Sequence anticipation and spike-timing-dependent plasticity emerge from a predictive learning rule

Area of Science:

• Computational Neuroscience
• Synaptic Plasticity
• Predictive Coding

Background:

• Intelligent behavior relies on the brain's predictive capabilities.
• The precise learning rules enabling neuronal prediction remain largely unknown.
• Anticipatory signaling is crucial for processing sensory input.

Purpose of the Study:

• To propose and investigate a novel plasticity rule based on predictive processing.
• To elucidate how neurons learn to predict future sensory inputs.
• To explain the development of anticipatory signaling and recall.

Main Methods:

• Introduced a plasticity rule where neurons model synaptic input dynamics.
• Utilized a low-rank model of synaptic input within the neuron's membrane potential.
• Analyzed sequence learning and anticipatory signaling in a recurrent neural network.

Main Results:

• Neurons amplify synapses that best predict other inputs based on temporal relations.
• This mechanism enables neurons to learn sequences over long timescales.
• The rule explains anticipatory signaling, recall, and observed spike-timing-dependent plasticity (STDP) mechanisms.

Conclusions:

• Predictive processing offers a viable mechanism for neuronal learning and prediction.
• This single-neuron plasticity rule can orchestrate anticipatory signaling and sequence learning.
• Prediction serves as a fundamental principle guiding synaptic plasticity in the brain.