Activation Model Specification
Version: 0.1.0-draftOverview
The activation model defines how nodes decide whether to process incoming signals. It replaces traditional rate limiting with a biologically-inspired threshold system.Activation State
Each node MUST maintain an activation state:Activation Sequence
- Signal arrives at node via synapse
- Signal contribution is calculated:
contribution = signal.weight * synapse.weight - Contribution is added to node’s potential:
potential += contribution - If
potential >= thresholdAND node is not in refractory period: a. Node fires (processes the signal) b. Potential resets to 0.0 c. Node enters refractory period - If potential < threshold OR node is in refractory: a. Signal is queued for accumulation b. No processing occurs
Dynamic Threshold
The threshold SHOULD adjust dynamically based on node load:load_factor = current_queue_depth / max_queue_depth.
This ensures that overloaded nodes become more selective, providing natural backpressure.
Refractory Period
After firing, a node MUST enter a refractory period during which it cannot fire again. The RECOMMENDED default is 10 milliseconds. During the refractory period:- Incoming signals still accumulate potential
- No processing occurs
- No signals are dropped (they queue)
Activation Functions
Implementations MUST support thestep activation function. Implementations SHOULD support sigmoid and leaky functions.
| Function | Behavior |
|---|---|
| Step | Binary: fires when potential >= threshold |
| Sigmoid | Probabilistic: firing probability increases smoothly |
| Leaky | Always passes a small fraction (leak rate = 0.01) |