Extracting Value Coding Features from Individual Serotonin Neurons

Abstract

Adaptive behaviour requires animals to continually reevaluate the appetitive or aversive quality of their surroundings. Dorsal raphe serotonin neurons, the main source of serotonergic input to the forebrain, have been implicated in both signaling the quality of an animals environment and regulating reward-seeking and punishmentavoiding behaviour, but the precise quantity signaled by these neurons has remained unclear, as well as how these neurons relate with behaviour. Using open-access recordings of serotonergic neurons of the dorsal raphe nucleus while animals perform a dynamic Pavlovian task, we compare firing rate and behavioral data with a model that considers reward history accumulated over a tunable timescale. Our Bayesian parameter estimation supports that serotonergic neurons are consistent with reward history being estimated over about a hundred trials on average, with a heterogeneity across individual neurons spanning 30 to 300 trials. Anticipatory licking also correlated with reward history at multiple timescale, but could not be dissociated from that of a time/thirst nuisance variable and otherwise mostly on a timescale faster than seen in serotonergic cells. These results provide a more precise picture of the dynamics of serotonergic cells under a dynamic Pavlovian task.