Inefficient Visual Search Strategies in the First-Episode Schizophrenia Spectrum

Abstract

Background:

Knowledge is lacking regarding deficits in selective attention and their underlying biological mechanisms during early stages of schizophrenia. The present study examined the N2pc, a neurophysiological index of covert spatial attention, and its cortical sources at first psychotic episode in the schizophrenia spectrum (FESz).

Methods:

Neurophysiological responses measured simultaneously with magnetoencephalography (MEG) and electroencephalography (EEG) during pop-out and serial search tasks were compared between 32 FESz and 32 matched healthy controls (HC). Mean scalp-recorded N2pc was measured from a cluster of posterior-lateral EEG electrodes. Cortical source-resolved MEG activity contributing to the N2pc signal was derived for the intraparietal sulcus (IPS) and lateral occipital complex (LOC).

Results:

Group differences in EEG N2pc varied by task demand. FESz exhibited reduced N2pc amplitude during pop-out (p<.01), but not serial search (p=.11). Furthermore, group differences in N2pc-related MEG cortical activity varied by task demand and cortical region. Compared to HC, FESz exhibited greater IPS during serial search (p<.01).

Discussion:

Reductions in EEG N2pc amplitude indicate an impairment of visuo-spatial attention evident at an individual’s first psychotic episode, specifically during conditions emphasizing bottom-up processing. Examination of its cortical sources with MEG revealed that, compared to HC, FESz engaged parietal structures to a greater extent during the serial search condition. This pattern suggests a less efficient, more resource intensive strategy employed by FESz in response to a minimal demand on attention. The greater reliance on this controlled attentional network may negatively impact real-world functions with much greater complexity and attentional demands.

1. Introduction

Deficits in selective attention are a hallmark of schizophrenia, highlighted in early conceptualizations of the disorder (Bleuler 1911). Importantly, deficits in selective attention contribute significantly to disease burden and functional outcomes (Cloutier et al 2016; Green et al 2000). Most research examining selective attention deficits have been conducted in chronic schizophrenia despite the positive impact of early disease identification and intervention on outcome measures (Eack et al 2010; Hegelstad et al 2012). While previous studies have identified impairments on broad measures of attention early in disease course (Albus et al 2006; Zabala et al 2010), the isolation of selective attention processes and their neural concomitants is needed to disentangle the complex, multidimensional domain of attention. Thus, understanding the pathophysiology underlying deficits in selective attention early in disease course may provide novel information about primary disease processes and identify targets for therapeutic interventions.

Within the visual processing stream, the posterior contralateral N2 (N2pc) event-related potential, an electroencephalography (EEG) brainwave recorded from posterior scalp electrodes contralateral to attended items, is a well-validated index for the allocation of covert visuo-spatial attention. Although there remains some debate as to whether the N2pc represents processes of target selection (Eimer 1996; Li et al 2018) or distractor inhibition (Boehler et al 2011; Luck and Hillyard 1994), this potential is closely tied to the deployment of attentional resources in the service of target identification. An initial investigation in more chronic stages of schizophrenia revealed no difference in N2pc amplitude or latency to highly salient lateralized target stimuli (Luck et al 2006). However, in a follow-up study, individuals with schizophrenia exhibited a reduced N2pc response during conditions with increased cognitive demand (Verleger et al 2013). This finding is in-line with previous work demonstrating greater impairment on tasks requiring more top-down relative to bottom-up processing among individuals with schizophrenia (Fuller et al 2006; Luck and Gold 2008)].

Localization of the cortical activity contributing to the N2pc using magnetoencephalography (MEG) in healthy individuals has revealed two spatially and temporally distinct sources: an early, superior parietal source and a later, inferior occipito-temporal source (Hopf et al 2000). Given the role of the parietal cortex, specifically the intraparietal sulcus (IPS), in the frontoparietal attention network (Ptak 2012; Vossel et al 2014), this early source likely reflects the reorientation of attentional focus during visual search. This function of the IPS is also consistent with the traditional descriptions of the dorsal stream or “where” pathway of the visual system for locating targets in space. Interestingly, the parietal source of the N2pc is not always present, a phenomenon associated with the attentional demands of the specific visual search paradigm (Becke et al 2015; Hopf et al 2002). The more ventral source of the N2pc overlaps with portions of the lateral occipital complex (LOC), a region constituting the “what” pathway of the visual system traditionally associated with object recognition (Grill-Spector et al 2001; Malach et al 1995). As such, activity within these two regions provides information regarding the particular search strategy utilized to identify target items within a visual array based on task demands.

