Abstract

Previous studies have reported abnormal prefrontal and cingulate activity during attentional control processing in schizophrenia. However, it is not clear how variation in attentional control load modulates activity within these brain regions in this brain disorder. The aim of this study in schizophrenia is to investigate the impact of increasing levels of attentional control processing on prefrontal and cingulate activity. Blood oxygen level–dependent (BOLD) responses of 16 outpatients with schizophrenia were compared with those of 21 healthy subjects while performing a task eliciting increasing levels of attentional control during event-related functional magnetic resonance imaging at 3 T. Results showed reduced behavioral performance in patients at greater attentional control levels. Imaging data indicated greater prefrontal activity at intermediate attentional control levels in patients but greater prefrontal and cingulate responses at high attentional control demands in controls. The BOLD activity profile of these regions in controls increased linearly with increasing cognitive loads, whereas in patients, it was nonlinear. Correlation analysis consistently showed differential region and load-specific relationships between brain activity and behavior in the 2 groups. These results indicate that varying attentional control load is associated in schizophrenia with load- and region-specific modification of the relationship between behavior and brain activity, possibly suggesting earlier saturation of cognitive capacity.

Introduction

Attentional control allows the flexible allocation of attentional resources to relevant stimuli while suppressing those that are less relevant. This cognitive process provides a top-down bias for analysis and representation of relevant information in the face of concurrent and nonrelevant information (Desimone and Duncan 1995). In doing so, several subprocesses within attentional control are engaged, including allocation of attentional resources (i.e., a representation of the attentional demands of a task that can be used to bias processing in favor of task-relevant stimuli), and detection of conflict (i.e., the ability to detect conflicting information within the task) (MacDonald et al. 2000).

Electrophysiological studies in animal models as well as imaging studies in healthy humans have suggested that specific brain regions are crucially involved during attentional control, including prefrontal and parietal areas, as well as anterior and dorsal regions of the cingulate cortex (Desimone and Duncan 1995; Corbetta and Shulman 2002; Botvinick et al. 2004; Blasi et al. 2006). Moreover, in an earlier study, we have demonstrated a complex nonlinear relationship between activity in the anterior cingulate and dorsolateral prefrontal cortex with load of attentional subprocesses such as attentional allocation and conflict detection (Blasi et al. 2007).

Abnormal behavior and physiology in prefrontal and cingulate cortices during attentional processing in schizophrenia have been suggested by previous literature (Barch et al. 2001; Yucel et al. 2002; Weiss et al. 2003; Ford et al. 2004; Heckers et al. 2004; Honey et al. 2005; Kerns et al. 2005; MacDonald et al. 2005; Gur et al. 2007). In particular, some studies have found greater or reduced engagement of these brain regions during different cognitive subprocesses within the attentional domain (Barch et al. 2001; Yucel et al. 2002; Perlstein et al. 2003; Weiss et al. 2003; Heckers et al. 2004; Holmes et al. 2005; Honey et al. 2005; Kerns et al. 2005; Laurens et al. 2005; MacDonald et al. 2005; Arce et al. 2006; Gur et al. 2007). These studies have well-characterized regionally specific deficits in patients with schizophrenia during attentional processing. On the other hand, an open question pertains to the relationship between brain activity and individual variation in capacity to handle environmental inputs requiring attentional control and its putative saturation in schizophrenia.

Capacity limitations have been characterized in the working memory domain as being reflected by incrementally decreased behavioral performance in response to increasing working memory load (Just and Carpenter 1992). Previous studies investigating the relationship between saturation of working memory capacity and brain activity (Callicott et al. 2000) have demonstrated that before capacity saturation occurs, activity in prefrontal cortex of healthy subjects increases linearly with increasing working memory loads (Callicott et al. 1999). After capacity has been saturated, activity in prefrontal cortex drops, identifying a nonlinear relationship between working memory load and prefrontal activity suggesting an inverted-U curve (Callicott et al. 1999). Hypothetical pathophysiological models have also suggested that earlier capacity saturation may shift the nonlinear relationship to the left in schizophrenia (Callicott et al. 2000; Manoach 2003; Jansma et al. 2004).

In general, experimental paradigms involving increasing demands for cognitive processing may be useful to address the relationship between brain activity and variation in capacity (Callicott et al. 1999; Manoach 2003). In particular, parametric designs allow the study of capacity limitation in cognitive processing. Furthermore, they permit to analyze the physiological response during increasing cognitive loads, allowing in turn to compare brain activity associated with multiple levels of behavioral performance. Moreover, designs with these characteristics make it possible to compare the effect of increasing demands of cognitive load between groups, exploring differential physiological correlates of capacity limitation. In this context, we have recently developed a cognitive paradigm, the Variable Attentional Control (VAC) task, that requires increasing levels of attentional control processing (Blasi et al. 2005, 2007; Zhang et al. 2007). With this task, we have reported that activity in prefrontal and cingulate cortex increases linearly with increasing attentional control demands before capacity saturation occurs in healthy subjects (Blasi et al. 2007).

The aim of this study with the VAC task was to investigate in schizophrenia the relationship between varying load of attentional control and activity in prefrontal and cingulate cortex as measured with functional magnetic resonance imaging (fMRI). Based on previous studies showing abnormalities in prefrontal and cingulate activity during attentional processing and on evidence of earlier saturation of capacity in patients with schizophrenia during cognitive processing, we hypothesized that abnormalities in these brain regions might be differentially identified in prefrontal and cingulate cortex as a function of the load of attentional control required to perform the task.

Materials and Methods

Subjects

Twenty-five patients with schizophrenia on stable antipsychotic treatment for at least 1 month were recruited from the outpatient unit of the Psychiatric Clinic of the University of Bari and were enrolled in the study. Diagnosis was made according to the Diagnostic and Statistical Manual of Mental Disorder - fourth edition, as assessed with the Structured Clinical Interview for DSM-IV (SCID) (First et al. 1996). Symptoms were assessed on the day of scanning by a trained psychiatrist using the Positive and Negative Syndrome Scale (PANSS). Social-economic status (Hollingshead and Redlich 1958) and handedness (Oldfield 1971) were also measured. Five patients were excluded for excessive head movement during fMRI scanning, 4 performed at chance level at the task and were also excluded (vide infra). Therefore, 16 patients were included in all final analyses (Table 1). All subjects were treated with second-generation antipsychotics: 12 with olanzapine, 2 with risperidone, and 2 with clozapine. Two of these subjects received both olanzapine and haloperidol. The dose of treatment with antipsychotics expressed in chlorpromazine equivalents was 700.7 ± 353.2. Duration of illness was 9.0 ± 7.1 years. Patients included in the fMRI analyses did not differ from the excluded subjects in terms of demographics and symptoms (all P > 0.4).

