Functional Neuroanatomy of Auditory Working Memory in Schizophrenia: Relation to Positive and Negative Symptoms V. Menon,* , † , ‡ R. T. Anagnoson,* , ‡ D. H. Mathalon,§ G. H. Glover, ¶ and A. Pfefferbaum§ *Departments of Psychiatry and Behavioral Sciences, ¶ Department of Radiology, and †Program in Neuroscience, Stanford University School of Medicine, Stanford, California 94305-5719; ‡VA Palo Alto Health Care System, Palo Alto, California 94304; and §SRI International, Menlo Park, California 94025 Received June 19, 2000 Functional brain imaging studies of working memory (WM) in schizophrenia have yielded inconsistent results regarding deficits in the dorsolateral prefrontal (DLPFC) and parietal cortices. In spite of its potential importance in schizophrenia, there have been few inves- tigations of WM deficits using auditory stimuli and no functional imaging studies have attempted to relate brain activation during auditory WM to positive and negative symptoms of schizophrenia. We used a two- back auditory WM paradigm in a functional MRI study of men with schizophrenia (N  11) and controls (N  13). Region of interest analysis was used to investigate group differences in activation as well as correlations with symptom scores from the Brief Psychiatric Rating Scale. Patients with schizophrenia performed signifi- cantly worse and were slower than control subjects in the WM task. Patients also showed decreased lateraliza- tion of activation and significant WM related activation deficits in the left and right DLPFC, frontal operculum, inferior parietal, and superior parietal cortex but not in the anterior cingulate or superior temporal gyrus. These results indicate that in addition to the prefrontal cortex, parietal cortex function is also disrupted during WM in schizophrenia. Withdrawal-retardation symptom scores were inversely correlated with frontal operculum acti- vation. Thinking disturbance symptom scores were in- versely correlated with right DLPFC activation. Our findings suggest an association between thinking distur- bance symptoms, particularly unusual thought content, and disrupted WM processing in schizophrenia. © 2001 Academic Press INTRODUCTION Schizophrenia is characterized by broad range of cognitive impairments (Heinrichs et al., 1998). Work- ing memory (WM), the ability to hold and manipulate information online in the brain (Baddeley et al., 1974; Goldman-Rakic, 1994; Smith et al., 1999), is among the most significantly disrupted cognitive functions in schizophrenia (Goldman-Rakic, 1994; 1991; Spindler et al., 1997; Salame et al., 1998; Stone et al., 1998). The component processes involved in WM— encoding, re- hearsal, storage, and executive processes on the con- tents of stored memory—represent key cognitive oper- ations of the human brain. Smith and Jonides (Smith et al., 1999) have argued that analysis of WM is critical for understanding not only memory systems, but thought itself. Goldman-Rakic (1994) has hypothesized that WM dysfunction may be a fundamental feature of formal thought disorder, a predominant positive symp- tom of schizophrenia. Functional and structural neuroimaging in subjects with schizophrenia suggests that cognitive deficits re- sult from prefrontal pathophysiology (Weinberger et al., 1996; Shenton et al., 1997; Nestor et al., 1998; McCarley et al., 1999). Regional cerebral blood flow (rCBF) studies have found evidence for decreased pre- frontal cortex blood flow (“hypofrontality”) in subjects with schizophrenia (Weinberger et al., 1988), with the largest decreases occurring during cognitive tasks in- volving executive function (Young et al., 1998). A num- ber of previous imaging studies of prefrontal cortex deficits in schizophrenia have used neuropsychological tasks that have a WM component (Schroeder et al., 1994; Volz et al., 1997, 1999; Andreasen et al., 1992). Although these studies have found deficits in prefron- tal cortex function in schizophrenia, they have used tasks, such as the Wisconsin Cart Sorting Test, which are generally quite complex and engage a number of cognitive processes that are unrelated to WM per se. More recently, researchers have focused attention on tasks that are generally considered to involve the core operations underlying WM (Carter et al., 1998). These tasks can be generally categorized into two types (1) delayed matching to sample tasks involving WM de- lays of 3–5 s and (2) n-back tasks generally involving shorter delays (Gevins et al., 1993). For example, one study (Stevens et al., 1998) used auditory word and tone recognition in a delayed matching to sample task design and found decreased fMRI activation in subjects with schizophrenia in the lateral frontal cortex and NeuroImage 13, 433– 446 (2001) doi:10.1006/nimg.2000.0699, available online at http://www.idealibrary.com on 433 1053-8119/01 $35.00 Copyright © 2001 by Academic Press All rights of reproduction in any form reserved.