INTRODUCTION

Alterations of functional connectivity of the brain have been implicated in the pathophysiology of schizophrenia (Friston and Frith, 1995; Meyer-Lindenberg et al, 2001; Weinberger et al, 1992). Although initial studies focused on abnormal dynamic interactions among functionally segregated brain regions co-activated during a task (Meyer-Lindenberg et al, 2001), a growing number of reports has recently suggested that disruptions in the default mode network (DMN) may have a role in the cognitive deficits found in schizophrenia (see (Broyd et al, 2009) for a review).

The DMN comprises a set of brain regions whose activity increases during rest and is attenuated during an active task (Mazoyer et al, 2001; Raichle et al, 2001). This brain network includes midline areas such as the ventromedial prefrontal cortex (vmPFC), the posterior cingulate cortex (pCC), the precuneus, as well as bilateral inferior parietal lobule (IPL) (Raichle et al, 2001). Activity in the DMN has been associated with stimulus-unrelated thoughts, lapses of attention and mind wandering (Mason et al, 2007). Different roles have been hypothesized for this network ranging from monitoring external environment (‘watchfulness’), to internal mentation, for example, autobiographical memory, theory of mind, moral decision making and prospective thoughts (Gusnard et al, 2001). Interestingly, activity of the DMN shows a strong negative correlation, ‘anticorrelation’, with activity in task-related networks (Fox et al, 2005). This finding suggests the existence of two different modes of information processing: one with predominant DMN activity, which is relatively unfocused on external objects and characterized by mental exploration, whereas the other is focused on active processing of external stimuli (see (Buckner et al, 2008) for a review). Importantly, integrity of the DMN is also important for allocation of attentional resources needed for cognitive processing and has been associated with working memory (WM) performance (Hampson et al, 2006; Sambataro et al, 2008).

Although brain functions proposed for the DMN are impaired in patients with schizophrenia (Buckner et al, 2008), neuroimaging studies have reported inconsistent findings in the DMN of these individuals. Increased (Garrity et al, 2007; Harrison et al, 2007; Whitfield-Gabrieli et al, 2009; Zhou et al, 2007) or decreased (Liang et al, 2006; Pomarol-Clotet et al, 2008) activity and connectivity have been reported in the DMN network at the level of individual regions as well as of the global network. Different issues may have contributed to lack of consistent findings, including the method used for estimation of connectivity (Murphy et al, 2009). For example, the use of global signal regression, a standard preprocessing step of univariate seed-based connectivity analyses, has been questioned for the study of DMN as having a potential biasing effect in the estimation of the magnitude of the anticorrelations. Given that the use of this technique is still controversial (Fox et al, 2009; Murphy et al, 2009; Weissenbacher et al, 2009), other statistical approaches that do not rely on this preprocessing method such as independent component analysis (ICA) may yield more reliable results.

Another crucial determinant potentially impacting data consistency and DMN activity may be the effect of pharmacological treatment with antipsychotics. Although some reports are questioning the size of the effect and specificity of the cognitive effects of second-generation antipsychotics (Goldberg et al, 2007; Keefe et al, 2007), some evidence shows that this treatment is associated with improvement of some of the cognitive deficits of schizophrenia including WM (Bertolino et al, 2004; Purdon et al, 2000). Furthermore, second-generation antipsychotic treatment appears to modulate brain physiology underlying WM function with reduced activation of dorsolateral prefrontal cortex (DLPFC) and parietal cortex for better levels of performance, suggesting increased neural efficiency (Bertolino et al, 2004). Moreover, other studies have indicated that systemic administration of levodopa and apomorphine in healthy humans or in patients with Parkinson’s disease can modify the activity of the DMN (Argyelan et al, 2008; Nagano-Saito et al, 2009) further suggesting the potential relationship between dopamine signaling and DMN modulation. On the other hand, despite these earlier studies, the effect of treatment with antipsychotics on the DMN is still unknown.

In this study, we hypothesized that the effects of treatment with olanzapine, a second-generation antipsychotic drug, on WM brain physiology and behavioral performance can also be associated with modulation of connectivity within the DMN. Thus, we performed a longitudinal study in previously untreated patients with schizophrenia undergoing 8 weeks of treatment with olanzapine monotherapy. Patients were studied using spatial group ICA with fMRI during WM at 4 (t1) and 8 weeks of treatment (t2). To control for potential well-known effects of task practice and learning (Goldberg et al, 2007), we also scanned normal controls twice (t1, t2) with a time interval similar to that of patients.

