Theta sequences as eligibility traces: a biological solution to credit assignment
This addresses a fundamental challenge in biological neural networks for reinforcement learning, offering a plausible mechanism for credit assignment.
The paper tackles the credit assignment problem in reinforcement learning by proposing theta sequences as a biological solution, showing they enable bootstrap-free credit assignment without long memory traces, equivalent to TD(λ) eligibility traces.
Credit assignment problems, for example policy evaluation in RL, often require bootstrapping prediction errors through preceding states \textit{or} maintaining temporally extended memory traces; solutions which are unfavourable or implausible for biological networks of neurons. We propose theta sequences -- chains of neural activity during theta oscillations in the hippocampus, thought to represent rapid playthroughs of awake behaviour -- as a solution. By analysing and simulating a model for theta sequences we show they compress behaviour such that existing but short $\mathsf{O}(10)$ ms neuronal memory traces are effectively extended allowing for bootstrap-free credit assignment without long memory traces, equivalent to the use of eligibility traces in TD($λ$).