the N2pc, including the early portion. If the SPCN is defined as a red item). A clear N2pc was observed in several just an ongoing N2pc, and if attention to two digits causes a experiments in which contingent capture was observed, dem- larger N2pc, then a larger N2pc should have been observed in onstrating that contingent capture was at least partly due to the usual time window for the N2pc (i.e., 180–280 ms), and the capture of visual–spatial attention. Most interestingly, in not only in the latter portion, as we observed. The complete none of these experiments was an SPCN found to follow the absence of a load effect in the usual N2pc time window and N2pc. We believe that the consistent absence of an SPCN in the strong effect observed in the SPCN time window provide these experiments is quite telling and suggests that subjects evidence for a functional and neural dissociation between the have some good control over what entered visual short-term 170 BRAIN RESEARCH 1215 (2008) 160– 172 memory. And, in contrast, they had less control over where that control further processing of the stimuli encoded into the spatial attention was deployed (involuntary capture). When visual short-term memory to operate on the stimulus long the goal of the subject was to process a red item at fixation, a enough to compute subsequent representations and/or re- red item in the periphery captured visual–spatial attention, sponses. In our view, passage into visual short-term memory which generated an N2pc. However, subjects succeeded in is required for discriminative control of responses (and more keeping this item out of visual short-term memory, and hence generally of all cognitive operations controlled on the basis of no SPCN followed the N2pc (despite the use of stimulus dis- visual discriminations of lateralized visual stimuli). The fact plays that were very similar to those in which both an N2pc that we observe prolonged posterior lateralization of electrical and an SPCN have been observed, when the task required brain activity that is lateralized as a function of the stimulus encoding the peripheral target). These results provide addi- position, rather than as a function of the motor response, shows tional converging evidence for a functional dissociation of the that the SPCN is closely tied to stimulus processing, rather than N2pc and the SPCN. A top-down control setting to filter out response processing (although stimulus processing in some distractors is insufficient to prevent capture of visual–spatial cases may be critical for response generation and control). attention by stimuli sharing the target-defining feature, lead- Of course, we cannot prove this interpretation on the basis of ing to an N2pc, but it is sufficient to prevent access to visual a single experiment, or even several experiments; instead we short-term memory, leading to the suppression of the SPCN. propose this theoretical interpretation as a framework within A number of other researchers have examined ERPs later- which we can interpret the results of a large number of ex- alized as a function of attention to a lateralized stimulus (N2pc) periments and guide further research. and as a function of a lateralized motor response such as the In summary, the existant literature suggested good reasons lateralized readiness potential (LRP) in the same experiments to believe that the N2pc and the SPCN reflect different under- (e.g., Wascher, 1995; Wascher and Wauschkuhn, 1996; Wausch- lying cognitive functions. However, this functional dissociation kuhn et al., 1998). These experiments generally produce an early had not been tested directly until now. In the present work we ERP lateralization that is negative contralateral to the stimulus performed a test designed to dissociate the N2pc and SPCN in a to be processed that appears to be the same component as the choice task and found that the number of items to be processed N2pc (called posterior contralateral negativity, PCN, by Wascher, had a large effect on the amplitude and duration of the SPCN, Wauschkuhn, and their colleagues), and one or two later con- with no significant effects on the N2pc. The results provide tralateral negativities. One of these may well correspond to the converging support for the view that the N2pc and the SPCN SPCN discussed in this article, although clear interpretations in reflect different neural functions. The N2pc likely reflects the terms of memory are difficult because the experiments involved shifting and the engagement of visual–spatial attention on a lateralized responses that could be correlated with stimulus lateralized visual target. The present results and the existant