MATERIALS AND METHODS

Subjects

A total of 19 patients with schizophrenia and 19 healthy controls, matched for age, gender, race, handedness participated (Oldfield, 1971) and socioeconomic status (Hollingshead and Redlich, 1958) participated in this study (Table 1). All subjects underwent clinical assessment following the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-IV; First et al, 1996). Exclusion criteria included past history or presence of any medical, neurological disorders, drug or alcohol abuse, active drug use in the past year, past head trauma with loss of consciousness. Normal controls were excluded if presented with past history or current diagnosis of psychiatric disorders as documented by SCID-IV.

Table 1 Demographics of the Sample
Full size table

At the time of enrollment, all patients were experiencing acute exacerbation of psychosis requiring hospitalization. Nine patients were diagnosed with schizophreniform disorder and all of them were diagnosed with schizophrenia at a later follow-up. In all 13 patients were drug-naive and 6 had been drug-free for at least 2 weeks from oral antipsychotics or from at least two cycles of depot antipsychotic treatment. Clinical symptoms in patients were assessed on the day of study entry, at 4 weeks (t1), and at 8 weeks (t2) by a trained psychiatrist using the Positive and Negative Syndrome Scale (PANSS; see Table 2 for clinical data). All patients were treated with oral olanzapine monotherapy. Titration was allowed for the first 2 weeks, and then the dose was kept constant throughout the remaining 6 weeks of treatment (mean dose 20 mg/day; SD=6.9 mg/day). Given the lack of reliable algorithms to handle missing data points in statistical imaging, only subjects with two longitudinal fMRI scans were selected for this study.

Table 2 Clinical Data
Full size table

All participants gave written informed consent, approved by the Institutional Review Board of the University of Bari, Bari (Italy), to take part in the experiment.

Experimental Task

All subjects performed the N-back WM task as described elsewhere (Callicott et al, 1999). Briefly, subjects saw numbers (1–4) shown in pseudorandom sequence and displayed at the corners of a diamond-shaped box. A non-memory guided control condition (0-back) that required subjects to identify the number currently seen, alternated with the 2-back WM condition, in which subjects were asked to recall the number seen two stimuli before, while continuing to encode additionally incoming stimuli. Performance was recorded through a fiber optic response box which allowed measurement of behavioral data as the number of correct responses (accuracy) and reaction time (RT). Data recordings of two patients were lost because of a computer glitch. The stimuli were arranged in a block-design, consisting of eight 30-s blocks: four blocks of the control condition alternating with four blocks of the 2-back WM condition.

Image Acquisition

Blood-oxygen level dependent fMRI was performed on a GE Signa (Milwaukee, WI) 3T scanner at t1 and t2. A gradient echo Blood-oxygen level dependent-EPI pulse sequence was used to acquire 120 images per run. Each functional image consisted of 20 6-mm-thick axial slices covering the entire cerebrum and most of the cerebellum (TR=2000 ms; TE=30 ms; field of view=24 cm; flip angle=90°).

Data Analysis

Demographics, behavioral, and clinical data

Two sample t-tests and repeated measures ANOVAs were used to compare demographics and behavioral data, respectively. Chi-square analyses were performed to compare categorical variables. Clinical data were compared using paired t-tests.

Image analysis

Preprocessing: Data were pre-processed and analyzed using Statistical Parametrical Mapping 2 (SPM; http://www.fil.ion.ucl.ac.uk). The first four volumes were discarded to allow for T1 equilibration effects. All functional volumes were realigned to the first volume acquired and spatially normalized to the Montreal Neurological Institute (MNI) standard brain in the space of Talairach and Tournoux (1988). Smoothing was carried out with a 10-mm full width half maximum isotropic 3D Gaussian kernel.

