A stable, distributed code for cue value in mouse cortex during reward learning

Abstract

The ability to associate reward-predicting stimuli with adaptive behavior is frequently attributed to the prefrontal cortex, but the stimulus-specificity, spatial distribution, and stability of neural cue-reward associations are unresolved. We trained headfixed mice on an olfactory Pavlovian conditioning task and measured the coding properties of individual neurons across space (prefrontal, olfactory, and motor cortices) and time (multiple days). Neurons encoding cues and licks were most common in olfactory and motor cortex, respectively. By classifying cue-encoding neurons according to their responses to the six odor cues, we unexpectedly found value coding, including coding of trial-by-trial reward history, in all regions we sampled. We further found that prefrontal cue and lick codes were preserved across days. Our results demonstrate that individual prefrontal neurons stably encode components of cue-reward learning within a larger spatial gradient of coding properties.

Multi-areal neural dynamics encode human decision making

“…As HFA is highly correlated with fMRI BOLD signal (30,44), this observation directly implicates neuronal activity in choice, and helps bridge fMRI studies with animal studies that implicate SUA or MUA activity, which is also reflected in high-frequency LFP activity (24,42,44), in decision-making processes. Second, we demonstrate a highly distributed nature of human decision-making processes across regions, consistent with observations in animal models (6,7,50,53), while providing evidence for a more modular encoding of several decisionrelated variables such as risk (58), win probability (57) and motor activity(28) consistent with observations from non-invasive studies. Given their highly distributed nature, our sEEG recordings cover a wide number of areas and provide a more complete depiction of electrophysiological activity than in most animal model studies which are limited to a few areas (68,69).…”

“…The ubiquity of choice-related L/HFA encoding across areas in our study is consistent with the view that value-based decision-making is a distributed process engaging multiple brain areas, likely simultaneously. The growing consensus on the existence of widespread brain activation for multiple cognitive processes including decision-making is supported by an increasing amount of evidence from primate single unit recordings(4, 5, 49-51), rodent (6,7), and human imaging studies (52). Our data support a distributed model of decision-making (50,53), which extends 'action-plan' models (54)(55)(56) by proposing that decision-making is a parallel, distributed process that weighs information across all representational levels, including those of good-based valuation and up until action.…”

“…As above, the first PC for each region-powerband combination was used in LDS modeling. The appropriate dimensionality of the model (number of latent variables, LVs) was determined empirically by re-running the model for a range of dimensions (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15) and examining the predictability of neural dynamics. When increasing the dimension size does not improve the predictability of the neural dynamics, the optimal dimensionality has been reached.…”

Widespread coding of navigational variables in prefrontal cortex

“…The modular approach to understanding functional neuroanatomy has been challenged by a growing set of studies that highlight the broad distribution of variables otherwise thought to be more narrowly circumscribed. These include motor signals (Musall et al, 2019a and b; Stringer et al, 2019; Steinmetz et al, 2019) and reward signals (Vickery et al, 2011; Shin et al, 2021; Ottenheimer et al, 2022). These findings raise the possibility that like other cognitive functions, navigation may also be more distributed than previously assumed, perhaps as part of a gradient characterized by gradual untangling rather than strict functional borders (Fine et al, 2022; Fuster, 2000 and 2001; Yoo et al, 2018).…”

A stable, distributed code for cue value in mouse cortex during reward learning

“…The data and code for this manuscript are publicly available at 10.5281/zenodo.6686927 (Ottenheimer et al, 2022).…”