For the present study, the attentional processes indexed by N2pc and their cortical sources were compared between healthy controls (HC) and individuals with a schizophrenia spectrum diagnosis following their first psychotic episode (FESz). The visual target detection task comprised conditions that differentially emphasized parallel search strategies aided by stimulus saliency versus a controlled serial search of stimuli necessitating the recruitment of cognitive control networks. EEG and MEG were simultaneously recorded during task performance. EEG-measured N2pc was used to validate the paradigm. MEG-based N2pc source analysis was used to measure cortical activity in our a priori generator ROIs. We predicted FESz would exhibit impaired N2pc surface components and underlying cortical source activity relative to HC.

2. Methods and Materials

2.1. Participants

Eighty-three participants completed the study. Participants with insufficient segments of quality data (6 HC; 6 FESz) were excluded. The remaining 32 HC and 39 FESz were matched on age, gender, IQ, and parental socioeconomic status leading to final sample of 32 HC and 32 FESz. All participants completed the MATRICS Cognitive Consensus Battery (Table 1).

Table 1.

Demographics and assessment data (Mean +SD) by group

	HC (n=32)	FESz (n=32)	t/χ2	p
Age	21.8±4.2	22.7±5.7	−0.7	0.46
Female/Male	10/22	10/22	0.0	1.00
Education (years)	14.2±3.1	12.4±2.6	2.56	0.01
Race/Ethnicity			4.29	0.21
Caucasian	21	15
African American	7	14
Asian	3	3
Hispanic	1	0
WASIa	108.3±11.0	108.5±14.6	−0.05	0.96
MATRICS-Total	48.0±7.5	38.8±14.2	3.22	0.01
SESb	33.9±14.1	29.7±15.6	1.11	0.27
PSESc	51.7±14.0	46.7±15.6	1.3	0.20
SANSd	-	29.2±7.2
SAPSe	-	17.7±10.9
Unmedicated/Medicated		11/21
Medication (CPZf mg/day)	-	189.1±111

^aWechsler Abbreviated Scale of Intelligence;

^bSocioeconomic status;

^cParental socioeconomic status;

^dScale for the Assessment of Negative Symptoms;

^eScale for the Assessment of Positive Symptoms;

^fChlorpromazine equivalent dose

FESz were recruited from UPMC Western Psychiatric Hospital inpatient and outpatient services and participated within 2 months of their first clinical contact. They had less than 2 months of lifetime antipsychotic medication exposure. Eleven (34%) were unmedicated at time of testing. Diagnosis was based on the Structured Clinical Interview for DSM-IV (SCID-P). Twenty-two FESz were diagnosed with schizophrenia, 2 with schizoaffective disorder (depressed subtype), and 8 with psychotic disorder NOS. Symptoms were rated using the Scale for the Assessment of Negative Symptoms (SANS) and Scale for the Assessment of Positive Symptoms (SAPS) (Table 1).

Participants were screened for colorblindness using pseudoisochromatic plates and had at least nine years of schooling. None had a history of concussion, alcohol or drug addiction, or neurological comorbidity. Procedures were approved by the University of Pittsburgh IRB. Participants provided informed consent and were paid for participation.