Independent component analysis: One group of spatial ICA was performed on preprocessed data using the Group ICA of fMRI Toolbox [GIFT; http://icatb.sourceforge.net, (Calhoun et al, 2001)] entering the data of all subjects for both time points. The dimensionality of functional data for each subject was reduced using three consecutive steps of Principal Component Analysis alternating with data concatenation across the subjects, resulting in one aggregate mixing matrix for all subjects. An ICA decomposition using the Infomax algorithm was used to extract 21 independent components (IC), consisting of group spatial maps and the related time courses (TC) of the estimated signal. Individual participant's spatial maps for each IC, which are the voxel-wise IC loadings, represent local strength of functional connectivity and reflect the correspondence between the estimated TC in each voxel and the average TC of the network itself. Given that in spatial ICA, the individual subject TC is assumed to be constant across the entire brain, the voxel-wise estimated signal deviations from the average network TC reflects local changes in the strength of functional connectivity within a given network. The minimum description length criteria were used to estimate the order selection that is the number of ICs from the smoothed data sets after taking into account the spatial and temporal correlation of the fMRI data (Li et al, 2007). These components were then used to back-reconstruct to individual ICs using the aggregate mixing matrix created during the reduction steps of dimensionality data. Each voxel value of the IC map for each participant was calibrated using z-scores for between participant comparisons.

Eventually, two components that resulted unstable as suggested by a coefficient of stability (Iq) lower than 0.95 calculated by 50 bootstrapped permutated estimations of the ICs (ICASSO; Himberg et al, 2004) or showed artifactual patterns defined by those ICs with a spatial correlation greater than R2=0.02 with white matter and R2=0.05 for CSF (Kim et al, 2009) were removed from the analysis. The ICs that had (1) the highest negative Pearson's r correlation between its TC and the 2-back condition (a box-car function convolved with a double-Gaussian hemodynamic response function) and (2) the highest spatial correlation with a DMN template provided with the ICA toolbox were selected for further analyses.

Spatial analyses: Each subject's DMN-IC was entered into SPM5 and analyzed using second level random-effects analyses. Repeated measures ANCOVA with the normalized 2-back performance (ratio between 2-back and 0-back accuracy) as a covariate of no interest and time as a within subjects variable was used to compare DMN-IC spatial maps across diagnostic groups at each time point. Additional planned comparisons were performed to assess differences across diagnostic groups at each time point. To examine differences in positive and negative spatial maps separately, we created a positive and negative mask, respectively, using a one-sample t-test across all subjects. Average connectivity strength was extracted in significant clusters and correlated with WM performance computed as a ratio between percent accuracy at 2-back and at 0-back.

A statistical threshold of p<0.05 with False-Discovery-Rate (FDR; Genovese et al, 2002) small-volume-correction with q-FDR=0.05 was used to identify significant differences within regions of interest. Regions of interest were created using WFU pickatlas software version 1.04 (Functional MRI Laboratory at the Wake Forest University School of Medicine, http://www.rad.wfubmc.edu/fmri) and comprised the following a priori regions as part of the standard default mode: vmPFC, anterior cingulate cortex (BA24/32), and the IPL, as well as the posterior cingulate, and precuneus (Raichle et al, 2001). All coordinates are reported in MNI system.

Stimulus-related response: Averaged IC responses to the WM task were calculated for the DMN-IC TC for patients and normal controls separately during the 2-back condition using the deconvolution algorithm by Eichele et al (2008). Stimulus-related response depicts the extent of the modulation of the DMN TC during presentation of a WM stimulus of null duration. Averaged TCs were overlaid for both groups for 17 s following the stimulus onset to estimate the full extent of the TC response to stimuli.

RESULTS

Behavioral Performance

There was a significant effect of diagnosis [F(1, 34)=13.596, p=0.0007] on task accuracy: patients with schizophrenia had reduced performance relative to normal controls. There was also an effect of time [F(1, 34)=9.151, p=0.004] indicating that subjects at t1 had lower performance relative to t2. There was no interaction of diagnosis by time of observation (p>0.1). Even though both groups of subjects tended to improve accuracy over time, a t-test for dependent samples demonstrated that the difference was significant in patients (t16=2.05, p=0.05) but not in controls (t18=1.61, p=0.12), suggesting an effect of olanzapine on WM accuracy in patients. There was a significant effect of diagnosis [F(1, 34)=5.1299, p=0.03] on reaction time, with patients with schizophrenia being slower relative to normal controls. There was no effect of time of observation (p>0.1) or interaction of diagnosis by time of observation (p>0.1) on reaction time.

Imaging Results

Default mode network component

One IC spanning across the DMN regions and anticorrelated with the task was identified for all subjects across both time points (Figure 1a). This IC included midline regions vmPFC, pCC, precuneus (BA5/7), as well as bilateral IPL. This IC had high spatial correlation with a DMN template, R2=0.33 and high temporal anticorrelation with the timing of the 2-back condition of task, r=−0.65 (Figure 1b).