location (as in the Simon effect). We discuss here the inter- literature converge nicely to show that the SPCN reflects neural pretation of a component that Wauschkuhn et al. (1998) called activity specifically related to the maintenance of representa- the L-400, a contralateral negativity that was subsequent to tions in visual short-term memory, even in tasks that require an another one they called L-250. The L-250 appears to have been immediate choice response. Passage through visual short-term an N2pc and the L-400 may have been an SPCN. Indeed, we memory appears to be a necessary intermediate step in order tentatively re-interpret the L-400 component as related to the to make an overt response to information presented in visual involvement of visual short-term memory for the purpose of displays. controlling the response on the basis of a more detailed analysis of the stimulus. In contrast to this interpretation, Wauschkuhn et al. (1998) argued that their L-400 reflected the enhancement of 4. Experimental procedures an attentional shift subsequent to an initial shift of attention reflected in the L-250 (or N2pc). The experimental design used 4.1. Subjects by Wauschkuhn et al. (1998) involved several tasks, including a go/no-go condition (and several other conditions). The L-400 The subjects were neurologically normal undergraduate stu- was weak or absent for no-go trials, and generally weak or dents at the Université de Montréal who were paid for their absent in cases where the critical stimulus did not need to be voluntary participation. Fourteen subjects were tested, but re- interpreted further in order to determine the response required sults from 4 were excluded from the analyses (for reasons by the task. In our view, the L-400 was weak or absent in cases outlined in subsequent sections), leaving 10 subjects (4 men, for which there was no need to transfer the target stimulus into 6 women) who had a mean age of 21.4 years (19–30). All had visual short-term memory, which we argue is a necessary step normal or corrected-to-normal vision. for further analysis of the stimulus. Such transfer would be required particularly when the task required control over the 4.2. Stimuli response on the basis of discriminations of stimulus attributes, rather than the mere presence, or location, of the stimulus The stimuli were the digits 1–9 presented on a 17-inch cathode (although a weak, and perhaps short, SPCN may be observed if ray tube at a width of .6° of visual angle and a height of .9°. Four stimulus location is critical for the task). In other words, we are different digits, chosen at random without replacement from arguing that visual short-term memory is not merely a passive the set of 9, were displayed on each trial (see Fig. 1). Two of them short-term store used only for retention. Rather, passage into were on the horizontal midline (in line with the fixation symbol), visual short-term memory not only permits the maintenance of 2.5° to each side of fixation. Two were located 1.3° under the the stimulus information, but also allows other neural circuits others. Two were yellow (luminance=21.0 cd/m2, CIE, xy BRAIN RESEARCH 1215 (2008) 160– 172 171 coordinates: x=.483, y=.458), one on each side of fixation, one logram (VEOG), recorded as the voltage difference between was green (luminance=21.3 cd/m2, x=.272, y=.523) and the two electrodes placed above and below the left eye, was used other was red (luminance=21.8 cd/m2, x=.387, y=.274). The red to detect eye blinks. A bandpass filter of .01–40 Hz was applied and the green digits were never on the same side. They were and the EEG and EOG signals, digitized at 256 Hz, during offline presented on a near-black background (luminance=.98 cd/m2, analysis. x=.270, y=.300). The fixation point was grey (luminance= Trials with eye blinks (VEOG>80 µV), large horizontal eye 21.1 cd/m2, x=.274, y=.301). movements (HEOG>35 µV), and/or artefacts at electrode sites of interest (i.e., >80 µV at O1, O2, PO7, PO8, P7, and/or P8 electrode 4.3. Procedure sites) were rejected. Four participants with less than 50% of trials (i.e., 160 trials) remaining in at least one condition were excluded Each trial was initiated by pressing of the space bar, which from the analysis. The HEOG criterion was lowered to 20 µV for caused the disappearance of performance feedback from the one participant and to 17 µV for another participant so that previous trial and initiated the trial sequence. After a delay of residual averaged HEOG corresponding to eye movements 283 ms, the digits appeared and remained in view for 100 ms and towards the target would be less than 3.3 µV while retaining were followed by a