Heterogeneous medial prefrontal cortex ensembles exhibit reproducible internal temporal dynamics during choice behavior

Abstract

Decision-making requires the coordinated integration of sensory information, internal states, and learned rules across time. While the medial prefrontal cortex (mPFC) is known to contribute to this process, how population-level neural activity in the mPFC is temporally organized during decision-making remains unclear. Here, we investigated neuronal population activity in the mPFC of male mice using in vivo calcium imaging while animals performed a touchscreen-based visual cue-dependent location-choice task. We found that mPFC population activity exhibited a stable, low-dimensional, and continuous temporal structure that emerged reproducibly across trials. Using circular phase decoding, we found that these internal population dynamics tracked trial progression across pre-stimulus, stimulus, and post-stimulus epochs. At the single-neuron level, a large proportion of mPFC neurons exhibited reliable, time-locked activity changes around task events. Hierarchical clustering summarized these heterogeneous responses into temporally organized response profiles spanning different phases of the task. Choice-related decoding was observed across these mPFC response profiles, whereas decoding performance in the primary somatosensory cortex (SSp) remained near chance. Together, these results suggest that mPFC population activity exhibits reproducible temporal organization during visual cue-dependent choice behavior, linking single-neuron heterogeneity with population-level progression dynamics. Significance statement Understanding how the medial prefrontal cortex (mPFC) contributes to decision-making is a central question in neuroscience. Using calcium imaging in male mice performing a touchscreen-based cue–location task, we found that mPFC neurons exhibited heterogeneous but reproducible temporal activity patterns across trials. Population-level analyses revealed low-dimensional temporal dynamics that tracked progression through the task, whereas hierarchical clustering provided a descriptive summary of temporally organized response profiles. Choice-related decoding was observed across these response profiles. These findings suggest that mPFC activity during choice behavior is organized at both single-neuron and population levels, providing insight into how diverse neuronal responses are coordinated over time during goal-directed behavior.