Table 1

Demographics and behavioral data of the 2 groups

 Patients (n = 16; M = 13, F = 3)
 
Controls (n = 21; M = 14, F = 7)
 
 Mean ± SD Mean ± SD 
Age 32.5 ± 8.2 28.8 ± 6.5 
Handedness (Edinburgh Inventory) 0.6 ± 0.6 0.7 ± 0.4 
Parental socio-economical status (Hollingshead Scale) 22.9 ± 15.4 29.7 ± 19.8 
PANSS total score 69.6 ± 20.6  
Duration of illness (years) 9.0 ± 7.1  
Chlorpromazine equivalents (mg) 700.7 ± 353.2  
Behavioral data 
    Accuracy at HIGH 78.0 ± 13.5 93.5 ± 5.0 
    Accuracy at INT 83.1 ± 15.3 95 ± 3.6 
    Accuracy at LOW 92.3 ± 11.1 99.5 ± 1.1 
    Reaction time at HIGH 978.8 ± 224.5 1023.4 ± 153.3 
    Reaction time at INT 938.9 ± 196.6 961.5 ± 152.5 
    Reaction time at LOW 818.5 ± 127.2 815.1 ± 145.0 
 Patients (n = 16; M = 13, F = 3)
 
Controls (n = 21; M = 14, F = 7)
 
 Mean ± SD Mean ± SD 
Age 32.5 ± 8.2 28.8 ± 6.5 
Handedness (Edinburgh Inventory) 0.6 ± 0.6 0.7 ± 0.4 
Parental socio-economical status (Hollingshead Scale) 22.9 ± 15.4 29.7 ± 19.8 
PANSS total score 69.6 ± 20.6  
Duration of illness (years) 9.0 ± 7.1  
Chlorpromazine equivalents (mg) 700.7 ± 353.2  
Behavioral data 
    Accuracy at HIGH 78.0 ± 13.5 93.5 ± 5.0 
    Accuracy at INT 83.1 ± 15.3 95 ± 3.6 
    Accuracy at LOW 92.3 ± 11.1 99.5 ± 1.1 
    Reaction time at HIGH 978.8 ± 224.5 1023.4 ± 153.3 
    Reaction time at INT 938.9 ± 196.6 961.5 ± 152.5 
    Reaction time at LOW 818.5 ± 127.2 815.1 ± 145.0 

Note: See text for statistics. Only correct responses at the VAC task have been used for SPM analysis.

Twenty-one healthy subjects (Table 1), recruited by word of mouth, were also enrolled in the study after exclusion of any psychiatric diagnosis with the SCID interview. Healthy subjects and patients with schizophrenia included in this study had no history of drug or alcohol abuse, of head trauma with loss of consciousness, or of any significant medical condition. All subjects underwent fMRI while performing a task requiring increasing levels of attentional control (Blasi et al. 2005, 2007; Zhang et al. 2007).

The present experimental protocol was approved by the local institutional review board. After complete description of the study to the subjects, written informed consent was obtained.

Behavioral Task

The VAC task was used to elicit increasing demands of attentional control processing. This task was identical to that published in previous studies (Blasi et al. 2005, 2007; Zhang et al. 2007). Briefly, each stimulus was composed of arrows of 3 different sizes pointing either to the right or to the left; small arrows were embedded in medium-sized arrows that were in turn embedded in a large arrow. Subjects were instructed by a cue word (big, medium, or small) displayed above each stimulus to press a button corresponding to the direction of the large, medium, or small arrows (right or left). To increase the level of attentional control required, the direction of the arrows was congruent or incongruent across all 3 sizes. This resulted in the following conditions:

  • –Low level of attentional control: All 3 sizes of arrows were congruent in direction with each other. The cue was the word BIG.

  • –Intermediate level of attentional control. Two stimuli were used: The big arrow was incongruent in direction to the small and the medium arrows in both; the cue was BIG in one of them, SMALL in the other one.

  • –High level of attentional control. Two stimuli were used: the medium-sized arrows were incongruent in direction to the big and the small arrows in both; the cue was SMALL in one of them, MEDIUM in the other one.

  • –a simple bold arrow pointing either to the left or right was used as a sensorimotor control condition.

Although all controls performed above 50% in all conditions, most of the patients were not able to perform above chance to all 5 stimuli of the task. On the other hand, 21 patients performed above chance (>50%) to the stimulus eliciting low level of attentional control (LOW), to the stimulus in which the big arrow was incongruent in direction to the small and the medium arrows with SMALL as cue (intermediate level of attentional control, INT), and to the stimulus in which the medium-sized arrows were incongruent in direction to the big and the small arrows with SMALL as cue (high level of attentional control, HIGH) (Fig. 1). Because our objective was to evaluate neural responses to varying levels of attentional control, we focused our analysis on blood oxygen level–dependent (BOLD) responses elicited by these stimuli in the 21 patients who were able to perform above chance to these stimuli and in controls. Therefore, 21 healthy subjects and 21 patients with schizophrenia were initially entered in fMRI analyses. However, 5 of 21 subjects of the patients group were further excluded after imaging quality check (head movement during fMRI scan, vide infra). Thus, the sample of patients in the final analyses includes 16 subjects.

Figure 1.

Stimuli of the VAC task used to elicit low (A), intermediate (B), and high (C) load of attentional control processing.

Figure 1.

Stimuli of the VAC task used to elicit low (A), intermediate (B), and high (C) load of attentional control processing.

