Hypersensitivity to Reward in Problem Gamblers Johannes Hewig, Nora Kretschmer, Ralf H. Trippe, Holger Hecht, Michael G.H. Coles, Clay B. Holroyd, and Wolfgang H.R. Miltner Background: Recent research has begun to examine the neurophysiologic basis of pathological gambling. However, direct evidence of a behavioral deficit and an accompanying neurofunctional deviation in a realistic gambling context such as Black Jack has not yet been reported. Methods: Electroencephalogram was recorded while 20 problem gamblers and 21 control participants played a computerized version of Black Jack. Participants were asked to decide at point scores between 11 and 21 whether they wanted to take another card (“hit”) to arrive closer to 21 than the opponent (simulated by computer) or not to take another card (“sit”) to avoid going over 21 (“bust”). Results: At a critical point score of 16, problem gamblers decided more often to hit despite losses due to a bust on the preceding trial, whereas control participants decided more often to sit under these conditions. Furthermore, problem gamblers showed more reward- related positive amplitudes in the event-related brain potential than control participants after successful hit decisions at 16. Conclusions: Here we provide experimental evidence for high-risk taking behavior in gamblers and its correlate in event-related brain potentials. Our results suggest that high-risk-taking behavior in problem gamblers is associated with an increased reward-related neural response to infrequent successes of this behavior. Key Words: Addiction, anterior cingulate cortex, Black Jack, deci- sion making, problem gambling, risk T he pathological gambler is a high risk taker. Infrequent wins are accompanied by frequent losses, yet the patho- logical gambler persists despite negative consequences. In learning theory terms, such persistence may be attributed either to insensitivity to punishment associated with frequent losses or to hypersensitivity to the reward associated with infrequent wins. Accordingly, it has been suggested that the midbrain dopamine system, as a central reinforcement mechanism, plays an impor- tant role in pathological gambling (1,2,3–5). The event-related brain potential (ERP) between 250 and 350 ms after a feedback stimulus is believed to be sensitive to the arrival of dopamine signals in the anterior cingulate cortex (6). Unexpected punishing events such as negative performance feedback and monetary losses lead to decreases in the activity of midbrain dopamine neurons (7), resulting in a negative potential at the scalp (6,8,9). In contrast, unexpected rewarding events such as positive performance feedback and monetary gains lead to increases in the activity of midbrain dopamine neurons, resulting in relatively positive scalp amplitudes (10). It has recently been proposed that the negative potential represents the default response and that this response is modulated by the positive deflection associated with rewarding events (11). We used these brain potential measures to evaluate the response of 20 problem gamblers and 21 normal control partic- ipants to critical events in a computerized version of Black Jack (12,13). Participants referred to as “problem gamblers” could be classified as “pathological” according to two of the three mea- sures used to define the gambling status (for details, see Supple- ment 1). The players started with a point score between 11 and 21 and then decided whether to “hit,” that is, to take another card (value between 2 and 11) to approach a score of 21, or to “sit,” that, is not to take another card in the hope of beating an opponent with the current score. If the additional card increases the score beyond 21, then the player loses (“busts”). Decisions to hit at increasing scores become increasingly risky as the proba- bility of a bust increases. In previous research, we showed busts elicited more negative ERPs than “no-busts” and that high-risk hit decisions elicited negative ERPs and activated anterior cingulate cortex (12,13). We also showed that control participants hit on 50% of the trials with scores of 16. This behavior is high-risk because, for scores of 15 and higher, taking an additional card is associated with a lower expected value than sitting at the current score. Therefore, we focused on the decision making of problem gamblers and control participants at scores of 16 by computing the probability of hitting or sitting at this score. Furthermore, because gamblers tend to persist in gambling in the face of losses, this analysis evaluated these probabilities as a function of the outcome of the previous trial (bust or no-bust) with the same score. In addition, we evaluated the ERPs following hit decisions at scores of 16 as a function of the outcome (bust or no-bust). Methods and Materials Participants Forty-three male participants were recruited from the student population of the Friedrich-Schiller-Universität and the Univer- sity of Applied Sciences at Jena (21 problem gamblers; mean age: 23.00 years, SD  3.2 years; 22 control participants; mean age: 23.54 years, SD  4.5 years). Participants were initially selected from a group of 529 respondents on the basis of responses to the Short Questionnaire for Gambling (Kurzfragebogen zur Glücksspielsucht, 14). The 30 highest scorers were assigned to the provisional problem gambler group, and the lowest 30 scorers were assigned to the provisional control group. Final selection was based on a diagnostic interview (see Supplement 1 for more details). The problem gamblers (PGs) had to meet the DSM-IV-TR criteria for pathological gambling. Some also fulfilled criteria of specific phobia (three persons), social phobia (four persons), and alcohol abuse (three persons). Control participants From the Institute of Psychology (JH, NK, RHT, HH, WHRM), Friedrich- Schiller-University Jena, Jena, Germany; Donders Centre for Cognitive Neuroimaging (MGHC), Nijmegen, The Netherlands; and Department of Psychology (CBH), University of Victoria, Victoria, Canada. Address correspondence to Johannes Hewig, Dr. rer. nat. Dipl.-Psych., Insti- tut für Psychologie, Friedrich-Schiller-Universität Jena, Am Steiger 3, Haus 1, D-07743 Jena; E-mail: hewig@biopsy.uni-jena.de. Received Aug 27, 2009; revised Oct 15, 2009; accepted Nov 7, 2009. BIOL PSYCHIATRY 2010;67:781–783 0006-3223/10/$36.00 doi:10.1016/j.biopsych.2009.11.009 © 2010 Society of Biological Psychiatry