Metastable attractors explain the variable timing of stable behavioral action sequences
Natural animal behavior displays rich lexical and temporal dynamics, even in a stable environment. The timing of self-initiated actions shows large variability even when they are executed in reliable, well-learned sequences. To elucidate the neural mechanism underlying this mix of reliability and stochasticity, we trained rats to perform a stereotyped sequence of self-initiated actions and recorded neural ensemble activity in secondary motor cortex (M2), known to reflect trial-by-trial action timing fluctuations. Using hidden Markov models, we established a dictionary between ensemble activity patterns and actions. We then showed that metastable attractors, with a reliable sequential structure yet high transition timing variability, could be produced by coupling a high-dimensional recurrent network and a low-dimensional feedforward one. Transitions between attractors in our model were generated by correlated variability arising from the feedback loop between the two networks. This mechanism predicted aligned, low-dimensional noise correlations that were empirically verified in M2 ensembles. Our work establishes a novel framework for investigating the circuit origins of self-initiated behavior based on correlated variability.