Subjects were instructed to respond to task stimuli with the right hand using a button box (right button for “right” response, left button for “left” response), and to press the response button as fast and accurately as possible. Furthermore, they were asked to move their thumb to a small plastic knob placed between buttons after each response. All subjects were trained on the task prior to the fMRI session to stabilize their performance. More specifically, just before fMRI scanning, subjects were instructed on task rules and performed the task outside the fMRI environment until their average behavioral performance did not grossly vary across trials in terms of behavioral accuracy (±10 percentage points across trials). Each stimulus was presented for 800 ms. The order of the stimuli was randomly distributed across the session (Friston et al. 1999). The total number of stimuli was 116: 25 HIGH, 34 INT, 57 LOW; duration of the task was 10 min 8 s. A fixation cross hair was presented during the interstimulus interval, which ranged from 2000 to 6000 ms. Stimuli in the fMRI setting were presented via a back-projection system, and responses were recorded through a fiber optic response box, which allowed the measurement of the accuracy and reaction time for each trial. We report behavioral performance relative to the task performed in the scanner during the fMRI experiment.

BOLD fMRI

BOLD fMRI was performed on a GE Signa 3T scanner (gradient-echo-planar-imaging sequence, time repetition/time echo = 2000/30; 26 interleaved slices, thickness = 4 mm, gap = 1 mm; voxel size 3.75 × 3.75 × 5 mm; scans = 300; flip angle = 90°; field of view = 24 cm; and matrix = 64 × 64) while subjects performed the VAC task. The first 4 scans were discarded to allow for signal saturation.

Data Analysis

Demographics and Behavioral Data

Analysis of variance (ANOVA) and χ2 was used to compare demographics and behavioral data. Tukey's test was used for post hoc analyses.

fMRI Data

Analysis was completed using the event-related module within Statistical Parametric Mapping (SPM) 2 (http://www.fil.ion.ucl.ac.uk/spm). Images for each subject were realigned, spatially normalized into the Montreal Neurological Institute (MNI) template (12-parameter affine model), and spatially smoothed (10-mm Gaussian filter). After realignment, data sets were also screened for small motion correction. In particular, subjects with head movement during scanning greater that 2.5 mm in translation and 2° in rotation were excluded from further investigation. Using these criteria, 5 of the 21 patients initially included based on their behavioral performance were not entered in following imaging analyses. Therefore, the final sample used in our study was composed of 21 healthy subjects and 16 patients with schizophrenia. fMRI responses were modeled using a canonical hemodynamic response function and temporally filtered using a high-pass filter of 128 Hz and an hrf-shape low-pass filter. Vectors were created for each condition using the timing of correct responses. Timing of presentation of all other stimuli was also convolved with Hemodynamic Response Function but not considered for further analysis. To account for putative differences in head movement between groups, residual movement was also modeled as regressor of no interest. A t statistic was then used to produce a statistical image for BOLD responses relative to brain processing of stimuli associated with correct responses for each level of attentional control (HIGH, INT, and LOW).

A random effects ANOVA was used to investigate the main effect of increasing level of attentional control, of diagnosis, as well as the effect of diagnosis on each level of attentional control. These analyses were constrained by a mask obtained by combining the HIGH, INT, and LOW group activation maps of both groups of subjects (P < 0.05). Brodmann's areas were assigned to activated clusters using the Talairach Daemon (http://ric.uthscsa.edu/projects/talairachdaemon.html) after converting the MNI coordinates of the local maxima in the activated clusters to Talairach coordinates (http://www.mrc-cbu.cam.ac.uk/Imaging/Common/mnispace.shtml).

Because of the strong a priori evidence of the involvement of the cingulate and of the prefrontal cortex in attentional control processing and our use of a rigorous random effects statistical model, we used for all the fMRI analyses a statistical threshold of P < 0.001, minimum cluster size (k) = 8, with further family wise error small volume correction at P < 0.05 applied on the activated clusters to control for potential type I errors. The volume of interest for such correction was obtained combining regions of interest built using 10-mm spheres centered around the coordinates in prefrontal and cingulate regions published in previous studies on attentional processes in healthy subjects and patients with schizophrenia (Duncan and Owen 2000; Kerns et al. 2005; Laurens et al. 2005; MacDonald et al. 2006).

To further explore load-dependent differences between patients and controls, ANOVAs outside of SPM were also used on BOLD signal change extracted using MarsBar (http://marsbar.sourceforge.net/) from clusters differentiating controls from patients in fMRI analyses. Furthermore, polynomial and linear regressions were performed to test for putative nonlinear or linear relationships between attentional control loads and signal changes from these clusters (Yacubian et al. 2007). Finally, to investigate the relationship between brain activity and behavior during increasing demands for attentional control processing, Pearson's correlation analysis was then performed between signal changes and both accuracy and reaction time during the VAC task.

Results

Demographics and Behavioral Data

Patients did not differ from controls for age, gender, socioeconomic status, and handedness (all P > 0.1). Repeated measures ANOVA on accuracy during the VAC task showed a main effect of increasing level of attentional control (F2,70 = 22.3; P < 0.0001), of diagnosis (F1,35 = 21.9; P < 0.0001), and an interaction between these 2 factors (F2,70 = 3.6; P = 0.03). Post hoc analysis revealed significant higher number of errors for patients at the higher (P = 0.01) level of attentional control, and a strong statistical trend in a similar direction at the intermediate level (P = 0.08) (Table 1). No statistical difference between groups was found at the lower level of attentional control (P > 0.6) (Table 1). Repeated measures ANOVA on reaction time data of correct responses showed a main effect of increasing level of attentional control (F2,70 = 83.7; P < 0.0001), no effect of diagnosis (F1,35 = 21.9; P = 0.2), and no interaction (F2,70 = 1.0; P = 0.4). Post hoc analysis showed statistically significant differences between reaction time at the higher and at the intermediate level of attentional control (P = 0.007), as well as between those at intermediate and the lower levels (P = 0.0001) (Table 1).

fMRI Data

Consistent with previous studies (Blasi et al. 2005, 2007), ANOVA showed a main effect of attentional control load in bilateral dorsolateral and medial prefrontal areas, including the anterior cingulate. Furthermore, there was an effect of diagnosis in regions of prefrontal and cingulate cortices. In particular, healthy subjects had greater activity in clusters of the left lateral prefrontal cortex (BA6/8), as well as in the cingulate cortex (BA24) when compared with patients. On the other hand, the inverse contrast showed that patients had greater activity in left dorsolateral prefrontal cortex (BA9) relative to the control group (Table 2).

