Medial orbitofrontal cortex codes relative rather than absolute value of financial rewards in humans R. Elliott, Z. Agnew and J. F. W. Deakin Neuroscience and Psychiatry Unit, Room G907, Stopford Building, University of Manchester, Oxford Road, Manchester M13 9PT, UK Keywords: amygdala, fMRI, motivation, reinforcement, striatum Abstract Functional imaging studies in recent years have confirmed the involvement of orbitofrontal cortex (OFC) in human reward processing and have suggested that OFC responses are context-dependent. A seminal electrophysiological experiment in primates taught animals to associate abstract visual stimuli with differently valuable food rewards. Subsequently, pairs of these learned abstract stimuli were presented and firing of OFC neurons to the medium-value stimulus was measured. OFC firing was shown to depend on the relative value context. In this study, we developed a human analogue of this paradigm and scanned subjects using functional magnetic resonance imaging. The analysis compared neuronal responses to two superficially identical events, which differed only in terms of the preceding context. Medial OFC response to the same perceptual stimulus was greater when the stimulus predicted the more valuable of two rewards than when it predicted the less valuable. Additional responses were observed in other components of reward circuitry, the amygdala and ventral striatum. The central finding is consistent with the primate results and suggests that OFC neurons code relative rather than absolute reward value. Amygdala and striatal involvement in coding reward value is also consistent with recent functional imaging data. By using a simpler and less confounded paradigm than many functional imaging studies, we are able to demonstrate that relative financial reward value per se is coded in distinct subregions of an extended reward and decision- making network. Introduction Animal and human studies have demonstrated that orbitofrontal cortex (OFC) is a component of distributed reward circuitry (Schultz, 2000, 2005; Kringelbach & Rolls, 2004; Elliott & Deakin, 2005). Specifically, it has been suggested that OFC neurons are involved in ‘higher level’ aspects of reward processing, such as expectation, decision-making and, critically for this study, representation of value. Electrophysiological studies show that OFC responses to food reward depend on relative motivational value of the food (Critchley & Rolls, 1996; Rolls et al., 1998). This suggests that OFC neurons code current incentive value rather than sensory properties. Analogous functional magnetic resonance imaging (fMRI) studies have also demonstrated value-dependent responses to primary rewards. Enhanced OFC responses to pleasant tastes and smells have been observed for hungry compared with satiated subjects (O’Doherty et al., 2000; Small et al., 2001; Gottfried et al., 2003). In a seminal experiment, Tremblay & Schultz (1999) directly assessed relative reward preference in primate OFC. Animals were taught to associate three visual cues with raisins, apple and cereal in a spatial delayed-response task. Raisins were preferred over apple over cereal. In trial blocks, animals were exposed to only two of the three cue–reward pairings. Firing of OFC neurons to apple cues depended on context, being greater for apple–cereal pairing (where they were associated with the preferred food), than for apple–raisin pairing (where they were associated with the less-preferred food). This suggests that OFC responses depended on relative preference among available rewards. Other electrophysiological studies have also suggested that orbitofrontal neurons code relative, rather than absolute, incentive values of rewards (Schoenbaum et al., 1998; Gallagher et al., 1999; Watanabe, 1999; Baxter et al., 2000; Roesch & Olson, 2004). In human fMRI studies, financial reinforcers have the advantage of directly quantifiable value. O’Doherty et al. (2001) showed that medial OFC response correlated with amount of abstract (‘play’) money won on a decision-making task. We have also demonstrated a relationship between medial OFC response and reward value (Elliott et al., 2004). Studies have also shown OFC responses to cues predicting rewarding outcomes, suggesting a role in expectancy (Breiter et al., 2001; Knutson et al., 2001, 2005). However, many fMRI studies of financial reward use complex paradigms with several variables. In a recent review O’Doherty (2007) identified multiple putative functions of OFC within the reward-processing domain, which are often conflated. There is a relative dearth of studies which focus on one specific function. As theorists attempt to develop models of OFC function (Bechara & Damasio, 2005; Kringelbach, 2005), there is a need for simple paradigms to isolate component reward processes. In this study we employed a human analogue of the task used by Tremblay & Schultz (1999), using cues predicting three different values of financial reward. We hypothesized that BOLD response to the cue predicting the intermediate value would be enhanced in OFC when it is the larger, rather than the smaller, of two possible outcomes. That is, OFC response to identical perceptual input, predicting identical outcome, will depend on relative value context. Correspondence: Dr Rebecca Elliott, as above. E-mail: rebecca.elliott@man.ac.uk Received 11 May 2007, revised 11 February 2008, accepted 12 March 2008 European Journal of Neuroscience, Vol. 27, pp. 2213–2218, 2008 doi:10.1111/j.1460-9568.2008.06202.x ª The Authors (2008). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd