Tensions in Practice: The next update can change before the state does¶
A toy two-speed update rule
A toy system stores value 10 and normally accepts a stimulus when its size reaches threshold 3; an accepted stimulus adds 1 to the stored value. Compare keeping that threshold fixed with a slower governor that checks recent activity once per batch. At this checkpoint the recorded activity count is 8, which makes the governor set threshold 6 for the next batch. That threshold change leaves value 10 untouched. Only the next stimulus, of size 5, reveals the difference in readiness.
Keep prompt responsiveness
Continue accepting stimuli of the same size even after recent activity.
Restrain accumulated updating
Raise the next batch’s update threshold after substantial recent activity.
Why these aims pull against each other
Changing the rule of change can suppress another update without changing the current stored state. The same restraint that limits repeated updating can also block a useful stimulus.
Choose an arrangement to see what changes and what remains difficult.
Stored is the base value; Threshold is the minimum stimulus size for a +1 update. Update is the change at that row’s event. The checkpoint changes only the threshold; stimulus 5 then tests the held rule.
What this choice protects
What it costs
When it fits
Compare the arrangements
Keep threshold fixed
At the slow checkpoint, keep threshold 3. The next size-5 stimulus passes and changes the stored value from 10 to 11.
| Stored | Threshold | Update | |
|---|---|---|---|
| Before check | 10 | 3 | None |
| After check | 10 | 3 | None |
| Stimulus 5 | 11 | 3 | +1 |
- What it protects
- The next moderate stimulus still produces an update despite the prior activity count.
- What it costs
- This rule supplies no activity-dependent brake on repeated qualifying stimuli. Any protection against unwanted repeated changes must come from elsewhere.
- When it fits
- Plausible when continued responsiveness is valuable and recent activity does not justify lowering readiness.
Illustration note: This is an editorial, deliberately bounded illustration. Its stated rules and any numbers are invented, not observations, recommended settings, or predictions.
Raise readiness threshold
The once-per-batch governor sets threshold 6 because the prior activity count is at least 8. It retains that threshold across the next batch’s fast stimuli; size 5 now fails it.
| Stored | Threshold | Update | |
|---|---|---|---|
| Before check | 10 | 3 | None |
| After check | 10 | 6 | None |
| Stimulus 5 | 10 | 6 | 0 |
- What it protects
- Recent activity reduces subsequent updating while preserving the existing stored answer at the checkpoint.
- What it costs
- A useful size-5 signal is also rejected; the governor requires a history counter and a separate update schedule.
- When it fits
- Plausible when repeated changes create a real cost and the activity rule can be calibrated without suppressing needed adaptation.
Illustration note: This is an editorial, deliberately bounded illustration. Its stated rules and any numbers are invented, not observations, recommended settings, or predictions.
What this illustration does—and does not—establish
The source establishes the structural tension; the concrete alternatives and their conditional costs are editorial synthesis. No arrangement is a universal recommendation.
- The values, thresholds, count cutoff and +1 update are invented. They are not a model fit to a brain, learner or organization.
- The governor runs on a slower batch clock and holds its parameter across fast stimuli. A fixed high threshold alone would not demonstrate this history-driven second layer.
- The illustration tests only one next stimulus. It proves neither long-term stability nor that rejecting it improves the stored answer.
Source entries
Metaplasticity
Metaplasticity: Anti-Runaway Governor versus Over-Damping (sign/direction) supplies the local tension. The setting, alternative arrangements, and stipulated consequences are editorial applications.
Anti-Runaway Governor versus Over-Damping (sign/direction)
The metaplastic governor that raises the threshold as activity accumulates is the standard defense against runaway potentiation — but the same brake, set too aggressively, suppresses legitimate adaptation, leaving a system that cannot learn when it should.
Current Behavior versus Current Readiness (sign/direction)
Two systems identical in current output can respond oppositely to the next disturbance because one is primed and the other saturated.