The neuropsychology of ventral prefrontal cortex: Decision-making and reversal learningIntroductionThe cognitive neuroscience of decision-makingHuman lesion studiesFunctional neuroimaging studiesAnimal lesion studiesPsychopharmacological studiesSpecificity of decision-making effects to ventral prefrontal cortexThe cognitive neuroscience of reversal learningAnimal lesion studiesHuman lesion studiesFunctional neuroimaging studiesPsychopharmacological studiesThe development of decision-making and reversal learningSynthesisAcknowledgementsReferencesThe neuropsychology of ventral prefrontal cortex:Decision-making and reversal learningL. Clark,*R. Cools, and T.W. RobbinsDepartment of Experimental Psychology, University of Cambridge, Cambridge CB2 3EB, UKAccepted 14 April 2003Available online 4 February 2004AbstractConverging evidence from human lesion, animal lesion, and human functional neuroimaging studies implicates overlappingneural circuitry in ventral prefrontal cortex in decision-making and reversal learning. The ascending 5-HT and dopamine neuro-transmitter systems have a modulatory role in both processes. There is accumulating evidence that measures of decision-making andreversal learning may be useful as functional markers of ventral prefrontal cortex integrity in psychiatric and neurological disorders.Whilst existing measures of decision-making may have superior sensitivity, reversal learning may offer superior selectivity, partic-ularly within prefrontal cortex. Effective decision-making on existing measures requires the ability to adapt behaviour on the basis ofchanges in emotional significance, and this may underlie the shared neural circuitry with reversal learning.Ó 2003 Elsevier Inc. All rights reserved.Keywords: Orbitofrontal; Basal ganglia; ParkinsonÕs disease; Addiction; Psychopathy; Risk taking1. IntroductionFollowing brain injury to the orbitofrontal andventromedial prefrontal cortex (PFC), human patientsdisplay gross alterations in social and emotional be-haviour with largely preserved perception, language,memory, and even executive function (Bechara, Tranel,& Damasio, 2000; Damasio, 1994; Malloy, Bihrle,Duffy, & Cimino, 1993; Rolls, 1999). Characterisationof this profile using cognitive testing has been thetarget of considerable research, not least because thebehaviour of patients with ventral prefrontal lesionsresembles aspects of symptomatology seen in psychi-atric conditions including psychopathy (Lapierre,Braun, & Hodgins, 1995) and substance abuse (Bec-hara & Damasio, 2002). Two cognitive domains havereceived particular attention in recent years: decision-making and reversal learning. The development ofseveral measures of decision-making has stemmed lar-gely from observations by Damasio, Bechara and col-leagues, that patients with ventromedial prefrontalcortex damage are impaired in their ability to makesuccessful everyday decisions regarding employment,relationships, and personal finances. Specifically, it hasbeen proposed that these patients are unable to usepast experiences to guide their ongoing decision-mak-ing (Ômyopia for the futureÕ) (Bechara et al., 2000;Damasio, 1994). Recent interest in reversal learning, incontrast, has developed from pre-clinical research overmore than three decades demonstrating that rodentsand non-human primates with lesions to the orbito-frontal cortex are unable to adapt their respondingfollowing changes in stimulus–reward contingencies(Butter, 1969; Jones & Mishkin, 1972).The purpose of this article is to review convergingevidence for the involvement of ventral prefrontal cortexin decision-making and reversal learning, from (1) hu-man lesion studies, (2) animal lesion studies, and (3)human functional neuroimaging studies. Evidence forthe contribution of the ascending 5-HT and dopamineneurotransmitter systems to these domains will also bedescribed. Recent cognitive research in a number ofclinical groups has begun to investigate the sensitivityand selectivity of decision-making and reversal learni ngdeficits as indices of ventral prefrontal dysfunction.*Corresponding author. Fax: +44-1223-333564.E-mail address: [email protected] (L. Clark).0278-2626/$ - see front matter Ó 2003 Elsevier Inc. All rights reserved.doi:10.1016/S0278-2626(03)00284-7Brain and Cognition 55 (2004) 41–53www.elsevier.com/locate/b&c2. The cognitive neuroscience of decision-makingDecision-making requires the evaluation of multipleresponse options, followed by the selection of the re-sponse considered optimal. Each response option maybe characterised in terms of the reward and punishmentoutcomes wi th which it is associated. Response optionsmay vary in terms of (1) the magnitude of reward andpunishment, (2) the probability of receiving reward orpunishment, and (3) the delay to reward or punishment.This framework provides scope for a range of decision-making abnormalities in clinical groups. Deficits maybecome apparent in terms of increased sensitivity toreward or reduced sensitivity to punishment, or at amore complex level under situations of conflict; for ex-ample, a failure to avoid rewards with long-term nega-tive consequences, or the preference for a smallimmediate reward over a larger but delayed reward.This latter phenomenon is known as temporal (or delay)discounting, and exemplifies the relationship betweendecision-making and impulsivity, given that an opera-tional definition of impulsive behaviour is the tendencyto choose a small or inferior immediate reward over alarger delayed reward (Evenden, 1999; Logue, 1988).2.1. Human lesion studiesNeuropsychological studies of decision-making inhumans have utilised two paradigms in recent years: theIowa Gambling Task (Bechara, Dama sio, Damasio, &Anderson, 1994) and the Cambridge Gamble Task(Rogers, Everitt, et al., 1999). The Iowa Gambling Taskis described in detail elsewhere in this issue (Bechara,this issue). The task emphasises the learning of rewardand punishment associations in order to gu ide ongoingdecision-making. Healthy subjects performing the IowaGambling Task learn to avoid ÔriskyÕ card decks thatoffer high immediate rewards with a concomitant risk ofoccasional very high punishment. They develop a pref-erence instead for ÔsafeÕ card decks where the immediaterewards are smaller but there is a low risk of punish-ment. Patients with bilateral damage to the ventrome-dial PFC do not acquire a preference for the safe deckson the Iowa Gambling Task, but instead prefer the