MesoNet: plasticity-state sketch

plasticity state: decay between updates, bump on a change. write the actual jump rule down; neighboring updates feeding it back could get out of hand

intended state: each synapse has a plasticity value that decays over time and increases when its weight is updated. Higher plasticity makes a connection more susceptible to further change.

equation recorded: dP/dt = -P/tau. The next line in the source writes an increment involving C/tau on an update, but uses derivative notation inconsistently; do not assume a unique discrete-time implementation from it.

design tension: local STDP may miss broader patterns, while wider update propagation can destabilize convergence. The extra state itself also adds per-synapse computation/storage.

Local STDP and distributed updates

Variable plasticity

• review questions: does an event increment P, its derivative, or a discrete accumulator? What are the units of C and tau? Are plasticity and confidence inversely or directly related—the source's informal wording needs a precise definition. Write an isolated decay-and-event test before coupling this state to graph incidence.

• References saved in this journal section

• https://doi.org/10.3390/brainsci12070863
• https://doi.org/10.48550/arXiv.2304.07193
• https://www.nature.com/articles/s41598-023-32120-7
• https://brian2.readthedocs.io/en/latest/resources/tutorials/2-intro-to-brian-synapses.html

Graph-plasticity sketches

Graph-attribution working notes