- Email: [email protected]
THE NEURAL BASIS OF ASSOCIATIVE EMOTIONAL LEARNING IN SCHIZOPHRENIA AND THE RELATIONSHIP WITH ALEXITHYMIA
Abstracts
Methods: Twenty-four psychotic patients indicated the presence of AVH during 3T fMRI scanning by squeezing a hand-held balloon. A one sample T-test was performed to reveal group-wise activation during AVH. To enable analysis of brain activation 6 to 0 seconds preceding AVH a tailored 'selective averaging' method without any a priori assumptions concerning the haemodynamic response profile was performed. To control for motor related activation, fifteen control subjects squeezed a balloon at matched time intervals. Results: Group-wise analysis during AVH revealed brain activation in bilateral, right more than left, language-related regions and bilateral motor-regions. Prominent deactivation preceding AVH was observed in the left parahippocampal gyrus. In addition, significant deactivation was found in the left superior temporal, right inferior frontal and left middle frontal gyri as well as in the right insula and left cerebellum preceding AVH. No significant signal changes were revealed prior to the matched balloon-squeezing in the control subjects. Discussion: Auditory verbal hallucinations in psychotic patients are consistently preceded by deactivation of the parahippocampal gyrus. The parahippocampus has been hypothesized to play a central role in memory recollection; sending recognized information from the hippocampus to the association areas. Dysfunction of this region could trigger inadequate activation of right language areas during AVH.
471
effect). Across 15 HCs at baseline, the largest region that demonstrated this self-referential effect was dMPFC. Furthermore, as dMPFC activity increased across 15 HCs the more accurate they were at correctly identifying self-generated information (r = .48, p = .03). In contrast, at baseline, all SZs showed deactivation in bilateral frontal regions. Paired t-tests indicate that after 16 weeks of computer-games compared to baseline, CG patients showed increased activation only in bilateral occipital gyri (but not in dMPFC) during self-referential processing, and revealed no change on identifying either more self-generated items (t = .436, p = .66) or more externally-derived items (t = .423, p = .674), compared to baseline. In contrast, after 16 weeks of TCT exercises compared to baseline, TCT patients showed increased activation in the same dMPFC region that the HCs revealed during self-referential processing, and also showed significant improvement on correctly identifying more self-generated items (t = 2.026, p = .049), as well as more externally-derived items (t = 3.367, p = .002) compared to baseline. Discussion: These fMRI results indicate a possible "restorative" effect of targeted-cognitive-training in schizophrenia patients, not observed in computer-games control patients, whereby behavioral performance on a self-referential reality-monitoring source memory task is improved and brain activation patterns are "normalized," similar to the healthy comparison subjects.
doi:10.1016/j.schres.2010.02.885 doi:10.1016/j.schres.2010.02.886
Poster 125 NEUROPLASTICITY-BASED COGNITIVE TRAINING IMPROVES REALITY MONITORING IN SCHIZOPHRENIA PATIENTS: BEHAVIORAL AND FMRI ASSESSMENTS
Karuna Subramaniam, Tracy Luks, Stephanie Aldebot, Adelaide Hearst, Arul Thangavel, Melissa Fisher, Gregory V. Simpson, Srikantan Nagarajan, Sophia Vinogradov University of California San Francisco San Francisco, CA, USA
Background: Prior research indicates that schizophrenia patients (SZs) are impaired at identifying themselves as the source of selfgenerated information (reality monitoring). They also show relatively decreased activation within the dorsal medial prefrontal cortex (dMPFC) compared to healthy comparison subjects (HCs) when engaged in this process (Vinogradov et al., 2008). In the present study, we investigated whether this deficit is amenable to a behavioral intervention. Methods: Thirty one SZs and 15 HCs underwent an fMRI sourcememory task at baseline. Fourteen SZs were then randomly assigned to 80 hours of computerized targeted-cognitive-training (TCT) that focused on training auditory, visual and social cognitive processes. The other fourteen SZs were assigned to a control condition of 80 hours of computer-games (CGs). All subjects repeated the task after the 16 week intervention period. Prior to scanning, subjects were presented with sentences, where the final target word was either supplied by the experimenter, or left blank for subjects to generate themselves. During scanning, subjects were presented with target words, and decided whether they were experimenter-presented or self-generated. BOLD fMRI activity was measured on a 3T GE scanner (EPI; TR = 1 sec, 14 slices) before and after the 16 week intervention. Images were analyzed using SPM2. Results: At baseline, SZs revealed significantly more impairments, compared to HCs, while recalling self-generated information (p < .0005). Whole-brain analyses focused on brain regions showing greater activation for correctly remembered self-generated items versus externally presented items (i.e., a self-referential
