Altered error-related activity in patients with schizophrenia

Abstract

Deficits in working memory (WM) and executive cognitive control are core features of schizophrenia. However, findings regarding functional activation strengths are heterogeneous, partly due to differences in task demands and behavioral performance. Previous investigators proposed integrating these heterogeneous findings into a comprehensive model of cortical inefficiency assuming the inverted U-shaped relationship between performance and neuronal activation to be shifted in patients.

The present study pursued a novel approach of precise performance balancing potentially resolving a number of discrepancies with regard to performance related cognitive activation patterns in schizophrenia. Error-related WM activity was examined in 55 patients and 55 controls by balancing the mean number of incorrect responses between the groups and analyzing remaining incorrect trials. A modified Sternberg Item Recognition task (SIRT) allowing for the segregation of encoding, executive maintenance and retrieval-related activation was applied.

Relative to healthy controls, patients showed extended hypoactivations in inferior temporal, superior parietal, inferior, middle and superior frontal as well as cerebellar regions during encoding of incorrectly remembered items. During erroneous retrieval of information patients exhibited a significantly decreased activation in an executive control network comprising inferior and middle frontal areas, precuneus and cerebellum.

Present data suggest that in patients with schizophrenia encoding of erroneously retrieved items as well as erroneous retrieval itself is associated with extended activation abnormalities in task-relevant regions even after balancing performance. Thus, present results clearly indicate that disorder-characteristic activation abnormalities become manifest during dysfunctional executive processing even when the moderating effect of performance is largely eliminated.

Introduction

Deficits in working memory (WM) and executive control are prevailing in patients with schizophrenia. They have been shown to contribute to poor functional outcomes (Green, 2006) and to persist even after remission of clinical symptoms (Park, Puschel, Sauter, Rentsch, & Hell, 1999). Multiple studies therefore made an effort to contribute to a better understanding of these deficits and their neural correlates. Their findings point towards a network of multiple areas, predominantly in the frontal and parietal lobe, which seems to be abnormally activated during WM and executive processing in patients with schizophrenia (Cairo et al., 2006, Honey and Fletcher, 2006, Tan et al., 2005). Results are still rather inconsistent, however. Methodological variability can be assumed to constitute one of the main reasons for this inconsistency. Studies vary significantly with regard to the cognitive processes that were investigated. They point to abnormalities in association with exclusive maintenance of stimulus material (Tan et al., 2005), executive processing (Cannon et al., 2005), as well as stimulus encoding (Cairo et al., 2006, Johnson et al., 2006). Task difficulty is another methodologically relevant factor that varies significantly between studies. Individually perceived task difficulty and resulting task performance is known to constitute one of the most potent moderating variables when comparing the neural correlates between patients and healthy controls (Holcomb, 2004, Manoach, 2003, Van Snellenberg et al., 2006). Previous investigators proposed integrating these heterogeneous findings into a comprehensive model of cortical inefficiency. This model assumes the inverted U-shaped relationship between cognitive demand and neuronal activation to be shifted in patients (Callicott et al., 2003, Manoach, 2003).

Against this background, in a preceding study we made an effort to eliminate the moderating influence of performance by matching patients and controls according to performance and analyzing correctly performed trials only (Schlosser et al., 2008). We found that during encoding of stimulus material patients exhibited a distinct hypoactivation in a wide-spread fronto-temporo-parieto-cerebellar network. During subsequent executive processing, however, patients showed increased activation in DLPFC compared to controls. We interpreted this finding in light of a stable physiological encoding deficit in patients efficiently compensated by subsequently increased activation in DLPFC.

To find out whether this compensatory increased activation in DLPFC would be detectable in association with dysfunctional processing, as well, the current study investigated error-related activity in a larger sample of patients with schizophrenia. As shown in a meta-analysis by Van Snellenberg et al. (2006) accuracy accounts for a large amount of the variability in activation differences between patients and healthy controls in the context of working memory processing provided that ceiling effects in the control group can be excluded. The present study therefore intended to examine the “genuine”, disorder-related cognitive deficits and their neural correlates in association with dysfunctional working memory processing which are not confounded by these group-specific accuracy effects. Thus, we investigated incorrect performance by using an innovative approach of a trial based performance balancing that takes the aspect of inter-group comparability in statistical power (i.e. accuracy differences between the groups) into account. As a WM paradigm we applied a modified Sternberg task which allows for a precise segregation of encoding, executive maintenance and retrieval—processes regarded as central within the multi-component model of WM (Repovs & Baddeley, 2006) and analyzed incorrect trials only by balancing the number of incorrect trials between the groups (for a more detailed description of the algorithm please refer to Section 2). Due to the lack of studies on error-related WM processing in schizophrenia, we based our hypotheses primarily on our recent results (Schlosser et al., 2008). We hypothesized that the presumed encoding and executive control deficits reported previously would be clearly noticeable when analyzing incorrect performance. We expected this dysfunctional processing to go along with significant hypoactivation in a task-relevant network in patients compared to controls. As opposed to our preceding study that revealed compensatory increased DLPFC activation in patients during executive maintenance in association with correct performance we expected that during incorrect performance this compensatory frontal cortex increase would be missing.