blank screen. Subjects pressed response keys more than 50% of trials. For the ten subjects retained for detailed on the numeric keypad to report the identity of the digit(s) in analysis, an average of 81.6% of encode-1 trials and 80.6% of the display. Subjects had been trained prior to the experiment encode-2 trials remained after trial rejection. None of these proper to make these responses without moving their eyes away participants had residual eye movements that deviated more from the fixation point. After their response(s), the screen re- than .2° (i.e., average HEOG>3.3 µV) towards the target after mained blank for another period ranging from 1000 ms to artefact rejection (Lins et al., 1993). We also examined the aver- 1500 ms (randomly determined at run time). aged HEOG corresponding to eye movements towards the target There were two types of trials that were presented in two sets as a function of the encoding condition. There was a slightly of blocks. In encode-1 blocks, subjects encoded and responded larger, .39 µV, deflection in the encode-2 condition than in the only to the red digit (for half of the subjects) or to the green digit encode-1 condition, in a 340–420 ms time window, but this (for the other half). These trials required a single response. In the difference was not significant, t(9)=1.09, p>.3.However,ina450– encode-2 block, subjects were to respond to both digits in 600 ms time window, the encode-2 condition had a greater HEOG succession, to the two digits on the side of the red digit (for half deflection of 1.14 µV than in the encode-1 condition that was of the subjects) or on the side of the green digit (for the other significant, t(9)=3.2, p<.012. According to Lins et al. (1993),the half). Response time was measured from the onset of the digit HEOG propagation factor due to horizontal ocular movements at display to the last response required to complete the trial. P7 and P8 electrodes is 3±4%, and 1±3% at O1 and O2. Therefore, Feedback was presented at the fixation point (a plus sign if the1.14µVwouldtranslatetoamaximumof.08µVdifferenceat the digit(s) had been correctly identified, or a minus sign in posterior electrode sites, whichismuchlessthanthe1.5µVSPCN case of one or more errors). difference observed. Consequently, the small differences in There were 5 consecutive blocks of 64 experimental trials residual tendency to move the eyes towards the target stimuli for each type of trial, each preceded by a practice block of 18 could not explain the differences in SPCN across conditions. trials. The first 10 practice trials had a longer stimulus duration to enable the subjects to adjust to the task. There were 640 REFERENCES experimental trials in the experiment. In half of the trials in each block the red item was on the left (green on the right) and in half of the trials the red item was on the right (green on the Brisson, B., Jolicoeur, P., 2007a. Electrophysiological evidence of left). The position of the digits on each side (red/green on the central interference in the control of visuospatial attention. – midline, or yellow on the midline) was determined at random Psychon. Bull. Rev. 14, 126 132. Brisson, B., Jolicoeur, P., 2007b. The N2pc component and stimulus at run time. All testing was carried out in an electrically duration. NeuroReport 18, 1163–1166. shielded room under dim illumination using MEL stimulus Brisson, B., Jolicoeur, P., 2007c. A psychological refractory period in presentation software (Schneider, 1988). access to visual short-term memory and the deployment of visual–spatial attention: multitasking processing deficits revealed 4.4. Electrophysiological recordings by event-related potentials. Psychophysiology 44, 323–333. Brisson, B., Jolicoeur, P., 2007d. Cross-modal multitasking processing deficits prior to the central bottleneck revealed by The EEG was recorded from 64 active Ag/AgCl electrodes event-related potentials. Neuropsychologia 45, 3038–3053. (BioSemi ActiveTwo system) mounted on an elastic cap and Dell'Acqua, R., Sessa, P., Jolicoeur, P., Robitaille, N., 2006. Spatial referenced to the average of the left and right mastoids. Elec- attention freezes during the attentional blink. Psychophysiology trodes were placed according to the International 10/10 sys- 43, 394–400. tem at Fp1, Fpz, Fp2, AF7, AF3, AFz, AF4, AF8, F7, F5, F3, F1, Fz, Eimer, M., 1996. The N2pc component as an indicator of F2, F4, F6, F8, FT7, FC5, FC3, FC1, FCz, FC2, FC4, FC6, FT8, T7, C5, attentional selectivity. Electroencephalogr. Clin. Neurophysiol. – C3, C1, Cz, C2, C4, C6, T8, TP7, CP5, CP3, CP1, CPz, CP2, CP4, CP6, 99, 225 234. Folk, C.L., Leber, A.B., Egeth, H.E., 2002. Made you blink! Contingent TP8, P9, P7, P5, P3, P1, Pz, P2, P4, P6, P8, P10, PO7, PO3, POz, PO4, attentional capture produces a spatial blink. Percept. PO8, O1, Oz, O2, and Iz sites. The horizontal electrooculogram Psychophys. 64, 741–753. (HEOG), recorded as the voltage difference between elec- Girelli, M., Luck, S.J., 1997. Are the same attentional mechanisms trodes placed lateral to the external canthi, was used to used to detect visual search targets defined by color, measure horizontal eye movements. The vertical electroocu- orientation, and motion? J. Cogn. Neurosci. 9, 238–253. 172 BRAIN RESEARCH 1215 (2008) 160– 172