Poster 126 THE NEURAL BASIS OF ASSOCIATIVE EMOTIONAL LEARNING IN SCHIZOPHRENIA AND THE RELATIONSHIP WITH ALEXITHYMIA Marte Swart1, Richard Bruggeman2, Rudie Kortekaas1, Durk Wiersma2, André Aleman1 1 Department of Neuroscience, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands; 2Department of Psychiatry, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands Background: Schizophrenia is characterized by deficits in emotional processing and regulation (Aleman & Kahn, 2005). Verbalizing of emotions is an important aspect of emotion regulation. To verbalize emotions it is crucial to make associations between feelings and words, which involves emotional memory. We investigated whether schizophrenia patients and controls show different brain activation during associative emotional learning. We also examined whether this was associated with alexithymia (or "no words for feelings"). We expected increased limbic activation in schizophrenia patients during emotional memory. Additionally, we expected schizophrenia patients to report lower verbalizing ability than the controls. Methods: Seventeen schizophrenia patients and seventeen controls (matched on age, gender and education) performed an associative emotional learning task with emotional and neutral picture-word pairs during fMRI scanning. Participants had to indicate if the picture and word were associated and had to remember the pairs. After scanning they were tested for their memory of the pictureword pairs. They also filled in the Bermond-Vorst Alexithymia Questionnaire (BVAQ). Functional images were acquired with a 3T Philips scanner using echo-planar imaging (EPI). We created contrast images for each participant for the emotional vs. neutral picture-word pairs. We used t tests for group comparisons between patients and controls (SPM5). Results: Patients demonstrated increased activation in bilateral superior temporal gyrus (STG), left amygdala/hippocampus (amy/
472
Abstracts
hipp) and in left cingulate gyrus (CG) in emotional compared to neutral picture-word pairs. Patients had a higher score on the cognitive component (i.e. were worse in verbalizing, analyzing and identifying emotions) of the BVAQ whereas on the emotional component, groups scored similarly. Patients remembered less picture-word pairs than controls. Discussion: The areas of increased activation in patients have been related to emotional learning (amy/hipp), reduced emotion regulation (amy) and theory of mind (STG) during learning and associating emotional picture-word pairs. We suggest that the increases in activation reflect increased recruitment of neural resources to counter the difficulties that patients have in performing this task. This is consistent with the behavioral and questionnaire results in which patients reported more difficulties in verbalizing, analyzing and identifying their emotions and the fact that they remembered less pairs during the memory test after scanning. doi:10.1016/j.schres.2010.02.887
Poster 127 INTERACTION BETWEEN DARPP-32 AND DRD2 GENETIC VARIANTS ON ANTERIOR CINGULATE CORTEX ACTIVITY DURING ATTENTIONAL CONTROL IN HEALTHY SUBJECTS Paolo Taurisano1, Giuseppe Blasi1, Apostolos Papazacharias1, Raffaella Romano1, Gianluca Ursini1, Annabella Di Giorgio1, Leonardo Fazio1, Linda Antonucci1, Barbara Gelao1, Grazia Caforio1, Rita Masellis1, Luciana Lo Bianco1, Lorenzo Sinibaldi2,3, Teresa Popolizio4, Wolfgang Sadée5,6, Alessandro Bertolino1 1 Depatment of Neurologic and Psychiatric Science, University of Bari Bari, Bari, Italy; 2IRCCS-CSS Mendel Roma, Roma, Italy; 3Department of Sperimental Medicine, University “La Sapienza Roma, Roma, Italy; 4IRCCS Casa Sollievo della Sofferenza S.Giovanni Rotondo, Foggia, Italy; 5Department of Psychiatry, Division of Human Genetics, College of Medicine, and College of Pharmacy Columbus, Ohio, USA; 6Division of Biostatistics, College of Public Health, Ohio State University Columbus, Ohio, USA Background: Previous studies have revealed that dopamine D2 receptors are involved in modulation of cognitive processing. Other studies have also indicated that DARPP-32, a dual-function protein Kinase/phosphatase, crucially mediates effects of D2 stimulation. Genetic variation in DRD2 (encoding D2) and DARPP-32 is known to be functional. In particular, an intronic single nucleotide polymorphism (SNP) within the D2 receptor gene (DRD2, rs1076560, G > T) shifts splicing of the two protein isoforms (D2 short, mainly presynaptic, and D2 long), affecting D2short/D2 long ratio, and it has been associated with modulation of working memory performance and related brain activity in patients with schizophrenia. Furthermore, a SNP in the DARPP-32 gene (also called PPP1R1B, rs907094 A > G) has been shown to modulate cognitive behavior in humans, and it has been associated with DARPP-32 post-mortem mRNA expression, prefrontostriatal activity and diagnosis of schizophrenia. Aim of this study is to investigate effects of DRD2 rs1076560 and DARPP-32 rs907094 on cingulate activity during attentional processing, a phenotype associated with dopamine signaling and schizophrenia. Methods: 128 healthy subjects (44M, age 25 ± 5.8) were genotyped for DRD2 rs1076560 and DARPP-32 rs907094 (N=59 GG/AA; 61 GG/ Gcarriers; 10 GT/AA; 15 GT/Gcarriers) and performed a task eliciting increasing levels of attentional control (Variable Attentional Control TaskVAC) during event-related functional magnetic resonance imaging at 3 T. Results: No genotypes effects on behavioral data were found. On the other hand, fMRI data indicated an interaction between DRD2 rs1076560, DARPP-32 rs907094 and attentional load on Anterior Cingulate Cortex (ACC) activity (p< 0.05, FDR corrected). Analysis of