Table 2

Local maxima of brain regions showing differential activity between patients (n = 16) and controls (n = 21) during the VAC task

Brain region BA Talairach coordinates
 
k Z 
x y z 
Effect of diagnosis: patients > controls 
    Left inferior frontal gyrus −45 31 3.67 
Effect of diagnosis: controls > patients 
    Left middle frontal gyrus 6/8 −38 11 55 15 4.21 
    Left anterior cingulate 24 −1 16 20 12 3.79 
Effect of diagnosis during HIGH: controls > patients 
    Left anterior cingulate 24 −4 16 20 13 4.76 
    Right middle frontal gyrus 6/8 34 10 48 18 3.69 
    Left middle frontal gyrus 6/8 −38 11 55 11 3.61 
Effect of diagnosis during INT: patients > controls 
    Left inferior frontal gyrus −45 27 14 3.61 
Brain region BA Talairach coordinates
 
k Z 
x y z 
Effect of diagnosis: patients > controls 
    Left inferior frontal gyrus −45 31 3.67 
Effect of diagnosis: controls > patients 
    Left middle frontal gyrus 6/8 −38 11 55 15 4.21 
    Left anterior cingulate 24 −1 16 20 12 3.79 
Effect of diagnosis during HIGH: controls > patients 
    Left anterior cingulate 24 −4 16 20 13 4.76 
    Right middle frontal gyrus 6/8 34 10 48 18 3.69 
    Left middle frontal gyrus 6/8 −38 11 55 11 3.61 
Effect of diagnosis during INT: patients > controls 
    Left inferior frontal gyrus −45 27 14 3.61 

Further analysis revealed that, at the high level of attentional control, healthy subjects had greater activity bilaterally in lateral prefrontal cortex (BA6/8) and in anterior cingulate (BA24). No brain regions crossed the statistical threshold on the inverse comparison (Fig. 2, Table 2). On the other hand, at the intermediate level of attentional control, patients with schizophrenia showed greater activity in left BA9. The inverse contrast did not show any significant difference (Fig. 2, Table 2). There was no significant difference between the 2 groups at the lower level of attentional control.

Figure 2.

Upper row: (A) Coronal and sagittal sections showing prefrontal (bilateral BA6/8) and cingulate regions with greater activity in controls (n = 21) than in patients (n = 16) during high loads of attentional control processing. (B) Coronal section showing the prefrontal region (BA9) with greater activity in patients than in controls during intermediate loads of attentional control processing. Lower row: Plots of activity in patients and controls in prefrontal (central plot: x −45, y 5, z 27—BA9; right plot: x 34, y 10, z 48—BA6/8) and cingulate (left plot: x −4, y 16, z 20) regions differentially activated during different loads of attentional control processing elicited by the VAC task, as shown in Figure 2, upper row. A linear increase of brain responses is present in each region in controls. Consistent with significant polynomial regressions in patients, relationships between brain responses and attentional loads fitted nonlinear inverted-U shaped curves. See text for statistics.

Figure 2.

Upper row: (A) Coronal and sagittal sections showing prefrontal (bilateral BA6/8) and cingulate regions with greater activity in controls (n = 21) than in patients (n = 16) during high loads of attentional control processing. (B) Coronal section showing the prefrontal region (BA9) with greater activity in patients than in controls during intermediate loads of attentional control processing. Lower row: Plots of activity in patients and controls in prefrontal (central plot: x −45, y 5, z 27—BA9; right plot: x 34, y 10, z 48—BA6/8) and cingulate (left plot: x −4, y 16, z 20) regions differentially activated during different loads of attentional control processing elicited by the VAC task, as shown in Figure 2, upper row. A linear increase of brain responses is present in each region in controls. Consistent with significant polynomial regressions in patients, relationships between brain responses and attentional loads fitted nonlinear inverted-U shaped curves. See text for statistics.

Statistical analyses on signal change extracted from clusters differentially activated between patients with schizophrenia and healthy subjects (BA6/8—x 34, y 10, z 48; BA24—x −4, y 16, z 20; BA9—x −45, y 5, z 27) were also performed (Fig. 2). In BA24, there was an effect of diagnosis (F1,35 = 6.80; P = 0.01), no effect of load (F2,70 = 1.71; P = 0.18), and an interaction between load and diagnosis (F2,70 = 5.69; P = 0.005). Post hoc analysis indicated greater signal change in controls relative to patients at the higher attentional demand (P = 0.0003). In BA6/8, there was an effect of diagnosis (F1,35 = 7.76; P = 0.009), of load (F2,70 = 12.79; P = 0.00002), and a load by diagnosis interaction (F2,70 = 6.37; P = 0.003). Post hoc analysis indicated that controls had greater BOLD signal than patients at the high attentional demand (P = 0.007). In BA9, there was an effect of diagnosis (F1,35 = 5.58; P = 0.01), of load (F2,70 = 9.78; P = 0.0002), and an interaction between load and diagnosis (F2,70 = 2.96; P = 0.05). Post hoc analysis indicated that patients had greater signal change relative to controls at the intermediate attentional demand (P = 0.04). Finally, polynomial regression indicated a nonlinear relationship between signal change from all regions and attentional load in patients (BA24: F = 7.5; P = 0.008; BA6/8: F = 6.4; P = 0.06; BA9: F = 11.2;P = 0.001), whereas the relationship was linear in controls (linear regression BA24: F = 7.0; P = 0.01; BA6/8: F = 20.7; P = 0.00002; BA9: F = 8.1; P = 0.006).

Correlation Analyses

Correlation analyses were performed within each attentional control load to investigate the relationship between brain activity and behavior. At the higher load, BA24 BOLD responses correlated inversely with accuracy in controls only (controls: r = −0.44; P = 0.045, patients: r = −0.006; P > 0.9). At the intermediate attentional load, BA9 activity was inversely correlated with accuracy in patients only (patients: r = −0.49; P = 0.05; healthy subjects: r = 0.22; P > 0.3) (Fig. 3). No significant correlations between signal change and reaction time were present.

Figure 3.

Scatterplots of correlations between brain activity and accuracy during the VAC task. (A) In controls (n = 21), a negative correlation was present between accuracy and cingulate activity (x −4, y 16, z 20) during high loads of attentional control processing. (B) In patients with schizophrenia (n = 16), a negative correlation was present between accuracy and left dorsolateral prefrontal activity (x −45, y 5, z 27) during intermediate loads of attentional control processing. See text for statistics.

Figure 3.