Figure 1
figure 1

Default mode network (DMN). One DMN-IC comprising spatial maps and time courses was identified across both time points for all the subjects. (a) Axial maps display the spatial pattern of the independent components identified. (b) Time courses represent the temporal profile of DMN each component (blue continuous line) overlaid on the paradigm ‘box-car’ design (red dashed line). All images are thresholded at p=0.05 corrected for multiple comparisons. Color bar indicates t-scores.

Full size image

Between group comparisons.

Spatial extent: Effect of Diagnosis Patients with schizophrenia had increased connectivity with the DMN in left IPL (BA39, x=−56, y=−67, z=23, Z=4.03, p=0.00003 FDR corrected) as well as in precuneus (BA7, x=4, y=−79, z=45, Z=3.33, p=0.0004 with q-FDR=0.09). In addition, patients had decreased connectivity with the DMN in pCC (BA23/31, x=4, y=−52, z=26, Z=3.34, p=0.0004 FDR corrected; Figure 2).

Figure 2
figure 2

Effect of diagnosis on brain connectivity in the DMN. Patients with schizophrenia show decreased connectivity strength within the DMN in pCC (a) as well as increased connectivity in left-Inferior Parietal Lobule (BA39) and precuneus (BA7) relative to normal controls (b). Statistical t-map of DMN connectivity overlaid on the MNI brain template. Color bar indicates t-scores.

Full size image

Effect of time There was no significant effect of time.

Time by diagnosis There was a time by diagnosis interaction in vmPFC (BA10, x=0, y=49, z=−7, Z=3.02, p=0.001 FDR corrected; Figure 3a). In patients with schizophrenia the strength of connectivity increased at t2 relative to t1 with respect to normal controls (Figure 3b). Post hoc analysis revealed that patients with schizophrenia had significantly greater connectivity strength extracted in the vmPFC cluster at t2 relative to t1 (p=0.01). However, this was not true in normal controls (p>0.1). In addition, at t2 connectivity strength in patients in this region was greater relative to normal controls (p=0.01). Strength of connectivity in vmPFC was positively correlated with WM accuracy in patients, but not in normal controls (controls: t1: r=−0.20, p>0.1; t2: r=−0.15, p>0.1; Figure 3c). Furthermore, in patients the correlation was significant at t1 but not at t2 (patients: t1: r=0.49, p=0.05; t2; r=0.16, p>0.1).

Figure 3
figure 3

Time by diagnosis interaction on brain connectivity in the DMN. In patients with schizophrenia (SCZ) connectivity strength within the DMN increased in ventromedial prefrontal cortex (vmPFC) (a); at t2 (red) relative to t1 (blue) when compared with normal controls (NC) (b); connectivity strength in vmPFC was positively correlated with task performance in patients at t1 (r=0.45, p=0.04) (c); statistical t-maps of DMN connectivity are overlaid on the MNI brain template. Color bar indicates t-scores. Bar plots report parameter estimates of the time by diagnosis contrast from the ANCOVA extracted from the vmPFC cluster and measured in arbitrary units (a.u.). Dashed lines indicate the trendline. *P=0.01.

Full size image

Average stimulus-related responses

Average stimulus responses of DMN IC to the WM condition had a smaller negative modulation of this signal in patients relative to normal controls. This difference in WM-related responses decreased at t2 relative t1 (Figure 4).

Figure 4
figure 4

Time by diagnosis interaction on stimulus-related responses. At t1 patients had a smaller stimulus-related decrease of the time course amplitude relative to normal controls, and this difference decreased at t2 Y axis indicates the difference (patients—normal controls, SCZ—NC) of time course amplitude of the DMN-IC patients with schizophrenia relative to normal controls in arbitrary units (a.u.) as calculated by deconvolving the estimated signal for the 2-back condition, X axis indicates latency from the working memory stimulus onset expressed in seconds. The dashed arrow indicates the effect of time on the group differences.

Full size image

Negative DMN component

The negative component of DMN had a pattern similar to the typical WM task described during 2-back task (Supplementary Figure S1). This network of brain regions included bilaterally dorsolateral (DLPFC) and ventrolateral prefrontal cortex, frontal pole, caudate, anterior cingulate, and lateral posterior parietal cortex. This network demonstrated main effects and interaction between diagnosis and treatment in the region of DLPFC previously shown to be responsive to treatment with olanzapine as in earlier studies (Bertolino et al, 2004; see Supplementary Materials for further details). In normal controls (t1: r=0.40, p=0.09; t2=0.39, p=0.1) but not patients (t1: r=−0.03, p>0.1; t2=−0.14, p>0.1) there was a statistical trend for correlation between DLPFC connectivity strength and WM accuracy.