the pattern of response indicated linear increases of activity in ACC as a function of the increase in attentional load for all groups with the exception of DRD2 GT/ DARP-32 AA subjects. In this latter group there was a non-linear relationship between activity and attentional control load suggestive of an inverted "U" shaped response. Discussion: These findings suggest a genetically determined interaction of DARPP-32 and D2 in modulating ACC activity during attentional control. Further studies should address how these genetically mediated effects may be relevant to the pathophysiology of schizophrenia. doi:10.1016/j.schres.2010.02.888
Poster 128 BRAIN ACTIVATION DURING THE N-BACK TASK AS A PREDICTOR OF FUTURE TREATMENT RESPONSE IN FIRST EPISODE PSYCHOSIS PATIENTS Heather Taylor1, Tiago Reis Marques1, Andy Simmons3, A.A.T. Simone Reinders1,2, Rowena Handley1, Valeria Mondelli1, Sara Pozzoli1, Marta Di Forti1, Anthony David1, Robin M. Murray1, Carmine M. Pariante1, Paola Dazzan1 1 Psychosis Clinical Academic Group, Insititute of Psychiatry, Kings Health Partners, Kings College London London United Kingdom; 2BCN Neuroimaging Centre, Department of Neuroscience, Uniiversity of Groningen Groningen Netherlands; 3Centre for Neuroimaging Sciences, Institute of Psychiatry, Kings college London London United Kingdom Background: Treatment of psychotic disorders is currently conducted on a trial and error basis with antipsychotic medication. However, approximately a third of patients will show no response. Previous research has shown that psychosis is associated with abnormalities in brain function, already at the early stages of illness, for example during the execution of working memory tasks. It has been proposed that a dysfunction in dopamine transmission may underlie some of these cognitive impairments. It remains unclear if brain functional alterations that accompany the performance of these cognitive tasks characterise a subgroup of individuals with a poor response to treatment with antipsychotics. The aim of this study was to explore the relationship between brain functional activation during a working memory task at onset of psychosis and treatment response at 3 months. Methods: 20 First episode psychosis patients were recruited at their first contact with services. All patients had less than 12 weeks of antipsychotic medication. They were scanned using a 3.0 Tesla scanner. 186 volumes were acquired with a slice thickness of 3.5 mm over 6 minutes and 40 seconds during the execution of the N-Back working memory task. This task consists of 4 parts – 0, 1, 2 and 3 back (0 back being the control condition, and 3 back the hardest condition). Clinical (PANSS and CGI) data was collected at MRI and again 12 weeks later. According to change in PANSS and CGI scores patients were categorised as responders (n = 8) or non-responders (n = 12). Results: Behavioural data were investigated across all 4 N-Back conditions. There was no significant difference between responders and non-responders on the 0, 1, and 3 back. On the 2 back condition, responders showed quicker response times and a higher percentage of correct responses than non-responders. The functional MRI data was then analysed using SPM5. Applying a whole brain correction (p=0.001), there was an increased activation in responders in an area centred on the substantia nigra only in the 2 back vs control condition. Since this was the only contrast that showed also a behavioural difference, brain activation was analysed further for this contrast. Multiple corrections were performed on preselected brain regions of interest. While there was no difference in activation between responders and non-responders in frontal areas, responders showed an increased
Methods: Twenty-four psychotic patients indicated the presence of AVH during 3T fMRI scanning by squeezing a hand-held balloon. A one sample T-test was performed to reveal group-wise activation during AVH. To enable analysis of brain activation 6 to 0 seconds preceding AVH a tailored 'selective averaging' method without any a priori assumptions concerning the haemodynamic response profile was performed. To control for motor related activation, fifteen control subjects squeezed a balloon at matched time intervals. Results: Group-wise analysis during AVH revealed brain activation in bilateral, right more than left, language-related regions and bilateral motor-regions. Prominent deactivation preceding AVH was observed in the left parahippocampal gyrus. In addition, significant deactivation was found in the left superior temporal, right inferior frontal and left middle frontal gyri as well as in the right insula and left cerebellum preceding AVH. No significant signal changes were revealed prior to the matched balloon-squeezing in the control subjects. Discussion: Auditory verbal hallucinations in psychotic patients are consistently preceded by deactivation of the parahippocampal gyrus. The parahippocampus has been hypothesized to play a central role in memory recollection; sending recognized information from the hippocampus to the association areas. Dysfunction of this region could trigger inadequate activation of right language areas during AVH.