Scatterplots of correlations between brain activity and accuracy during the VAC task. (A) In controls (n = 21), a negative correlation was present between accuracy and cingulate activity (x −4, y 16, z 20) during high loads of attentional control processing. (B) In patients with schizophrenia (n = 16), a negative correlation was present between accuracy and left dorsolateral prefrontal activity (x −45, y 5, z 27) during intermediate loads of attentional control processing. See text for statistics.

Discussion

The present results suggest abnormal behavior and brain activity during attentional control processing in patients with schizophrenia. Patients had reduced behavioral accuracy and abnormal activity of prefrontal and cingulate regions at higher loads of attentional control. These anomalies were differentially identified as a function of the level of attentional control elicited by the VAC task. More specifically, no significant difference between groups was found at the lower level of attentional control. At the intermediate level of attentional control, patients with schizophrenia had greater activity in dorsolateral prefrontal cortex. In contrast, at the higher level of attentional control, patients had reduced activity in lateral prefrontal and cingulate cortices in the context of reduced behavioral accuracy. These abnormalities were identified in strongly interconnected brain regions that are part of a broader network associated with attentional control processing. Within this brain circuitry, previous studies have suggested that each of these areas are not segregated but specialize in different subprocesses within attentional control. In particular, the dorsolateral prefrontal cortex has been associated with representation and maintenance of the attentional demand of the task (MacDonald et al. 2000; Blasi et al. 2007), the premotor cortex and BA8 with top-down modulation of attention and motor preparation (Corbetta and Shulman 2002), and the anterior cingulate with conflict processing (Botvinick et al. 2004). Thus, our results suggest that, depending on attentional load, there is differential engagement of cortical areas associated with processing of attentional stimuli. This contention is further corroborated by analysis of signal changes that was consistent with SPM results. Furthermore, although in healthy volunteers there was an increase in activity in left and right lateral prefrontal as well as cingulate cortices with increasing demand for attentional control suggesting that capacity had not been saturated, slopes of relationships of BOLD responses to cognitive load indicated in patients nonlinear relationships in these regions, as shown by polynomial regression.

These data show region-specific hypo- and hyperactivity of different brain areas as a function of cognitive demand. They suggest that the differential brain response between patients with schizophrenia and healthy volunteers during attentional processing is region specific and load dependent. Within this conceptual framework, they are in line with data relative to other cognitive domains. In particular, several investigators have proposed that different manifestations of abnormal prefrontal activity in schizophrenia (hypo- vs. hyper-frontality) may be consistent with the putative physiological inverted-U relationship (Callicott et al. 1999) being shifted to the left in patients (Manoach 2003). In other words, capacity in these subjects may be saturated earlier resulting in more effortful and less efficient activity at lower working memory loads (hyper-frontality) or in disengagement when capacity is by far exceeded (hypo-frontality). Even though speculative in nature, an interpretation of the results of the present study in the domain of attentional control may fit this model previously applied to the working memory field further suggesting that capacity limitations in patients may be manifest differently in specific brain regions. In particular, capacity may have not been breached in controls with the current version of the VAC task. However, the negative relationship between brain activity at the higher level of attentional control and behavioral performance in controls may suggest that capacity limitation has been reached and that if we had a more difficult set of stimuli brain activity may have been reduced in prefrontal and cingulate regions. Because we do not have data speaking directly to this issue, this interpretation is speculative (Fig. 4). Differently, in patients, BOLD responses in left BA9 peaked early relative to controls and then tended to fall off, consistent with a shift to the left of the physiological brain response (Fig. 4). On the other hand, in right BA6/8 and in cingulate cortex, patients showed hypo-activity only at the higher level of attentional control, consistent with a load-dependent failure to activate brain regions crucially involved in this cognitive process (Fig. 4). In other words, the profile of the relationship between brain activity and behavior during varying attentional control may be region specific in schizophrenia. This idea is in line with the fact that different regions specializing in different subprocesses within a cortical network can behave differently even though strongly interconnected (Callicott et al. 1999, 2000). Furthermore, our findings are also consistent with previous results revealing both hypo- and hyperactivity in prefrontal cortex in patients with schizophrenia during working memory (Callicott et al. 2003), suggesting complex relationships between prefrontal activity and cognitive processing in schizophrenia.

Figure 4.

Putative models of the relationship between attentional control load and brain activity in BA9, BA6/8, and in the cingulate cortex. In BA9, schizophrenia may be associated with a left shift of the putative physiological inverted-U response. In BA6/8 and in the cingulate cortex, a load-dependent failure to activate at higher loads may be present.

Figure 4.

Putative models of the relationship between attentional control load and brain activity in BA9, BA6/8, and in the cingulate cortex. In BA9, schizophrenia may be associated with a left shift of the putative physiological inverted-U response. In BA6/8 and in the cingulate cortex, a load-dependent failure to activate at higher loads may be present.

Another possible interpretation of the present results is that a shift to the left of the inverted-U curve in patients may be present in all brain regions differentially activated by the groups. In particular, our task may be designed so that the data points describing the relationship between brain activity and behavior identify the whole inverted-U curve in some regions but only a fraction of the profile in other regions. If this hypothesis were true, the cingulate and the right BA6/8 present a different profile compared with BA9 simply because the task design may engage the different 3 regions to a different extent. However, this interpretation also remains hypothetical because we do not have specific data to test it.

Irrespective of these possible interpretations, in sum, our data suggest that abnormal responses in nodes of the attentional control network may be differentially expressed in schizophrenia as a function of spatial localization and of the cognitive load required. Furthermore, they are consistent with other studies in schizophrenia in the domain of attentional control. Like our results at the higher level of attentional control, previous findings have shown abnormal cingulate activity during tasks eliciting attentional processes (Kerns et al. 2005; Polli et al. 2008). Furthermore, greater or reduced activity of prefrontal cortex during attentional tasks has been previously reported (Weiss et al. 2003; Honey et al. 2005; Kerns et al. 2005; Laurens et al. 2005; MacDonald et al. 2005; Gur et al. 2007). Moreover, our brain activity–behavior correlations are also in line with other findings showing a negative correlation between cingulate activity and accuracy at the higher attentional demand in healthy subjects (Blasi et al. 2006), consistent with previous models suggesting involvement of the cingulate cortex in conflict processing (MacDonald et al. 2000). In particular, lower conflict predicts lower cingulate activity and better behavioral performance (Botvinick et al. 1999; Casey et al. 2000; Blasi et al. 2006). On the other hand, the significant negative correlation between accuracy and left dorsolateral prefrontal response at the intermediate demand of attentional control in patients may suggest that hyperactivity of this prefrontal region may provide assistance for reduced cortical efficiency in other nodes of the attentional network consistent with redistribution of brain resources during attentional control processing.

