In summary, it is argued that lOFC is relatively more specialized for assigning credit for both rewards and errors to specific stimulus choices. When different types of reward GSI-IX chemical structure outcome are available then lOFC represents the assignment of a particular reward type to a particular stimulus. By contrast, it is argued that vmPFC/mOFC value representations are not so much of the specific identify of a reward outcome but of its value and that it is these value representations that determine the goals and choices that primates pursue. The few neuron recording studies that have compared the areas support this interpretation. Rolls (2008) reports that neurons encoding dimensions

of reward outcomes, such as taste and JQ1 nmr texture, are more prevalent in lOFC than vmPFC/mOFC in the macaque. Bouret and Richmond (2010) report that lOFC neurons are more active than vmPFC/mOFC neurons when macaques see visual stimuli that predict rewards. By contrast vmPFC/mOFC neurons have greater access to information about the macaque’s current motivational state; the activity of vmPFC/mOFC neurons, but not lOFC neurons, was modulated by satiety (Figure 6). A very influential observation has been the report of neurons encoding the values of potential choices (“offer-value”-correlated activity) and the values of choices that are actually taken (“chosen-value”-correlated activity) in the lateral bank of the medial orbital sulcus

of and the adjacent posterior orbitofrontal cortex (Padoa-Schioppa and Assad, 2006, Padoa-Schioppa and Assad, 2008 and Padoa-Schioppa, 2009), a region at the transition between vmPFC/mOFC and lOFC divisions (Ongür and Price, 2000). It is tempting to relate the activity of such neurons to human vmPFC/mOFC BOLD signals that reflect the values of available choices and of taken choices (Boorman et al., 2009, FitzGerald et al., 2009, Philiastides et al., 2010 and Wunderlich et al.,

2010) but it is not clear whether the frequency of such neural patterns changes between vmPFC/mOFC and lOFC. In many experiments it is assumed that during decision-making people first weigh and compare the values of all of the different options that are available in order to make a choice and second, that these values are learned from the experience of previously choosing these options. Neither of these assumptions may be true. Instead the choice made and the best alternative may each have a special status. Moreover, learning about the value of choices can sometimes occur even without taking the choice if the right feedback is provided. Recent studies of aPFC provide the key evidence for both of these propositions. The aPFC carries a very distinct signal to the vmPFC. While vmPFC/mOFC encodes the value of the choice that is being made the aPFC encodes information about the value of alternative options that are not chosen (Boorman et al., 2009).