2.2. Stimuli and Procedures

M/EEG were recorded simultaneously during task performance (Figure 1). Participants maintained gaze on a centrally presented fixation cross (1°, 500ms) that was replaced by a centrally presented color cue (0.65°, 500ms) indicating the color of the to-be presented target. The cue was followed by a target array of 6 eccentrically arranged annuli (each subtending 0.65° and positioned 2° from fixation, 500ms). Targets were presented either in the right (RVF) or left (LVF) visual field. To disentangle selective attention from motor response selection, preparation, and execution, there was a delay between the target array and the response array (500ms) at which time a small gap appeared on the right or left of each annulus. Participants indicated the side of the gap on the target annulus via button-press (2000ms). If the participant responded correctly on 0 or 1 of the previous 3 trials the gap was 0.65°, 0.33° if 2 of 3 correct responses, and 0.22° if 0 of 3.

Figure 1.

Target detection task depicting pop-out and serial search conditions. Participants were instructed to identify the target annulus based on the color cue then indicate via button-press the side on which the gap appears on the target.

The task consisted of two conditions run as separate blocks with order counterbalanced across participants. In the pop-out condition all non-target annuli were the same color and the target color remained consistent across trials. In contrast, all non-target annuli were different colors and the target color varied from trial-to-trial during the serial search condition. A common network of frontal, parietal, and occipital regions contribute to all forms of visual search (Ossandón et al 2012). There exists, however, some debate regarding the relative contributions of parietal and frontal sources depending on search type. While some studies argue an increased reliance on both regions during serial search (Corbetta and Shulman 2002; Kim et al 2012), others suggest divergent roles with greater reliance on parietal regions during pop-out as opposed to frontal cortices during serial search (Li et al 2010; Nobre 2003). Development of the task was based on previous work in non-human primates demonstrating the necessity of communications between cognitive control and sensory cortices for successful completion of the serial search condition (Rossi et al 2007). Each block consisted of 144 trials (72 RVF and 72 LVF targets). Participants were instructed to covertly shift attention towards the target to facilitate performance.

2.3. EEG/MEG Recordings and Pre-processing

M/EEG were recorded using a low impedance 60-electrode array based on the 10–10 system and a 306-channel MEG system (Eleckta Neuromag), with a sampling rate of 1000 Hz (online bandpass filter=0.1–330Hz). Head position was tracked continuously using 4 indicator coils placed on the EEG cap and a 3D-digitizer (ISOTRAK; Polhemus, Inc., Colchester, VT). EEG data were referenced to the left mastoid. The right mastoid was ground (a separate electrode was also used to record right mastoid activity). Bipolar leads placed above and below the left eye (VEOG) and lateral to the outer canthi (HEOG) monitored blinks and horizontal eye movements. ECG leads recorded cardiac activity. Structural MRIs were obtained for each participant. T1-weighted images were obtained using a Siemens TIM Trio 3 Tesla system with a multi-echo 3D MPRAGE sequence [TR/TE/TI=2530/1.74,3.6,5.46,7.32/1260ms, flip angle=7°, field of view (FOV) = 220×220mm, 1mm isotropic voxel size, 176 slices, GRAPPA acceleration factor = 2].

Neuromag MaxFilter software (http://imaging.mrc-cbu.cam.ac.uk/meg/Maxfilter_V2.2) was used to correct for head motion. Electromagnetic noise originating from outside the MEG helmet was separated from brain signal using the temporal extension of Signal Space Separation (Uusitalo et al 1997). EEGLAB Toolbox (Delorme and Makeig 2004) was used for further pre-processing. A high-pass filter (0.5Hz; 12dB/oct) was applied to remove DC offsets. Segments of noisy data and bad channels were removed. Adaptive mixture independent components analysis (AMICA) was performed to remove eye-blink and ECG components. Channels previous removed were interpolated.

2.4. EEG Sensor-Level Processing

Offline processing was conducted using BrainVision Analyzer2 (Brain Products GMBH). Data were re-referenced to the averaged mastoids and a low-pass filter (40Hz; 48dB/oct) was applied. Separate 600ms epochs were created for LVF and RVF targets, including a 100ms baseline. Following baseline correction, segments exceeding ±100µV in EEG or ±5pT in MEG sensors were rejected. Trials with a saccade to the target (voltage exceeding ±25µV during 200–500ms post-stimulus in HEOG channels) were rejected (Eimer 1996; Li et al 2018). Analysis of averaged HEOG waveforms from accepted trials revealed residual lateral eye movements of less than 0.1°. Participants had a minimum of 30 segments in each of the 4 possible segment categories (i.e., LVF pop-out, RVF pop-out, LVF serial search, and RVF serial search). Groups did not differ in the number of trials (p’s>.1).