Earlier studies have demonstrated that cortical dopamine signaling is crucial for a variety of cognitive functions including working memory and attention (Blasi et al. 2005; Bertolino et al. 2006). Studies in animals have characterized the relationship between dopamine stimulation and cortical activity during working memory describing an inverted-U function (Goldman-Rakic 1998). Several experiments in humans and in animals have suggested that stimulation and inhibition of dopamine signaling in prefrontal cortex may shift the curve to the right or to the left, respectively (Mattay et al. 2003; Vijayraghavan et al. 2007). Moreover, reduced cortical dopamine signaling has been implicated in explaining part of the cognitive deficits in schizophrenia (Carlsson et al. 1999). Therefore, it is possible to hypothesize that modifications of the relationship between BOLD responses and attentional load in patients may be associated with abnormal dopaminergic tone in patients with schizophrenia.

Because patients had lower behavioral accuracy at higher loads, the number of correct events included in the fMRI analyses is smaller in patients and this aspect may represent a limitation of our study. However, we did not find any correlation between the number of correct events included in the analysis and the number of significantly activated voxels in prefrontal and cingulate cortex during high (controls Pearson's r = 0.08; P > 0.7; patients r = 0.03; P > 0.9) and intermediate (controls Pearson's r = 0.29; P > 0.2; patients r = −0.17; P > 0.5) loads of attentional control. Furthermore, inclusion of a smaller number of correct events in the fMRI analyses may have lead to underestimation of the BOLD response in patients resulting in generalized reduced activation. Indeed, our results demonstrate attentional load related areas of both increased and reduced activation. These latter results are found for spatial extent of activation and for signal change and suggest that a reduced number of events in patients is unlikely to explain them.

Another limitation is the large number of patients that had to be excluded (over a third of the original sample). However, it is necessary to underline that many of these patients were excluded because they exceeded the stringent motion parameters used in our fMRI analyses (we also tried to minimize the effect of motion by covarying motion in our SPM analyses). The patients excluded because of limited behavioral performance were only 4, representing only 16% of our original sample, a proportion consistent with many earlier imaging studies. Furthermore, most of the patients were male (13 of 16). This characteristic of our sample may not allow us to generalize our results to female patients. On the other hand, as reported above, groups were matched for gender, suggesting that gender effects are not likely to affect the findings.

One can argue that the fMRI results may reflect a general ceiling effect in patients, with a nonspecific reduced function in prefrontal and cingulate cortex at the higher level of attentional load. However, we believe it is important to note that we have evaluated brain responses associated with 1) only correct responses at the task that has been shown to strongly involve attentional control subprocesses (Blasi et al. 2005, 2007); 2) all subjects performed above chance at each stimulus. In this regard, it is also important that prefrontal and cingulate dysfunction is manifest as both increased and reduced engagement at 2 different levels of attentional load. As detailed above, these data may be consistent with the putative physiological inverted-U relationship being shifted to the left in patients, as suggested in the working memory domain (Manoach 2003). It is also relevant that our findings were in brain regions repeatedly associated by previous studies with conflict processing (the cingulate cortex), attentional allocation (BA9), and motor preparation (BA6). Therefore, we do not believe that reduced activity at HIGH simply reflects nonspecific ceiling effects.

It is important to note that the task stimuli used in this study have 2 simultaneously varying dimensions: the congruence/incongruence and the local–global aspects. In particular, the local–global processing aspects of the different experimental conditions may be a confound in the hemispheric localization of the results. Although we cannot resolve these issues with the current data, we believe that these aspects may be relevant in determining the general cognitive process under evaluation here which is varying levels of attentional control as obtained with differential allocation of attentional resources throughout the levels of the VAC task. In fact, we have specifically manipulated these aspects to elicit differential request for allocation of attentional resources within attentional control. In particular, according to the global precedence effect (Navon 1977), global characteristics of stimuli are more readily processed relative to local ones. Therefore, higher top-down cognitive effort in allocating attentional resources on local characteristics might be required within attentional control processing. Consistently, in a previous study of our group with the VAC task in healthy subjects (Blasi et al. 2007), we have shown greater dorsolateral prefrontal cortex activity and longer RT while allocating attentional resources on local (small arrows) relative to global (big arrow) stimuli. Furthermore, in the present study, we found progressively deteriorating behavioral performance with the increase in attentional control level. On this basis, we believe that the allocation of attentional resources on more local or more global features of stimuli may have contributed to vary attentional control load.

Another criticism of our findings may be that the behavioral and pathophysiological findings of the present study could reflect state and not trait-related pathology. However, the clinical condition of patients who had been on stable pharmacological treatment for at least 4 weeks and lack of significant correlation between PANSS subscales scores (positive, negative, and general psychopathology subscales) with behavior (all r < 0.3; all P > 0.25), or brain activity (as performed with SPM) are consistent with trait pathology. Therefore, it does not seem likely that differences between patients and controls are heavily affected by variable clinical symptoms.

Interestingly, patients with schizophrenia included in the study did not show longer reaction times. On the other hand, this is a common finding across many kinds of tasks in schizophrenia. Under these conditions, the possibility of a speed–accuracy trade-off should be considered for the interpretation of the data of this study. Finally, antipsychotic medication may have an effect on the nature of functional and behavioral abnormalities observed in patients. However, no significant correlations were found between behavioral data or signal changes extracted from the areas differentially activated between groups and chlorpromazine equivalents (all r < ±0.3; all P > 0.15).

In conclusion, these results suggest that the cognitive load required during attentional control processing strongly modulates the neurobiological functional phenotype associated with schizophrenia. Regional degree of impairment may differentially impact how load-dependent pathophysiological correlates of attentional control processing in this disorder are functionally expressed in different areas of the brain.

Funding

2007 NARSAD Young Investigator award to G.B.

We thank Terry E. Goldberg and Brita Elveevag for helping with task design. We also thank Riccarda Lomuscio for her help in data acquisition and all people that participated in this study. None of the authors has any competing interests to disclose. Conflict of Interest: None declared.

