Orbitofrontal control of visual cortex gain promotes visual associative learning

Abstract

4Signaling of expected outcomes in the orbitofrontal cortex (OFC) is critical for 5 outcome-guided and learning behavior. The OFC projects to primary visual 6 cortex (V1), yet the function of this top-down projection is unclear. We found 7 that optogenetic activation of OFC projection to V1 reduced the amplitude of V1 8 visual responses via the recruitment of local somatostatin-expressing (SST) 9interneurons. Using mice performing a Go/No-Go visual task, we showed that 10 the OFC projection to V1 mediated the suppression of V1 responses to the 11 reward-irrelevant No-Go stimulus. Furthermore, the responses of V1-projecting 12 OFC neurons to No-Go stimulus were reduced when the mice's expectation was 13 incorrect. In addition, optogenetic inactivation of OFC projection to V1 14 impaired, whereas activation of SST interneurons in V1 improved the learning of 15 Go/No-Go visual task. Thus, OFC top-down projection to V1 is crucial to drive 16 visual associative learning by reducing the response gain of V1 neurons to non-17 relevant stimulus. 18 19The OFC is a critical brain region for using the information about expected 20 outcomes to guide learning and behavior 1-3 . Studies in rodents and monkeys have 21 demonstrated that the identity and expected values of specific outcomes are 22 represented by activities in the OFC 4-14 . Lesions or inactivation of the OFC impair 23 behavior guided by outcome expectancy and learning driven by the discrepancy 24 48 V1 visual response by recruiting SST interneurons. We thus hypothesized that OFC 49 projection to V1 may filter out non-relevant visual information, and tested this 50 hypothesis in mice performing a Go/No-Go visual task. We found that the OFC 51 projection to V1 contributed to the suppression of V1 responses to the No-Go 52 stimulus, which was not associated with reward. Optogenetic tagging of V1-53 projecting OFC neurons revealed that their responses to the No-Go stimulus were 54 reduced in wrong trials but not in correct trials. We further showed that optogenetic 55 inactivation of OFC projection to V1 slowed the learning of Go/No-Go visual 56 behavior. Thus, the OFC projection to V1 plays a key role in filtering out non-relevant 57 visual information to facilitate associative learning. 58 59Results 60 OFC top-down projection controls V1 response amplitude by activation of local 61SST interneurons. We injected Cholera toxin subunit B (CTB) in V1 and found that 62 the retrograde labeled neurons were in the ventrolateral OFC (vlOFC) (Fig. 1a), 63 consistent with the finding in a previous study 38 . By injecting rAAV2-retro-hSyn-Cre 64 in V1 and AAV-DIO-EYFP in the OFC, we found that the axons of OFC neurons 65 terminated in both superficial and deep layers of V1 (Fig. 1b). To examine how the 66 5 OFC top-down projection influences V1 neuronal responses, we expressed excitatory 67 opsin Channelrhodopsin-2 (ChR2) or ChrimsonR in the OFC, and measured V1 68 responses with and without laser stimulation of OFC axons in mice passively viewing 69 drifting gratings (Fig. 1c)....

A cortical circuit mechanism for coding and updating task structural knowledge in inference-based decision-making

“…Recent studies began to elucidate the functions of specific projections from OFC to certain cortical and subcortical regions [8][9][10] . However, the sensory discrimination learning 8 or the conventional reversal learning based on the go/no-go task 9 provides only a single prolonged learning curve, does not manifest model-based algorithm, and cannot distinguish rapid behavioral adaptation via model-based inference 2,3,11 from slow model-free reinforcement learning 1,12 . Therefore, it remains an open question how the structured knowledge is represented in the brain, and what type of circuit architecture and operation mechanism is responsible for using and updating of the internal models during flexible decision-making.…”

Value Shapes Abstraction During Learning

“…The facilitating effect of value on abstraction can be thus directly linked to changes in the early processing stage of visual information. In line with this interpretation, recent work in mice has elegantly reported how value governs a functional remapping in the sensory cortex by direct lateral OFC projections carrying RPE information 46 , as well as by modulating the gain of neurons to irrelevant stimuli 47 . While these considerations clearly point to a central role of the vmPFC and valuation in abstraction -by controlling sensory representations, it remains to be investigated whether this effect results in more efficient construction of abstract representations, or in better selection of internal abstraction-based RL algorithms.…”

“…a top-down (attentional) effect, to tag sensory information 45 . Work in rodents has reported strong top-down modulation of sensory cortices by OFC neurons implicated in value computations 46,47 . We thus hypothesized abstraction could result from a direct effect of value in the VC.…”

“…Artificially adding value in the form of reward to a feature representation in the VC resulted in increased use of abstractions. The direct lateral OFC projections carrying RPE information46 , as well as by modulating the gain of neurons to irrelevant stimuli47 . While these considerations clearly point to a central role of the vmPFC and valuation in abstraction -by controlling sensory representations, it remains to be investigated whether this effect results in more efficient construction of abstract representations, or in better selection of internal abstraction-based RL algorithms.Given the complex nature of our design, there are some limitations to this work.…”

Learning enhances behaviorally relevant representations in apical dendrites

“…In contrast to previous studies of discrimination learning (Henschke et al, 2020;Liu et al, 2020;Poort et al, 2015), we included an unrewarded post-conditioning session to examine whether learning-related effects persisted through extinction. Our results show that post-conditioning selectivity of the apical population remains significantly higher than pre-conditioning, even after animals stop licking in response to the CS+ (Fig.…”