N2pc was measured from a cluster of posterior EEG electrodes (i.e., PO3/4, PO7/8, O1/2). Difference waves were constructed by subtracting ipsilateral from contralateral target waveforms. N2pc was measured as the mean amplitude between 225–275ms (Figure 2).

Figure 2.

N2pc Difference waveforms for healthy control (HC) and first-episode psychosis (FESz) participants derived from pop-out and serial search conditions (A). Un-subtracted, grand average waveforms elicited in response to ipsilateral and contralateral targets for HC (B) and FESz (B) participants. The grey box indicates the 225 – 275ms N2pc mean amplitude time-window.

2.5. MEG Cortical Source Analysis

Dipole sources were constrained to the gray/white matter boundary, segmented from each participant’s MRI, which was tessellated into an icosohedron with 5mm spacing between vertices. The forward solution, modeled as a single sphere, and the noise covariance matrix calculated from baseline were used to create a linear inverse operator using an orientation constraint of 0.4 with depth weighting applied [Hämäläinen and Hari 2002; Lin et al 2004]. The current estimate at each location was normalized to pre-stimulus baseline variance to calculate the dSPM statistic that represents the signal-to-noise estimate at each vertex.

Cortical source activity estimates derived from RVF target trials were subtracted from trials with LVF targets to extract lateralized activity contributing to the N2pc. Because this subtraction resulted in the reverse of the N2pc difference wave (ipsilateral-minus-contralateral) for left hemisphere locations, the sign of these values was inverted prior to analysis. As with the N2pc component, dSPM within each ROI was averaged over the post-target 225–275ms time-window (Figure 3).

Figure 3.

(A) The intraparietal sulcus (IPS) and lateral occipital complex (LOC) ROIs selected for analysis of cortical source reconstructions. N2pc-related activity during pop-out (HC (B); FESz (C)) and serial search (HC (D); FESz (E)). Due to the subtraction used to generate N2pc-related activity (i.e., LVF – RVF), negative (blue) values on LH and positive (red) values on RH reflect greater activity. LH values were inverted in the quantification for purposes of comparison. dSPM values reflect averages over the 225–275ms time-window.

The IPS and LOC were selected as a priori ROIs based on previous literature (Hopf et al 2000; Hopf et al 2002). Despite the lack of a uniform anatomical definition of the LOC, there is a consensus that it includes the lateral occipital cortex (LO) in addition to ventral structures of the temporal-occipital boundary including the posterior fusiform gyrus and occipitotemporal sulcus (Grill-Spector and Weiner 2014; Sayres and Grill-Spector 2008). We therefore included LO and Brodmann’s area 37 in our ROI comprising the LOC. Given the role of the insula in saliency detection (Menon and Uddin 2010) and sensitivity of the superior temporal sulcus (STS) to task manipulations of search strategy (i.e. pop-out versus serial search) (Hayakawa et al 2003) as well as the strong signal observed from these structures in our data, both were included as additional ROIs in a supplementary analysis (see Supplementary Materials).

2.6. Data Analysis

Group demographics were compared using independent samples t-tests and chi-squared tests where appropriate. Behavioral data were subjected to a 2 (group: HC/FESz) × 2 (task: pop-out/serial search) repeated-measures ANOVA. A 2 (group) × 2 (task) × 3 (electrode site: PO3/4, PO7/8, and O1/2) × 2 (hemisphere: LH/RH) repeated-measures ANOVA was used to evaluate main effects and interactions on N2pc amplitude. A similar 2 (group) × 2 (task) × 2 (ROI: IPS and LOC) × 2 (hemisphere: LH/RH) repeated-measures ANOVA was used to evaluate main effects and interactions on cortical source activity. Given our primary interest in group differences, complex interactions involving group were first conducted within each electrode/ROI separately. When appropriate, subsequent interactions were explored within hemisphere, then within task condition ultimately allowing a direct comparison between groups. Exploratory correlations between neurophysiological measures, behavioral measures, and neurocognitive measures were assessed separately for HC and FESz. Effects of medication on outcome measures in FESz were explored using ANOVA with a between-group factor of medication status. Results were considered significant at p≤0.05.

