Experience shifts value updating from synaptic plasticity to neural dynamics

In mice undergoing repeated odor–outcome reversals, experience speeds value updating but also accelerates memory decay, and this trade-off is reproduced only by neural-dynamics-based learning.

• Why it matters: Rapid updating in changing environments must balance flexibility with durable value memories, yet the neuronal mechanism producing this stability–plasticity trade-off is unclear.
• What they did: Mice were tested across reversal schedules while RNNs with continual plasticity were trained to compare routes to faster learning, and dopamine neuron inhibition, CaMKII blockade, CaMKIIa deletion, and recurrent-weight freezing were used alongside amygdala/hippocampus recordings.
• The result: Neural dynamics, not more efficient synaptic plasticity, matched the updating–decay trade-off; dopamine or CaMKII disruption impaired early learning but spared expert reversal, and cue value plus latent context representations decayed with long inter-trial intervals.

AI summary of the abstract · may contain errors · not medical advice · read the paper.