Using negative feedback to guide behavior: Impairments on the first 4 cards of the Wisconsin Card Sorting Test predict negative symptoms of schizophrenia Sally J. Vogel a , Gregory P. Strauss a,b , Daniel N. Allen a, ⁎ a University of Nevada, Las Vegas, Department of Psychology, United States b University of Maryland School of Medicine, Department of Psychiatry and Maryland Psychiatric Research Center, Baltimore, MD, United States abstract article info Article history: Received 22 January 2013 Received in revised form 26 July 2013 Accepted 29 July 2013 Available online 6 November 2013 Keywords: Reward Reinforcement learning Decision-making Avolition Anhedonia Schizophrenia Deficit syndrome Research has demonstrated that individuals with schizophrenia fail to appropriately use negative feedback to guide learning. These learning deficits are thought to arise from abnormalities in midbrain dopamine activity. Primary and enduring negative symptoms are also associated with abnormal dopamine activity and are expected to produce more severe deficits in learning when they present in individuals with schizophrenia. The current study examines this matter by comparing individuals with deficit syndrome schizophrenia, which is characterized by primary and enduring negative symptoms, to individuals with nondeficit syndrome schizophre- nia and to normal controls in their use of positive feedback and negative feedback to guide learning on the first four cards of the WCST. Participants included 67 individuals with schizophrenia (15 deficit; 52 nondeficit syn- drome) and 51 healthy controls. Accuracy data from the first 4 cards of the WCST and measures of global test performance were examined. Individuals with schizophrenia were significantly less accurate than controls in their performance on early (pre-shift) WCST trials, and this impairment was significantly greater in patients with deficit than nondeficit schizophrenia. Additionally, accuracy across the first 4 WCST cards significantly predicted the number of categories completed and percentage of perseverative errors across the entire test. These findings suggest that negative symptoms of schizophrenia are associated with difficulty using negative feedback to adaptively guide behavior, and are consistent with the notion that abnormal DA signaling contributes to the higher-order executive functioning impairments seen in schizophrenia with severe negative symptoms. © 2013 Elsevier B.V. All rights reserved. 1. Introduction Individuals with schizophrenia typically perform poorly on the Wisconsin Card Sorting Test (WCST), completing fewer categories and showing a higher rate of perseverative errors (Van der Does and Van den Bosch, 1992; Koren et al., 1998), which may reflect difficulty abandoning a previously-correct response after receiving negative feed- back. Prentice et al. (2008) took a novel approach to investigating WCST performance by analyzing data from the first four WCST trials to exam- ine whether patients had greater deficits than controls in using rapid, trial-to-trial feedback to guide behavior. They found that individuals with schizophrenia showed poorer performance on cards 2–4 than con- trols and that performance on these trials predicted overall WCST success as indexed by the number of categories completed and percentage of perseverative errors. Poor performance on these pre-shift WCST trials suggests a basic impairment in the ability to use negative feedback to guide behavior. Prentice et al. (2008) suggest that the initial trials of the WCST can be conceptualized within a temporal difference error (TDE) reward learning framework (Schultz, 2002; Montague et al., 2004). TDE models propose that the outcomes of actions drive learning, so that actions that result in better than expected outcomes cause positive TDE signals that increase the likelihood of repeating the action, whereas actions that produce worse than expected outcomes result in negative TDE signals which decrease the likelihood of repeating the action (Hollerman and Schultz, 1998; Schultz, 2002; Montague et al., 2004). Dopaminergic signaling appears critical in TDE learning, as there is evidence that in primates transient increases and decreases in DA cell activity have been associated with the coding of positive and negative prediction errors (Hollerman and Schultz, 1998), and in humans positive and negative TDEs have been associated with increased activation of DA projection areas (Monchi et al., 2001; Ullsperger and von Cramon, 2003; Aron et al., 2004). As demonstrated by Prentice et al. (2008), it is possible to use behavior on the WCST first 4 cards to investigate reinforcement learning in a TDE framework, where repetitions of reinforced responses indicate learning from positive feedback and abandoning incorrect responses in favor of alternative actions reflects learning from negative outcomes. Given that individuals with schizo- phrenia display both reinforcement learning deficits (Juckel et al., 2006; Barch and Dowd, 2010) and abnormal DA function (Kapur et al., 2005), there is reason to expect that TDE learning can explain the types of reward-based impairments that occur in schizophrenia. Schizophrenia Research 151 (2013) 97–101 ⁎ Corresponding author at: Department of Psychology, University of Nevada, Las Vegas, Las Vegas, NV 89154, United States. Tel.: +1 702 895 0121; fax: +1 702 895 0195. E-mail address: daniel.allen@unlv.edu (D.N. Allen). 0920-9964/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.schres.2013.07.052 Contents lists available at ScienceDirect Schizophrenia Research journal homepage: www.elsevier.com/locate/schres