A Circuit for Associative Reward Learning
Wenliang Wang, Mark A. G. Eldridge, Tsuyoshi Setogawa, Spencer Webster-Bass, Nanami Miyazaki et al.
(see article e0491252025)
Many organisms associate items with the experience of reward to guide reward-based decision-making. Wang and colleagues explored the circuitry involved in visual associative learning in two male rhesus monkeys. In a previous study, the authors found that the rhinal cortex (Rh) and rostromedial caudate (rmCD) are causally linked to visual associative reward learning. Herein, this discovery led Wang et al. to assess the behavioral consequences of artificially silencing layer 5 neurons projecting from the Rh to the rmCD. To silence this projection, the authors combined a unilateral Rh lesion and contralateral expression of an inhibitory DREADD via a retrograde lentivirus injected into rmCD. Interruption of this connection had little effect on already learned stimulus–reward associations but slowed down the learning of new associations. The authors further localized the source of this learning impairment to perirhinal (PRh) neurons projecting to rmCD via local microinjections of deschloroclozapine. Thus, this projection from PRh to rmCD may be critical for learning—but not the retrieval of—visual reward-based associations.
Highlighted are brain regions encoding subjective salience/risk prediction errors more strongly than objective salience/risk prediction errors (left) and vice versa (right). See Kim et al. for more information.
Roles of Insula Subregions in Salient Risk Processing
Jae-Chang Kim, Lydia Hellrung, Stephan Nebe, and Philippe N. Tobler
(see article e2302242025)
When an outcome is better or worse than a person predicted, this is considered a salient, risky outcome. The anterior insula encodes these risky outcomes, regardless of whether they are appetitive (better than predicted) or aversive (worse than predicted) prediction errors. In this issue, Kim and colleagues advance understanding of the insula’s role during appetitive and aversive prediction errors by assessing about 40 people in a Pavlovian conditioning task. The authors compared insula signals to objectively (probability-based) and subjectively (rating-based) prediction errors. They discovered that the anterior insula and adjacent frontal cortex processed subjective salience prediction errors, while the mid-insula processed objective prediction errors. The authors also considered activity at the time of cues and found that anterior insula signals encoded subjective salience prediction errors during the presentation of aversive cues more strongly than during the presentation of appetitive cues. According to the authors, this work reveals domain-specific salience signals during cue delivery and suggests that subregions of the insula play distinct roles in the processing of salient risk prediction errors.
Footnotes
This Week in The Journal was written by Paige McKeon