DISCUSSION

In this study we investigated the effect of olanzapine on DMN connectivity in patients with schizophrenia. We found that patients have greater strength of connectivity in the IPL and precuneus as well as reduced connectivity in pCC. These effects were not significantly affected by treatment with olanzapine. On the other hand, our data indicate an interaction of time by diagnosis on connectivity strength in vmPFC where patients with schizophrenia had greater connectivity strength with the rest of DMN during olanzapine treatment. In patients with schizophrenia the extent of DMN connectivity strength in vmPFC positively correlated with WM accuracy at t1 but not at t2.

Previous imaging studies have reported widespread alterations of DMN in patients with schizophrenia (see Broyd et al, 2009 for a review). Our findings indicate decreased functional connectivity with the DMN in pCC as well as increased connectivity strength in parietal cortex, including precuneus and IPL, in patients with schizophrenia. pCC is an important node of the DMN and a crucial cortical hub for interconnecting segregated brain areas, that is, anterior and posterior midline systems that are mPFC and entorhinal cortex (Greicius et al, 2009). As a result of these connections pCC may have an important role in cognitive functions including episodic memory (Greicius et al, 2003) and WM (Hampson et al, 2006; Sambataro et al, 2008). In addition, disruption of functional coupling with pCC is associated with ADHD (Castellanos et al, 2008) and cognitive decline in physiological (Sambataro et al, 2008) and pathological aging (Greicius et al, 2004). The relative lack of disease specificity of these findings suggests that this mechanism reflects basic disruption of the brain systems associated with memory dysfunction across different categorical diagnostic conditions. Thus, decreased connectivity strength between pCC and the other regions of DMN may contribute to the WM deficits observed in patients with schizophrenia. On the other hand, connectivity of the parietal regions, that is, precuneus and IPL, with the DMN has crucial anatomical and functional differences within posterior DMN regions. Cytoarchitectonically, the precuneus is a fully differentiated isocortex as opposed to the limbic cortex in pCC, and it also has structural connections with different thalamic nuclei and with occipital, parietal, and frontal areas (Buckwalter et al, 2008). This region is associated with visual processing, motor planning (Cavada and Goldman-Rakic, 1989), cognitive functions, such as motor imagery, spatial navigation (Sato et al, 2006), memory retrieval (Lundstrom et al, 2005), and also self-perception and internal mentation (Gusnard et al, 2001). On the lateral parietal cortex, IPL which is a heteromodal association area, also has a crucial role in integrating incoming sensory information (Lynch, 1980) and in executive functions, specifically WM, sustained attention and dual task performance (Torrey, 2007). Increased connectivity of parietal regions within the DMN may be compensatory for deficits in other brain regions within the DMN, for example, pCC.

The DMN anti-correlated network comprised a set of brain regions that are recruited during WM tasks (Callicott et al, 1999). Patients with schizophrenia had increased connectivity in DLPFC which reflects the well-known prefronto-cortical dysfunction in schizophrenia (Weinberger et al, 1986). Consistent with these prefrontal effects during working memory, after 8 weeks of treatment with olanzapine, patients with schizophrenia had greater connectivity in vmPFC. Interestingly, connectivity in vmPFC correlated significantly with behavioral accuracy in patients at the first time point only and not in controls. On the other hand, DLPFC activation in healthy subjects correlated with WM behavioral performance both at t1 and t2. Alterations in vmPFC have been reported in patients with schizophrenia and this finding has been associated with abnormal prefronto-cortical function (Meyer-Lindenberg et al, 2001; Pomarol-Clotet et al, 2008). vmPFC is involved in monitoring and integrating emotional and cognitive processing through its projections to limbic regions, orbitofrontal cortex, ventral striatum, amygdale, and hypothalamus (Gusnard et al, 2001). This brain region has an important role in decision making as well as in emotion and motivation associated with a cognitive task. Neurons in vmPFC encode the relative current value and the outcome of chosen options during decisions by comparing available options (Boorman et al, 2009; Luk and Wallis, 2009). In addition, vmPFC regulates both the emotional context and motivational milieu during a cognitive task (Groenewegen and Uylings, 2000) so that this region inhibits emotional processing that may interfere with cognitive performance in a ‘dynamic interplay’ with the regions of DMN anticorrelated network and specifically DLPFC (Longe et al, 2009). In keeping with this, Uddin et al (2009) found that DMN activity in vmPFC ‘Granger-causes’ the activity in the anticorrelated network and not vice versa. Decreased temporal anticorrelation between the DMN and task-related network may impair task performance either for decreased allocation of attentional resources for the task at hand (Fransson, 2006) or for direct interference of non suppressed DMN (‘default mode interference hypothesis’ Sonuga-Barke and Castellanos, 2007). Our present data in vmPFC are consistent with this earlier literature. Moreover, they suggest that olanzapine treatment is associated with a more physiological modulation of the relationship between vmPFC connectivity and behavioral accuracy which is found only in patients and only at t1. This interpretation is also consistent with the possibility that olanzapine treatment may modulate DMN directly or acting on both anticorrelated networks (Bertolino et al, 2004; Blasi et al, 2009). As a result of this modulation, olanzapine can improve the temporal characteristics of brain response (Schlagenhauf et al, 2008) of these networks so that patients have a faster and deeper hemodynamic response to the task at hand (Figure 4) which may improve WM performances.