471
effect). Across 15 HCs at baseline, the largest region that demonstrated this self-referential effect was dMPFC. Furthermore, as dMPFC activity increased across 15 HCs the more accurate they were at correctly identifying self-generated information (r = .48, p = .03). In contrast, at baseline, all SZs showed deactivation in bilateral frontal regions. Paired t-tests indicate that after 16 weeks of computer-games compared to baseline, CG patients showed increased activation only in bilateral occipital gyri (but not in dMPFC) during self-referential processing, and revealed no change on identifying either more self-generated items (t = .436, p = .66) or more externally-derived items (t = .423, p = .674), compared to baseline. In contrast, after 16 weeks of TCT exercises compared to baseline, TCT patients showed increased activation in the same dMPFC region that the HCs revealed during self-referential processing, and also showed significant improvement on correctly identifying more self-generated items (t = 2.026, p = .049), as well as more externally-derived items (t = 3.367, p = .002) compared to baseline. Discussion: These fMRI results indicate a possible "restorative" effect of targeted-cognitive-training in schizophrenia patients, not observed in computer-games control patients, whereby behavioral performance on a self-referential reality-monitoring source memory task is improved and brain activation patterns are "normalized," similar to the healthy comparison subjects.
doi:10.1016/j.schres.2010.02.885 doi:10.1016/j.schres.2010.02.886
Poster 125 NEUROPLASTICITY-BASED COGNITIVE TRAINING IMPROVES REALITY MONITORING IN SCHIZOPHRENIA PATIENTS: BEHAVIORAL AND FMRI ASSESSMENTS
Karuna Subramaniam, Tracy Luks, Stephanie Aldebot, Adelaide Hearst, Arul Thangavel, Melissa Fisher, Gregory V. Simpson, Srikantan Nagarajan, Sophia Vinogradov University of California San Francisco San Francisco, CA, USA
Background: Prior research indicates that schizophrenia patients (SZs) are impaired at identifying themselves as the source of selfgenerated information (reality monitoring). They also show relatively decreased activation within the dorsal medial prefrontal cortex (dMPFC) compared to healthy comparison subjects (HCs) when engaged in this process (Vinogradov et al., 2008). In the present study, we investigated whether this deficit is amenable to a behavioral intervention. Methods: Thirty one SZs and 15 HCs underwent an fMRI sourcememory task at baseline. Fourteen SZs were then randomly assigned to 80 hours of computerized targeted-cognitive-training (TCT) that focused on training auditory, visual and social cognitive processes. The other fourteen SZs were assigned to a control condition of 80 hours of computer-games (CGs). All subjects repeated the task after the 16 week intervention period. Prior to scanning, subjects were presented with sentences, where the final target word was either supplied by the experimenter, or left blank for subjects to generate themselves. During scanning, subjects were presented with target words, and decided whether they were experimenter-presented or self-generated. BOLD fMRI activity was measured on a 3T GE scanner (EPI; TR = 1 sec, 14 slices) before and after the 16 week intervention. Images were analyzed using SPM2. Results: At baseline, SZs revealed significantly more impairments, compared to HCs, while recalling self-generated information (p < .0005). Whole-brain analyses focused on brain regions showing greater activation for correctly remembered self-generated items versus externally presented items (i.e., a self-referential
Poster 126 THE NEURAL BASIS OF ASSOCIATIVE EMOTIONAL LEARNING IN SCHIZOPHRENIA AND THE RELATIONSHIP WITH ALEXITHYMIA Marte Swart1, Richard Bruggeman2, Rudie Kortekaas1, Durk Wiersma2, André Aleman1 1 Department of Neuroscience, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands; 2Department of Psychiatry, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands Background: Schizophrenia is characterized by deficits in emotional processing and regulation (Aleman & Kahn, 2005). Verbalizing of emotions is an important aspect of emotion regulation. To verbalize emotions it is crucial to make associations between feelings and words, which involves emotional memory. We investigated whether schizophrenia patients and controls show different