14. Neuroimaging, Functional 221 resultsconfirm earlier findings of a deficient frontal tobe function in schizophrenia and support the hypothesis that this hypofiontality may be ameliorated by cognitive training [2]. However, performance gains were not only associated with a quantitative shift within the working memory network, but also with qualitative activation changes with involvement of regions initially not activated. Hence, it should be emphasized that patients did not reach normalization of task-related cerebral activation patterns. References: [1] Hempel et a1.(2002). Learning effects on cerebral activation in a functional MRI working memory paradigm. Scliizophr Res 53,110. [2] Wykes et al. (2002). Effects on brain of psychological treatment: cognitive reme- diation therapy. Br J Psychiat 181,144-152. META-ANALYSIS FAILS TO REVEAL EVIDENCE OF ACTIVATION HYPOFRONTALITY IN SCHIZOPHRENIA K. E. Hill,* L. Mann, E J. McKenna, C. Stephenson, K. Laws, I. Nimmo-Smith Fulborn Hospital, Cambridge, United Kingdom The concept of hypofrontality has been central to fnnctional imag- ing research in schizophrenia for nearly 20 years. This has been found in some studies under resting conditions but more consistent- ly in activation studies using executive, vigilance and memory tasks. Conventional meta-analysis of 22 studies using conventional 'region of interest' analysis to compare schizophrenic patients and normal subjects yielded an effect size of- 0.30 (CI - 0.45 to- 0.15), indi- cating significant activation hypofrontality. However, such a between subjects analysis is confounded by lower resting levels of prefrontal flow/metabolism in schizophrenic patients (effect size from 39 homogeneous studies: - 0.42, CI - 0.52/- 0.32). Within subjects analysis of 13 studies reporting both resting and activation data yield- ed an effect size for activation in schizophrenic patients of 0.30 (CI +0.09/+0.51) which was greater than that for activation in controls (ES = 0.18, CI - 0.05/+0.41) The novel technique devised by Dun- can and Owen can be used for combining data across voxel-based studies. This examines homogeneity of activated regions using a 3 dimensional version of the Kolmogorov-Smirnov test. Applying this technique to 8 studies comparing schizophrenic patients and normal controls revealed no significant difference for prefrontal (KS3 = 0.20, P = 0.97) or whole brain distributions of activations (KS3 = 0.14, P -- 0.96), and no difference in numbers of reported prefrontal activa- tions (Chi squared = 0.33, P=0.57). Although based on pooled data from small numbers of studies, these findings provide little to sup- port the widely accepted view that there is an altered pattern of func- tional prefrontal architecture in schizophrenia. THE NEURAL CORRELATES OF EPISODIC ENCODING AND RECOGNITION OF WORDS IN UNMEDICATED PATIENTS WITH ACUTE EPISODE OF SCHIZOPHRENIA: A FUNCTIONAL MRI STUDY A. Hofer,* E. M. Weiss, S. M. Golaszewski, C. M. Siedentopf, C. Brinkhoff, C. Kremser, S. Felber, W. W. Fleischhacker Dept. of Biological Psychiatry, lnnsbruck University Itospital, Innsbruck, Austria Memory impairment has been welt documented in schizophrenia. We have previously investigated patterns of brain activity during episodic encoding and recognition of words in remitted, stable out- patients, treated with novel antipsychotics, and therefore not pre- senting with the original dysfunctions of untreated schizophrenia. Using the same procedure we now investigated unmedicated patients with acute episode of schizophrenia. We used functional magnetic resonance imaging (tMRI) to study regional brain activation in ten healthy comparison subjects and ten patients with acute episode of schizophrenia during the performance of a modified version of War- rington Recognition Memory Test for Words. Despite intact recog- nition performance, patients with schizophrenia showed reduced activation of anterior prefrontal, posterior cingulate, retrosplenial, and posterior parietal cortices during word encoding. During word recognition, activation impairments of patients were found in dor- solateral prefrontal and limbic/paralimbic regions. Instead, a higher metabolism in bilateral anterior prefrontal cortices was observed. Acute psychosis may prevent training effects in patients with schiz- ophrenia, reflected by a failure to engage brain regions which have been associated with successful episodic memory retrieval in healthy people. However, relatively small time periods between encoding and recognition may enable the patients to maintain the encoded information in working memory and to engage retrieval processes, leading to intact recognition performance. VISUAL ENCODING DYSFUNCTION IN SCHIZOPHRENIC PATIENTS BEFORE AND AFTER LEARNING, AN EVENT RELATED FMRI STUDY H. H. Holcomb,* L. Beason-Held, M. Tagamets, T. Cullen, C. A. Tamminga Functional Neuroimaging Laboratory, Maryland Psychiatric Research Centel, Bahimore, MD, USA When making perceptual judgments, patients with schizophrenia, SZ, routinely fail to perform with the precision or appropriate diffi- culty-related changes in response time observed in normal healthy volunteers, NV. We have used event related fMRI to assess how SZ and NV encode a visual target in a visual delayed -match-to-sample- task, DMST. Nine NV (mean age 28) and ten medicated outpatient SZ (mean age 35) participated in this study. The DMST required the subject to make a judgment, indicated by button press in fight or left hand, as to whether two consecutively presented black rectangles (separated by 6 seconds) were the same height. Control trials always had identical stimuli in which the black rectangle was clearly dis- placed to either the right or left side of the frame containing the black rectangle. Control trials only required the subject to match his hand button with the side to which the rectangle was displaced. These data only concern trial sets that elicited a relatively high error rate, 25-35% errors. Functional MRI data were acquired using echoplanar proto- cols with a TR of 2 seconds. Four sets of high error scans were obtained before and alter a rigorous 2 week training period. Normal and Schizophrenic subjects exhibited similar improvements in speed and accuracy with learning. Functional MRI results were analyzed with Statistical Parametric Mapping software (SPM99) using a ran- dom effects model. Pro-training Task Encode signals for NV were significant for parietal and caudate regions. NV revealed no positive BOLD response for Control Encode measures. Post-training Task Encode signals in NV were significant in hippocampal, thalamic, insula, and medial frontal / cingulate regions. Pro-training Task Encode responses for the SZ group were confined to the parietal cor- tex. Identical responses were found in SZ Pro-training Encode Con- trol trials. Post-training Task Encode responses for SZ participants International Congress on Schizophrenia Research 2003