Decoding instrumental lever pressing from prefrontal, accumbens, and hippocampal local field potentials: conservation across sex, task, and dopamine depletion.

UNLABELLED: During operant behavior, oscillatory activity is coordinated across multiple brain regions and can change under different task conditions. In this study, we use a decoding approach to characterize these patterns and predict lever pressing behavior in rats from local field potentials (LFP) recorded bilaterally in the hippocampus, nucleus accumbens, and prefrontal cortex. We extract LFP features for band power and peak frequency and predict behavior across animals and conditions with a Poisson Generalized Linear Model (GLM). We use data from both male and female rats performing either fixed-ratio (FR40) or progressive (PROG) operant lever-pressing tasks, under vehicle (VEH) or tetrabenazine (TBZ), a vesicular monoamine transport (VMAT-2) inhibitor that depletes dopamine and induces depressive-like motivation dysfunction. We find that we can accurately predict lever pressing from multi-region LFP within-animals on a timescale of seconds with >30% variance explained. Although within-animal and within-condition predictions are highest, we also find, interestingly, that these decoders generalize across animals and across drug, sex, and task conditions, suggesting a stable association between LFP and behavior. We then evaluate whether LFP features can be used to decode long-term task variables, such as the number of presses since or until the next reinforcer. We find that LFP features can predict these long-term task variables, but that decoding relies on different LFP features for prediction than those used for immediate lever pressing. Altogether these results suggest that there are robust distributed patterns of LFP associated with lever-pressing behavior under multiple drug, task, and sex conditions.

SIGNIFICANCE STATEMENT: Although patterns of neural activity related to many observable behaviors, such as sleep and locomotion, have been well described, identifying stable neural patterns related to other states, such as motivation and effort, is an ongoing challenge. Here we use a decoding approach and aim to predict effort-based lever pressing from neural activity for a wide range of conditions, with rats of different sexes performing different tasks with and without a drug, tetrabenazine. We find that there is a stable pattern of neural features that allows us to robustly predict lever-pressing across these many conditions. This pattern may serve as a general indicator of effortful behavior, and deviations from the pattern could potentially indicate motivational dysfunction.