Olanzapine may affect DMN connectivity via dopaminergic system either directly on pCC or indirectly via PFC modulation. Evidence from animal studies indicates that olanzapine increases extracellular levels of dopamine in PFC (Gessa et al, 2000; Ichikawa et al, 2001). Dopamine can regulate brain networks by modulating the synchronization among different brain regions and membrane oscillation frequencies (Seamans and Yang, 2004) thus increasing stability of attentional networks, which ultimately results in decreased trial-by-trial variability and increased signal-to-noise ratio (Rolls et al, 2008). Decreased dopamine levels as resulting from carrying val alleles of COMTval158met polymorphism have been associated with increased brain deactivations in pCC during a prosaccade task (Ettinger et al, 2008) as well as a decreased sequence learning-related deactivation in mPFC (Argyelan et al, 2008). Thus increased levels of dopamine in the hubs of DMN may affect the connectivity in the whole network. Alternatively, treatment with olanzapine may also indirectly regulate the DMN connectivity via an increase of dopamine levels in the anticorrelated network (Bertolino et al, 2004).

A caveat to our study is the lack of fMRI data during WM at baseline that for clinical reasons could not be acquired. It is extremely difficult to obtain co-operation from patients to undergo complicated research experiments during acute exacerbations requiring hospitalization. We cannot exclude that patients may have already had an improvement at the level of brain physiology and of behavioral performance before 4 weeks of treatment with olanzapine relative to the baseline. Nevertheless, our data acquired within the timeframe of this study suggest that olanzapine modulates not only DMN connectivity, but, more importantly, its relationship with WM performance.

Another limitation of this study is the intrinsic reliability of the fMRI data in longitudinal studies. Manoach et al (2001), have reported a low test-retest reliability of the magnitude and extent of brain activations during a Sternberg Item recognition task in patients with schizophrenia across scan sessions. To control this variance, we used a bootstrapped ICA that is a technique well-known to be able to identify and isolate artifactual and noise components (Calhoun et al, 2001).

Finally, although estimation of DMN may be more optimal during non complex cognitive tasks (Esposito et al, 2006), some studies have reported similar spatial extent of DMN during rest and during variously demanding cognitive tasks (Calhoun et al, 2008; Smith et al, 2009). Moreover, Calhoun et al (2008) found similar results during rest and during an oddball task when comparing results between patients with schizophrenia and normal controls. Furthermore, most of the studies on DMN in patients with schizophrenia have used complex cognitive tasks such as an oddball task (Garrity et al, 2007), Multi-Source Interference Task (Harrison et al, 2007) and 1-back and 2-back (Pomarol-Clotet et al, 2008; Whitfield-Gabrieli et al, 2009). These studies have reported results that are similar to those obtained with rest data. The objective of our study was to investigate potential effects of treatment using a within subject design under the same task load conditions. Therefore, this design has allowed estimation and evaluation of our objective under very similar physiological conditions thus contributing to temper potential limitations associated with assessing DMN during cognition.

In conclusion, our data suggest that treatment with olanzapine can modulate brain connectivity in DMN and in its anticorrelated network, and their task-related responses. Furthermore, our findings shed new light on the functional anatomy underlying the DMN network, suggesting also potential different roles of posterior DMN regions in the pathogenesis of schizophrenia.