3. Results

3.1. Task Performance

FESz performed less accurately (F_1,62=4.7; p=.03) and more slowly (F_1,62-4.5; p=.04) than HC across tasks (see Table 2). Participants performed less accurately on serial search compared to pop-out trials (F_1,62=12.4; p<.01). There was no effect of task on response time or interaction between group and task on either behavioral measure (p’s>.4).

Table 2.

Behavioral and electrophysiological data for participants during task performance (M±SD).

	Pop-Out		Serial Search
	HC	FESz	HC	FESz
Accuracy (%)	97.9±2.2	96.2±5.0	96.8±2.6	94.5±5.5
Response Time (ms)	543.0±115.4	600.8±125.3	541.7±84.9	601.5±129.0
N2pc Amplitude (µV)	−1.12±0.95	−0.53±0.67	−0.60±0.75	−0.35±0.42
IPSa Activity (dSPM)	0.33±0.46	0.27±0.34	−0.01±2.18	0.43±0.49
LOCb Activity (dSPM)	0.18±0.62	0.18±0.52	0.24±0.40	0.33±0.46

^aIntraparietal sulcus;

^blateral-occipital complex

3.2. Scalp-recorded N2pc Component

Grand average N2pc difference and unsubtracted waveforms averaged across EEG electrode sites are depicted in Figure 2. N2pc amplitudes are presented in Table 2 and their scalp topographies are depicted in Supplemental Figure 1. FESz exhibited reduced N2pc amplitudes across conditions (F_1,62=6.3; p=.02) and the serial search condition elicited smaller N2pc amplitudes compared to pop-out independent of group (F_1,62=21.4; p<.01). An interaction between group and task condition was also observed (F_1,62=5.0; p=.03). FESz exhibited smaller pop-out (t₆₂=−2.83; p<.01), but not serial search (p=.11) amplitudes compared to HC. Furthermore, while there was only a trend level task effect among FESz (p=.07), smaller N2pc amplitudes were observed during serial search compared to pop-out among HC (t₃₁=−4.40; p<.01). The main effects of electrode (p=.17) and hemisphere (p=.69) were not significant, nor were any additional interactions (p’s>.05).

3.3. Cortical Source Localized N2pc Activity

N2pc-related cortical activity is depicted in Figure 4 and quantification of dSPM values for each ROI is presented in Table 2. Unsubtracted cortical activity generated by LVF and RVF targets are depicted in Supplemental Figure 2. Of particular interest to the present investigation, an interaction between group, task condition, and ROI (F_1,62=6.0; p=.02) was observed. Given our primary interest on group differences, follow-up analyses were initially conducted separately within each ROI and subsequently, when present, interactions were explored between groups within each task condition. Analysis of IPS activity revealed an interaction between group and task (F_1,62=13.0; p<.01). FESz exhibited greater activity than HC during serial search (t₆₂=−3.26; p<.01), but not pop-out (p=.54). FESz also exhibited a trend-level increase in activity during serial search compared to pop-out (p=.07) while HC exhibited a significant decrease (t₃₁=−3.12; p<.01). No significant effects of task condition, group, or their interaction was present within the LOC (p’s>.1) ROIs.

In addition to the 3-way interaction, a series of lower-level interactions and main effects were also present. An interaction between group and task (F_1,62=8.7, p<.01) was observed. FESz exhibited greater activity during serial search compared to pop-out (t₃₁=−2.8; p<.01) across ROIs and larger overall activity in FESz compared to HC during serial search (t₆₂=−2.8; p<.01). No difference between tasks was present in HC and no group difference was observed during pop-out (p’s>.1). An interaction between task and ROI (F_1,62=5.8, p=.02) was also present, driven by a pattern of larger activity in IPS compared to LOC during pop-out that was reversed during serial search. However, no statistically significant differences between ROIs was present during either task (p’s>.05).

