To be specific: The role of orbitofrontal cortex in signaling reward identity

Abstract

The orbitofrontal cortex (OFC) plays a prominent role in signaling reward expectations. Two important features of rewards are their value (how good they are) and their specific identity (what they are). Whereas research on OFC has traditionally focused on reward value, recent findings point toward a pivotal role of reward identity in understanding OFC signaling and its contribution to behavior. Here, we review work in rodents, nonhuman primates, and humans on how the OFC represents expectations about the identity of rewards, and how these signals contribute to outcome-guided behavior. Moreover, we summarize recent findings suggesting that specific reward expectations in OFC are learned and updated by means of identity errors in the dopaminergic midbrain. We conclude by discussing how OFC encoding of specific rewards complements recent proposals that this region represents a cognitive map of relevant task states, which forms the basis for model-based behavior. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Figure 1.. Responses of hypothetical neural units to rewards.

A) In a hypothetical experiment, two outcomes differ in value, as indicated by the relative size of the circles, and also have distinct identities, as indicated by color. B) In this scenario, a hypothetical neural unit that codes for value (i.e., higher response to higher-value reward than smaller-value reward) would be indistinguishable from a neural unit that responds specifically to “blue”. C) In another hypothetical experiment, the two distinct reward identities have a value-matched small-value (small circles) and large-value (large circles) version, which may be achieved by carefully manipulating reward magnitude. D) In this case, a value-coding unit has a distinct response pattern compared to “blue” and “red” units.

Figure 2.. Representations of specific reward identity in OFC.

A) Adapted from (McDannald et al., 2014). Left: schematic of the unblocking paradigm showing predictive odor cues (clouds) and flavored milk rewards (drops). Middle: Boxes show location of single unit recording in rat lateral OFC and agranular insula. Right: A portion of single units in lateral OFC responded to “flavor unblocked” odor cues. B) Adapted from (J. D. Howard et al., 2015). Left: Odors of distinct identity (pink=sweet, teal=savory) were individually chosen and matched for rated pleasantness (i.e. value) at two different intensity levels. Middle: two sets of visual cues were paired with odor rewards in a Pavlovian conditioning task while subjects underwent fMRI. Right: Decoding of patterns of fMRI activity evoked by predictive cues revealed significant information about reward identity encoded in right lateral OFC and hippocampus (HIP). C) Adapted from (J. D. Howard & Kahnt, 2017). Left: Information about specific reward identity was decoded in medial and lateral posterior OFC (pOFC) in the pre-satiety phase of a reinforcer devaluation task. Right: After one reward was selectively devalued, decoding of identity information for cues predicting the sated (i.e., devalued) reward was no longer significant in the post-satiety phase, whereas identity decoding remained intact for the non-sated (i.e. non-devalued) reward. D) Adapted from (Xie & Padoa-Schioppa, 2016). An example neuron recorded in macaque OFC showing preferential responding to the value of a specific reward identity in one block of binary choice trials (left: kiwi punch), and the same neuron responding preferentially to the value of a different identity (right: apple juice) in a subsequent block of binary choice trials. E) Adapted from (Takahashi et al., 2017). Top: location of single unit recording electrodes targeting midbrain dopaminergic neurons in the rat. Bottom: Response of midbrain dopamine neurons to violations in expected reward identity. F) Adapted from (J. D. Howard & Kahnt, 2018). Top left: Schematic of reversal learning task, in which associations periodically changed to induce violations in either reward value (i.e., a “value reversal”) or reward identity (i.e., an “identity reversal”). Top right: violations in both reward value and reward identity evoked prediction error responses in overlapping regions of the human midbrain. Bottom left: Pattern-based analysis of cue-evoked activity revealed information about expected reward identity encoded in left lateral OFC. Bottom middle: After an identity reversal, information in OFC activity patterns changed to reflect associations with the new reward identity. Bottom right: The change in OFC identity coding after an identity reversal (i.e., the “OFC identity update”) was directly related to the magnitude of the identity prediction error signal across subjects.