References

Arce
E
Leland
DS
Miller
DA
Simmons
AN
Winternheimer
KC
Paulus
MP
Individuals with schizophrenia present hypo- and hyperactivation during implicit cueing in an inhibitory task
Neuroimage
 , 
2006
, vol. 
32
 (pg. 
704
-
713
)
Barch
DM
Carter
CS
Braver
TS
Sabb
FW
MacDonald
A
3rd
Noll
DC
Cohen
JD
Selective deficits in prefrontal cortex function in medication-naive patients with schizophrenia
Arch Gen Psychiatry
 , 
2001
, vol. 
58
 (pg. 
280
-
288
)
Bertolino
A
Blasi
G
Latorre
V
Rubino
V
Rampino
A
Sinibaldi
L
Caforio
G
Petruzzella
V
Pizzuti
A
Scarabino
T
, et al.  . 
Additive effects of genetic variation in dopamine regulating genes on working memory cortical activity in human brain
J Neurosci
 , 
2006
, vol. 
26
 (pg. 
3918
-
3922
)
Blasi
G
Goldberg
TE
Elvevag
B
Rasetti
R
Bertolino
A
Cohen
J
Alce
G
Zoltick
B
Weinberger
DR
Mattay
VS
Differentiating allocation of resources and conflict detection within attentional control processing
Eur J Neurosci
 , 
2007
, vol. 
25
 (pg. 
594
-
602
)
Blasi
G
Goldberg
TE
Weickert
T
Das
S
Kohn
P
Zoltick
B
Bertolino
A
Callicott
JH
Weinberger
DR
Mattay
VS
Brain regions underlying response inhibition and interference monitoring and suppression
Eur J Neurosci
 , 
2006
, vol. 
23
 (pg. 
1658
-
1664
)
Blasi
G
Mattay
VS
Bertolino
A
Elvevag
B
Callicott
JH
Das
S
Kolachana
BS
Egan
MF
Goldberg
TE
Weinberger
DR
Effect of catechol-O-methyltransferase val158met genotype on attentional control
J Neurosci
 , 
2005
, vol. 
25
 (pg. 
5038
-
5045
)
Botvinick
M
Nystrom
LE
Fissell
K
Carter
CS
Cohen
JD
Conflict monitoring versus selection-for-action in anterior cingulate cortex
Nature
 , 
1999
, vol. 
402
 (pg. 
179
-
181
)
Botvinick
MM
Cohen
JD
Carter
CS
Conflict monitoring and anterior cingulate cortex: an update
Trends Cogn Sci
 , 
2004
, vol. 
8
 (pg. 
539
-
546
)
Callicott
JH
Bertolino
A
Mattay
VS
Langheim
FJ
Duyn
J
Coppola
R
Goldberg
TE
Weinberger
DR
Physiological dysfunction of the dorsolateral prefrontal cortex in schizophrenia revisited
Cereb Cortex
 , 
2000
, vol. 
10
 (pg. 
1078
-
1092
)
Callicott
JH
Mattay
VS
Bertolino
A
Finn
K
Coppola
R
Frank
JA
Goldberg
TE
Weinberger
DR
Physiological characteristics of capacity constraints in working memory as revealed by functional MRI
Cereb Cortex
 , 
1999
, vol. 
9
 (pg. 
20
-
26
)
Callicott
JH
Mattay
VS
Verchinski
BA
Marenco
S
Egan
MF
Weinberger
DR
Complexity of prefrontal cortical dysfunction in schizophrenia: more than up or down
Am J Psychiatry
 , 
2003
, vol. 
160
 (pg. 
2209
-
2215
)
Carlsson
A
Waters
N
Carlsson
ML
Neurotransmitter interactions in schizophrenia–therapeutic implications
Biol Psychiatry
 , 
1999
, vol. 
46
 (pg. 
1388
-
1395
)
Casey
BJ
Thomas
KM
Welsh
TF
Badgaiyan
RD
Eccard
CH
Jennings
JR
Crone
EA
Dissociation of response conflict, attentional selection, and expectancy with functional magnetic resonance imaging
Proc Natl Acad Sci USA
 , 
2000
, vol. 
97
 (pg. 
8728
-
8733
)
Corbetta
M
Shulman
GL
Control of goal-directed and stimulus-driven attention in the brain
Nat Rev Neurosci
 , 
2002
, vol. 
3
 (pg. 
201
-
215
)
Desimone
R
Duncan
J
Neural mechanisms of selective visual attention
Annu Rev Neurosci
 , 
1995
, vol. 
18
 (pg. 
193
-
222
)
Duncan
J
Owen
AM
Common regions of the human frontal lobe recruited by diverse cognitive demands
Trends Neurosci
 , 
2000
, vol. 
23
 (pg. 
475
-
483
)
First
MB
Gibbon
M
Spitzer
RL
Williams
JBW
Guide for the structured clinical interview for DSM-IV axis I disorders-research version
 , 
1996
New York
Biometrics Research
Ford
JM
Gray
M
Whitfield
SL
Turken
AU
Glover
G
Faustman
WO
Mathalon
DH
Acquiring and inhibiting prepotent responses in schizophrenia: event-related brain potentials and functional magnetic resonance imaging
Arch Gen Psychiatry
 , 
2004
, vol. 
61
 (pg. 
119
-
129
)
Friston
KJ
Zarahn
E
Josephs
O
Henson
RN
Dale
AM
Stochastic designs in event-related fMRI
Neuroimage
 , 
1999
, vol. 
10
 (pg. 
607
-
619
)
Goldman-Rakic
PS
The cortical dopamine system: role in memory and cognition
Adv Pharmacol
 , 
1998
, vol. 
42
 (pg. 
707
-
711
)
Gur
RE
Turetsky
BI
Loughead
J
Snyder
W
Kohler
C
Elliott
M
Pratiwadi
R
Ragland
JD
Bilker
WB
Siegel
SJ
, et al.  . 
Visual attention circuitry in schizophrenia investigated with oddball event-related functional magnetic resonance imaging
Am J Psychiatry
 , 
2007
, vol. 
164
 (pg. 
442
-
449
)
Heckers
S
Weiss
AP
Deckersbach
T
Goff
DC
Morecraft
RJ
Bush
G
Anterior cingulate cortex activation during cognitive interference in schizophrenia
Am J Psychiatry