3.4. Correlations between neurophysiological measures and performance

Among HC, faster response times (pop-out: r=−.48, p<.01; serial search: r=−.39, p=.03) and higher serial search accuracy (r=.49; p<.01) on the visual search task were associated with better overall MATRICS performance. Correlations conducted between neurophysiological measures and performance revealed an association between increased IPS activity during serial search and worse MATRICS performance (r=−.39, p=.03). No other correlations between neurophysiologic measures and task or MATRICS performance achieved significance (p’s>.05).

Among FESz, faster response times in both tasks were also associated with better overall MATRICS performance (pop-out: r=−.59, p<.01; serial search: r=−.43, p=.01). No correlations were observed between neurophysiologic measures and task or MATRICS performance (p’s>.05).

3.5. Medication effects

There was no effect of medication status on any neurophysiological (p’s>.4) or behavioral (p’s>.2) measure.

4. Discussion

Although an impairment in selective attention is a core feature of schizophrenia, its presence and degree to which it contributes to morbidity in early stages of the disease remain understudied. In the present investigation, scalp-recorded EEG revealed an impairment in the N2pc among FESz during a visual search task emphasizing bottom-up processing. This marker of covert spatial attention, however, was better-preserved during trials requiring a more controlled search of the stimulus array. Localization of N2pc-related cortical activity using MEG revealed a difference between groups based on task demand. In contrast to HC, FESz recruited parietal structures in response to the attentional demands of serial search trials. The N2pc impairment during pop-out and greater reliance on parietal structures during serial search compared to HC highlights a dysfunction within the stimulus-driven attention network and greater inefficiency within the broader visual processing system present even at early stages of psychosis.

The reduced N2pc in FESz during pop-out reflects an inability to deploy attention towards highly salient objects in space, a fundamental process of the visual system largely independent of top-down modulation. A deficit predominantly affecting bottom-up attention is somewhat unexpected given the literature identifying greater impairment of controlled cognitive functions (Fuller et al 2006; Luck and Gold 2008). However, given the reliance of bottom-up processes on an intact sensory system, this deficit may be explained by dysfunctions in upstream visual processes in schizophrenia (Donohoe et al 2008; Foxe et al 2001). This theory is supported by a subsequent analysis of our MEG data that revealed an impaired early sensory-perceptual response ~100ms following target presentation in FESz localized to primary visual cortex (Sklar et al 2020).

In healthy adults, reductions in the N2pc during serial search result from increased temporal variability related to shifts in attentional focus required to locate the target (Dowdall et al 2012). The reduced signal-to-noise caused by this variability likely contributed to the task effect observed in the present study. However, reduced temporal variability of the mechanisms guiding attentional selection is unlikely to explain the relative preservation of the serial search N2pc in FESz compared to HC. An alternative explanation for the observed interaction is the recruitment of additional cognitive resources by FESz in response to increasing demands on attention. While analysis of cortical N2pc-related activity revealed a differential task effect similar to that observed at scalp sensors, it was driven by the significantly larger IPS activity in FESz relative to HC during serial search. Importantly, the lack of a parietal response in HC during serial search was not due to bilateral activity within this region negated during subtraction procedures (Supplemental Figure 2).