brain activation during associative emotional learning. We also examined whether this was associated with alexithymia (or "no words for feelings"). We expected increased limbic activation in schizophrenia patients during emotional memory. Additionally, we expected schizophrenia patients to report lower verbalizing ability than the controls. Methods: Seventeen schizophrenia patients and seventeen controls (matched on age, gender and education) performed an associative emotional learning task with emotional and neutral picture-word pairs during fMRI scanning. Participants had to indicate if the picture and word were associated and had to remember the pairs. After scanning they were tested for their memory of the pictureword pairs. They also filled in the Bermond-Vorst Alexithymia Questionnaire (BVAQ). Functional images were acquired with a 3T Philips scanner using echo-planar imaging (EPI). We created contrast images for each participant for the emotional vs. neutral picture-word pairs. We used t tests for group comparisons between patients and controls (SPM5). Results: Patients demonstrated increased activation in bilateral superior temporal gyrus (STG), left amygdala/hippocampus (amy/
472
Abstracts
hipp) and in left cingulate gyrus (CG) in emotional compared to neutral picture-word pairs. Patients had a higher score on the cognitive component (i.e. were worse in verbalizing, analyzing and identifying emotions) of the BVAQ whereas on the emotional component, groups scored similarly. Patients remembered less picture-word pairs than controls. Discussion: The areas of increased activation in patients have been related to emotional learning (amy/hipp), reduced emotion regulation (amy) and theory of mind (STG) during learning and associating emotional picture-word pairs. We suggest that the increases in activation reflect increased recruitment of neural resources to counter the difficulties that patients have in performing this task. This is consistent with the behavioral and questionnaire results in which patients reported more difficulties in verbalizing, analyzing and identifying their emotions and the fact that they remembered less pairs during the memory test after scanning. doi:10.1016/j.schres.2010.02.887
Poster 127 INTERACTION BETWEEN DARPP-32 AND DRD2 GENETIC VARIANTS ON ANTERIOR CINGULATE CORTEX ACTIVITY DURING ATTENTIONAL CONTROL IN HEALTHY SUBJECTS Paolo Taurisano1, Giuseppe Blasi1, Apostolos Papazacharias1, Raffaella Romano1, Gianluca Ursini1, Annabella Di Giorgio1, Leonardo Fazio1, Linda Antonucci1, Barbara Gelao1, Grazia Caforio1, Rita Masellis1, Luciana Lo Bianco1, Lorenzo Sinibaldi2,3, Teresa Popolizio4, Wolfgang Sadée5,6, Alessandro Bertolino1 1 Depatment of Neurologic and Psychiatric Science, University of Bari Bari, Bari, Italy; 2IRCCS-CSS Mendel Roma, Roma, Italy; 3Department of Sperimental Medicine, University “La Sapienza Roma, Roma, Italy; 4IRCCS Casa Sollievo della Sofferenza S.Giovanni Rotondo, Foggia, Italy; 5Department of Psychiatry, Division of Human Genetics, College of Medicine, and College of Pharmacy Columbus, Ohio, USA; 6Division of Biostatistics, College of Public Health, Ohio State University Columbus, Ohio, USA Background: Previous studies have revealed that dopamine D2 receptors are involved in modulation of cognitive processing. Other studies have also indicated that DARPP-32, a dual-function protein Kinase/phosphatase, crucially mediates effects of D2 stimulation. Genetic variation in DRD2 (encoding D2) and DARPP-32 is known to be functional. In particular, an intronic single nucleotide polymorphism (SNP) within the D2 receptor gene (DRD2, rs1076560, G > T) shifts splicing of the two protein isoforms (D2 short, mainly presynaptic, and D2 long), affecting D2short/D2 long ratio, and it has been associated with modulation of working memory performance and related brain activity in patients with schizophrenia. Furthermore, a SNP in the DARPP-32 gene (also called PPP1R1B, rs907094 A > G) has been shown to modulate cognitive behavior in humans, and it has been associated with DARPP-32 post-mortem mRNA expression, prefrontostriatal activity and diagnosis of schizophrenia. Aim of this study is to investigate effects of DRD2 rs1076560 and DARPP-32 rs907094 on cingulate activity during attentional processing, a phenotype associated with dopamine signaling and schizophrenia. Methods: 128 healthy subjects (44M, age 25 ± 5.8) were genotyped for DRD2 rs1076560 and DARPP-32 rs907094 (N=59 GG/AA; 61 GG/ Gcarriers; 10 GT/AA; 15 GT/Gcarriers) and performed a task eliciting increasing levels of attentional control (Variable Attentional Control TaskVAC) during event-related functional magnetic resonance imaging at 3 T. Results: No genotypes effects on behavioral data were found. On the other hand, fMRI data indicated an interaction between DRD2 rs1076560, DARPP-32 rs907094 and attentional load on Anterior Cingulate Cortex (ACC) activity (p< 0.05, FDR corrected). Analysis of