 , 
2004
, vol. 
161
 (pg. 
707
-
715
)
Hollingshead
A
Redlich
F
Social class and metal illness
 , 
1958
New York
Wiley
Holmes
AJ
MacDonald
A
3rd
Carter
CS
Barch
DM
Andrew Stenger
V
Cohen
JD
Prefrontal functioning during context processing in schizophrenia and major depression: an event-related fMRI study
Schizophr Res
 , 
2005
, vol. 
76
 (pg. 
199
-
206
)
Honey
GD
Pomarol-Clotet
E
Corlett
PR
Honey
RA
McKenna
PJ
Bullmore
ET
Fletcher
PC
Functional dysconnectivity in schizophrenia associated with attentional modulation of motor function
Brain
 , 
2005
, vol. 
128
 (pg. 
2597
-
2611
)
Jansma
JM
Ramsey
NF
van der Wee
NJ
Kahn
RS
Working memory capacity in schizophrenia: a parametric fMRI study
Schizophr Res
 , 
2004
, vol. 
68
 (pg. 
159
-
171
)
Just
MA
Carpenter
PA
A capacity theory of comprehension: individual differences in working memory
Psychol Rev
 , 
1992
, vol. 
99
 (pg. 
122
-
149
)
Kerns
JG
Cohen
JD
MacDonald
AW
3rd
Johnson
MK
Stenger
VA
Aizenstein
H
Carter
CS
Decreased conflict- and error-related activity in the anterior cingulate cortex in subjects with schizophrenia
Am J Psychiatry
 , 
2005
, vol. 
162
 (pg. 
1833
-
1839
)
Laurens
KR
Kiehl
KA
Ngan
ET
Liddle
PF
Attention orienting dysfunction during salient novel stimulus processing in schizophrenia
Schizophr Res
 , 
2005
, vol. 
75
 (pg. 
159
-
171
)
MacDonald
AW
3rd
Becker
TM
Carter
CS
Functional magnetic resonance imaging study of cognitive control in the healthy relatives of schizophrenia patients
Biol Psychiatry
 , 
2006
, vol. 
60
 (pg. 
1241
-
1249
)
MacDonald
AW
3rd
Carter
CS
Kerns
JG
Ursu
S
Barch
DM
Holmes
AJ
Stenger
VA
Cohen
JD
Specificity of prefrontal dysfunction and context processing deficits to schizophrenia in never-medicated patients with first-episode psychosis
Am J Psychiatry
 , 
2005
, vol. 
162
 (pg. 
475
-
484
)
MacDonald
AW
3rd
Cohen
JD
Stenger
VA
Carter
CS
Dissociating the role of the dorsolateral prefrontal and anterior cingulate cortex in cognitive control
Science
 , 
2000
, vol. 
288
 (pg. 
1835
-
1838
)
Manoach
DS
Prefrontal cortex dysfunction during working memory performance in schizophrenia: reconciling discrepant findings
Schizophr Res
 , 
2003
, vol. 
60
 (pg. 
285
-
298
)
Mattay
VS
Goldberg
TE
Fera
F
Hariri
AR
Tessitore
A
Egan
MF
Kolachana
B
Callicott
JH
Weinberger
DR
Catechol O-methyltransferase val158-met genotype and individual variation in the brain response to amphetamine
Proc Natl Acad Sci USA
 , 
2003
, vol. 
100
 (pg. 
6186
-
6191
)
Navon
D
Forest before trees: the precedence of global features in visual perception
Cognit Psychol
 , 
1977
, vol. 
9
 (pg. 
353
-
383
)
Oldfield
RC
The assessment and analysis of handedness: the Edinburgh inventory
Neuropsychologia
 , 
1971
, vol. 
9
 (pg. 
97
-
113
)
Perlstein
WM
Dixit
NK
Carter
CS
Noll
DC
Cohen
JD
Prefrontal cortex dysfunction mediates deficits in working memory and prepotent responding in schizophrenia
Biol Psychiatry
 , 
2003
, vol. 
53
 (pg. 
25
-
38
)
Polli
FE
Barton
JJ
Thakkar
KN
Greve
DN
Goff
DC
Rauch
SL
Manoach
DS
Reduced error-related activation in two anterior cingulate circuits is related to impaired performance in schizophrenia
Brain
 , 
2008
, vol. 
131
 (pg. 
971
-
986
)
Vijayraghavan
S
Wang
M
Birnbaum
SG
Williams
GV
Arnsten
AF
Inverted-U dopamine D1 receptor actions on prefrontal neurons engaged in working memory
Nat Neurosci
 , 
2007
, vol. 
10
 (pg. 
376
-
384
)
Weiss
EM
Golaszewski
S
Mottaghy
FM
Hofer
A
Hausmann
A
Kemmler
G
Kremser
C
Brinkhoff
C
Felber
SR
Fleischhacker
WW
Brain activation patterns during a selective attention test-a functional MRI study in healthy volunteers and patients with schizophrenia
Psychiatry Res
 , 
2003
, vol. 
123
 (pg. 
1
-
15
)
Yacubian
J
Sommer
T
Schroeder
K
Glascher
J
Kalisch
R
Leuenberger
B
Braus
DF
Buchel
C
Gene-gene interaction associated with neural reward sensitivity
Proc Natl Acad Sci USA
 , 
2007
, vol. 
104
 (pg. 
8125
-
8130
)
Yucel
M
Pantelis
C
Stuart
GW
Wood
SJ
Maruff
P
Velakoulis
D
Pipingas
A
Crowe
SF
Tochon-Danguy
HJ
Egan
GF
Anterior cingulate activation during Stroop task performance: a PET to MRI coregistration study of individual patients with schizophrenia
Am J Psychiatry
 , 
2002
, vol. 
159
 (pg. 
251
-
254
)
Zhang
Y
Bertolino
A
Fazio
L
Blasi
G
Rampino
A
Romano
R
Lee
ML
Xiao
T
Papp
A
Wang
D
, et al.  . 
Polymorphisms in human dopamine D2 receptor gene affect gene expression, splicing, and neuronal activity during working memory
Proc Natl Acad Sci USA
 , 
2007
, vol. 
104
 (pg. 
20552
-
20557
)