Cooperation between model-free and model-based control systems is boosted by enhancing dopamine levels.

The medial prefrontal cortex and orbitofrontal cortex work in concert to match state representations from feedback to those at choice, and the strength of these common neural codes predict credit assignment precision.

A multi-compartment spiking neural network model demonstrates that biologically feasible deep learning can be achieved if sensory inputs and higher-order feedback are received by different dendritic compartments.

Neural activity in the rat nucleus accumbens provides a rich task representation that includes not only expected outcomes, but also the specific identity of the cues that predict these outcomes.

What can artificial intelligence learn from neuroscience, and vice versa?.

A proposal for a complete cellular/network implementation of trial-and-error motor learning in the olivo-cerebellar system.

Variability in individual neurons' temporal receptive fields of integration is found to explain the heterogeneity of neuronal responses observed in prefrontal cortex during reward-guided decision making.

Neuronal reactivation during dopamine modulation induces input-specific LTP at previously primed hippocampal synapses, suggesting a possible solution to the credit assignment problem and a mechanism for memory linking.

Human learning relies on short-term memories (eligibility traces) which provide a mechanism to reinforce sequences of actions from a single reward (one-shot).