the pattern of response indicated linear increases of activity in ACC as a function of the increase in attentional load for all groups with the exception of DRD2 GT/ DARP-32 AA subjects. In this latter group there was a non-linear relationship between activity and attentional control load suggestive of an inverted "U" shaped response. Discussion: These findings suggest a genetically determined interaction of DARPP-32 and D2 in modulating ACC activity during attentional control. Further studies should address how these genetically mediated effects may be relevant to the pathophysiology of schizophrenia. doi:10.1016/j.schres.2010.02.888
Poster 128 BRAIN ACTIVATION DURING THE N-BACK TASK AS A PREDICTOR OF FUTURE TREATMENT RESPONSE IN FIRST EPISODE PSYCHOSIS PATIENTS Heather Taylor1, Tiago Reis Marques1, Andy Simmons3, A.A.T. Simone Reinders1,2, Rowena Handley1, Valeria Mondelli1, Sara Pozzoli1, Marta Di Forti1, Anthony David1, Robin M. Murray1, Carmine M. Pariante1, Paola Dazzan1 1 Psychosis Clinical Academic Group, Insititute of Psychiatry, Kings Health Partners, Kings College London London United Kingdom; 2BCN Neuroimaging Centre, Department of Neuroscience, Uniiversity of Groningen Groningen Netherlands; 3Centre for Neuroimaging Sciences, Institute of Psychiatry, Kings college London London United Kingdom Background: Treatment of psychotic disorders is currently conducted on a trial and error basis with antipsychotic medication. However, approximately a third of patients will show no response. Previous research has shown that psychosis is associated with abnormalities in brain function, already at the early stages of illness, for example during the execution of working memory tasks. It has been proposed that a dysfunction in dopamine transmission may underlie some of these cognitive impairments. It remains unclear if brain functional alterations that accompany the performance of these cognitive tasks characterise a subgroup of individuals with a poor response to treatment with antipsychotics. The aim of this study was to explore the relationship between brain functional activation during a working memory task at onset of psychosis and treatment response at 3 months. Methods: 20 First episode psychosis patients were recruited at their first contact with services. All patients had less than 12 weeks of antipsychotic medication. They were scanned using a 3.0 Tesla scanner. 186 volumes were acquired with a slice thickness of 3.5 mm over 6 minutes and 40 seconds during the execution of the N-Back working memory task. This task consists of 4 parts – 0, 1, 2 and 3 back (0 back being the control condition, and 3 back the hardest condition). Clinical (PANSS and CGI) data was collected at MRI and again 12 weeks later. According to change in PANSS and CGI scores patients were categorised as responders (n = 8) or non-responders (n = 12). Results: Behavioural data were investigated across all 4 N-Back conditions. There was no significant difference between responders and non-responders on the 0, 1, and 3 back. On the 2 back condition, responders showed quicker response times and a higher percentage of correct responses than non-responders. The functional MRI data was then analysed using SPM5. Applying a whole brain correction (p=0.001), there was an increased activation in responders in an area centred on the substantia nigra only in the 2 back vs control condition. Since this was the only contrast that showed also a behavioural difference, brain activation was analysed further for this contrast. Multiple corrections were performed on preselected brain regions of interest. While there was no difference in activation between responders and non-responders in frontal areas, responders showed an increased