The robust IPS activity in FESz during serial search suggests a need to engage additional cognitive control structures to meet task demands and reflects a less efficient use of resources compared to HC. Interestingly, follow-up analyses investigating associations between early sensory-perceptual responses within primary visual cortex and the downstream N2pc revealed significant correlations in HC during both task conditions (pop-out: r=−.57, p<.01; serial search: r=−.45, p<.01). In contrast, similar correlations were only present during pop-out in FESz (pop-out: r=−.54, p<.01; serial search: r=−.18, p=.3). These results suggest that while HC relied largely on bottom-up processes for target selection across tasks, FESz employed an alternative, less efficient utilization of resources to meet serial search task demands. A similar, paradoxical increase in IPS activity has previously been reported in first-episode psychosis during an active visual pursuit task (Lencer et al 2011). As in the present study, the authors interpreted this finding as an attempt to compensate for an impairment in bottom-up perceptual processing identified during the passive viewing condition. Furthermore, the concept of efficiency has also been applied to working memory performance during chronic psychosis with hyperactivation within the dorsolateral prefrontal cortex during low-load conditions interpreted as an inefficiently functioning network (Johnson et al 2006; Kim et al 2010). As with this model of inefficient working memory load modulation, one would expect the compensatory activation observed in the present study to reach its limit in FESz at a lower attentional load compared to HC resulting in fewer cognitive resources available to address increases in task demands.

The scalp-recorded N2pc impairment observed in the present study appears to contradict results obtained by Luck et al (2006), which showed a preserved N2pc among a group of older individuals with schizophrenia. However, differences in the tasks used and illness severity between the two samples provide possible explanations for the divergent results. It is also important to note that a more recent study (Verleger et al 2013) observed reduced N2pc amplitudes in individuals with schizophrenia using a rapid serial visual presentation task to increase cognitive demand.

4.1. Limitations

Although both EEG and MEG data were collected, MEG data alone were used for localization of N2pc-related cortical sources. The poor signal quality of the MR images obtained made it difficult to extract reliable 3-shell boundary element models (BEM) for all participants. In contrast to MEG data, a more refined head model such as BEM is required to generate accurate forward solutions when using EEG data. While inclusion of data from both modalities may have improved our source solution, reliance of MEG data alone for cortical localization has been well validated and remains common practice in the literature.

The choice of electrodes and ROIs to include in analyses poses a challenge as there are no uniformly applied standards in the literature and their selection can have a significant impact on results. We chose to restrict our analyses to ROIs with evidence of their contribution to the lateralized N2pc response. However, because of significant N2pc-related activity within the insula and STS, these structures were included as additional ROIs in a post hoc analysis (see Supplementary Materials). Importantly, the inclusion of these ROIs did not alter our primary finding of a differential effect of task on N2pc-related cortical activity in our groups driven by large IPS activity in FESz during serial search (Supplemental Figure 3).

Differences between results obtained from neurophysiological and performance measures can be difficult to reconcile. While group differences in the N2pc and its related cortical activity varied based on task and brain region, similar interactions were observed for neither accuracy nor response times. Furthermore, if the observed pattern of cortical activity represents a successful compensatory mechanism employed by FESz, one might expect equivalent performance or reduced group differences on the serial search task. One possibility for this discrepancy is that neurophysiologic measures exhibit greater sensitivity to neurocognitive processes than gross measures of task performance.

4.2. Future Directions

A logical next step for this work is an examination of the oscillatory network dynamics underlying the observed impairments given the growing conceptualization of psychosis as a syndrome of aberrant connectivity. Manipulation of the target detection task to include a “hard” serial search condition with an increased number of distractors would also enhance our understanding of selective attention performance deficits and the limits of engaging compensatory search strategies.

4.3. Conclusions

The present study used M/EEG recordings to examine the N2pc and its cortical sources in FESz. Deficits in the N2pc reflect an impaired ability to deploy attentional resources towards highly salient stimuli and localization of its cortical activity revealed that FESz, unlike HC, were forced to rely on more inefficient search strategies in the face of increasing task demands. These findings highlight the dysfunction in processes of visual attention critical to daily functions present at what is currently the earliest stages of disease diagnosis. Given the presence of cognitive deficits during disease prodrome (Fusar-Poli et al 2012), deficits in the N2pc and its cortical sources provide information about the early primary pathology of schizophrenia. This work also has important implications for potential treatments of cognitive deficits given evidence for improved N2pc following transcranial direct current stimulation directed at parietal structures in individuals exhibiting poor performance on a visual short-term memory task (Tseng et al 2012).