Hopf, J.-M., Luck, S.J., Girelli, M., Hagner, T., Mangun, G.R., Scheich, McCollough, A.W., Machizawa, M.G., Vogel, E.K., 2007. H., Heinze, H.-J., 2000. Neural sources of focused attention in Electrophysiological measures of maintaining representations visual search. Cereb. Cortex 10, 1233–1241. in visual working memory. Cortex 43, 77–94. Hopf,J.-M.,Boelmans,K.,Schoenfeld,A.M.,Heinze,H.-J.,Luck,S.J., Pernier, J., Perrin, F., Bertrand, O., 1988. Scalp current density 2002. How does attention attenuate target-distractor interference fields: concept and properties. Electroencephalogr. Clin. in vision? Evidence from magnetoencephalographic recordings. Neurophysiol. 69, 385–389. Cogn. Brain Res. 15, 17–29. Robitaille, N., Jolicoeur, P., 2006. Fundamental properties of the Jolicoeur, P., Sessa, P., Dell'Acqua, R., Robitaille, N., 2006a. On the N2pc as an index of spatial attention: effects of masking. Can. J. control of visual spatial attention: evidence from human Exp. Psychol. 60, 79–89. electrophysiology. Psychol. Res. 70, 414–424. Schneider, W., 1988. MicroExperimental Laboratory: an integrated Jolicoeur, P., Sessa, P., Dell'Acqua, R., Robitaille, N., 2006b. system for IBM-PC compatibles. Behav. Res. Meth. Instrum. Attentional control and capture in the attentional blink Comput. 20, 206–217. paradigm: evidence from human electrophysiology. Eur. J. Sperling, G., Budiansky, J., Spivak, J.G., Johnston, M.C., 1971. Cogn. Psychol. 18, 560–578. Extremely rapid visual search: the maximum rate of Klaver, P., Talsma, D., Wijers, A.A., Heinze, H.-J., Mulder, G., 1999. scanning letters for the presence of a numeral. Science 174, An event-related brain potential correlate of visual short-term 307–311. memory. NeuroReport 10, 2001–2005. Urbach, T.P., Kutas, M., 2002. The intractability of scaling scalp Leblanc, É., Jolicoeur, P., 2005. The timecourse of the contingent distributions to infer neuroelectric sources. Psychophysiology spatial blink. Can. J. Exp. Psychol. 59, 124–131. 39, 791–808. Leblanc, É., Prime, D., Jolicoeur, P., 2006. Contingent capture: a Van der Lubbe, R.H.J., Jaskowski, P., Wauschkuhn, Verleger, R., visuospatial effect? Evidence from electrophysiology. Visual 2001. Influence of time pressure in a simple response task, a Cognition 14, 112–115. choice-by-location task, and the Simon task. J. Psychophysiol. Lins, O.G., Picton, T.W., Berg, P., Scherg, M., 1993. Ocular artifacts in 15, 241–255. EEG and event-related potentials. I: Scalp topography. Brain Vogel, E.K., Machizawa, M.G., 2004. Neural activity predicts Topogr. 6, 51–63. individual differences in visual working memory capacity. Luck, S.J., Hillyard, S.A., 1994. Spatial filtering during visual search: Nature 428, 748–751. evidence from human electrophysiology. J. Exp. Psychol. Hum. Wascher, E., 1995. The timing of stimulus localization and the Percept. Perform. 20, 1000–1014. Simon effect: an ERP study. Exp. Brain Res. 163, 430–439. Luck, S.J., Fan, S., Hillyard, S.A., 1993. Attention-related Wascher, E., Wauschkuhn, B., 1996. The interaction of stimulus- and modulation of sensory-evoked brain activity in a visual search response-related processes measured by event-related task. J. Cogn. Neurosci. 5, 188–195. lateralisations of the EEG. Electroencephalogr. Clin. Luck, S.J., Chelazzi, L., Hillyard, S.A., Desimone, R., 1997a. Neurophysiol. 99, 149–162. Mechanisms of spatial selective attention in areas V1, V2, and Wauschkuhn, B., Verleger, R., Wascher, E., Klostermann, W., Burk, V4 of macaque visual cortex. J. Neurophysiol. 77, 24–42. M., Heide, W., Kömpf, D., 1998. Lateralized human cortical Luck, S.J., Girelli, M.T., McDermott, M.A., Ford, M.A., 1997b. activity for shifting visuospatial attention and initiating Bridging the gap between monkey neurophysiology saccades. J. Neuropsysiol. 80, 2900–2910. and human perception: an ambiguity resolution Woodman, G.F., Luck, S.J., 2003. Serial deployment of attention theory of visual selective attention. Cogn. Psychol. 33, during visual search. J. Exp. Psychol. Hum. Percept. Perform. 29, 64–87. 121–138.