Ordinary-Training Comparator Protocol¶

Comparator protocol — instantiates Reopened Malleability Window

Runs a matched ordinary-input control arm so any gain can be credited to reopened capacity rather than to more practice, assistance, or expectancy.

The whole premise of a reopening is that the target had stopped responding to ordinary input — so the first thing to rule out is that ordinary input would have worked anyway. Ordinary-Training Comparator Protocol is the matched control arm that receives the same corrective input, at the same dose, without the reopening trigger. Its defining job is to make the reopening claim falsifiable: if the comparator improves just as much, the induction added nothing beyond more practice, more scaffolding, or the expectancy of a special treatment. It supplies the counterfactual against which every claimed gain is scored, so "it worked" can be separated from "it changed the capacity to update."

Example¶

An adult-language lab is testing whether a plasticity-reopening adjunct lets native Japanese speakers re-acquire the English /r/–/l/ contrast — a perceptual boundary that typically plateaus well short of native-like after childhood. The reopening arm gets the trigger plus discrimination training. The Ordinary-Training Comparator Protocol runs a matched arm that gets the identical training schedule, the same encouragement, and a credible inactive procedure — everything except the induction.

By the end, both arms are better at the specific trained syllables. The comparator makes the decisive cut visible: the ordinary arm's gains sit almost entirely on trained tokens and fade with new talkers, while only a difference that exceeds the comparator — especially transfer to untrained voices — can be credited to reopened capacity. Without the control arm, the trained-token improvement alone would have read as success.

How it works¶

What distinguishes this protocol is disciplined matching, not the training itself:

Match everything but the trigger — same input, dose, attention, and expectancy in both arms, so the only difference is the induction.
Include a compensation check — compare supported versus unsupported performance, catching gains that are scaffolding or strategy rather than a real change in capacity.
Score on the margin — the outcome is the difference between arms, weighted toward transfer and durability, not either arm's raw improvement.
Guard the counterfactual — pre-register the comparator and its match dimension so it cannot be reshaped after the results are in.

Tuning parameters¶

Match dimension — which variables are held equal (time, dose, attention, expectancy). Matching on the wrong one — equal time but not equal attention — lets an uncontrolled factor leak into the "reopening" effect.
Comparator type — no-treatment, active-sham, or supported-performance control. Stronger controls isolate capacity more cleanly but cost more to run.
Blinding — whether participants and assessors know the arm. Blinding is what neutralizes expectancy, the confound this protocol most exists to kill.
Transfer breadth — how far beyond trained items the outcome reaches. Wider transfer tests distinguish genuine capacity change from item-specific practice.
Design — parallel arms versus within-subject crossover. Crossover controls individual differences but risks carryover from a window that does not fully reclose.

When it helps, and when it misleads¶

Its strength is that it targets the three things most likely to masquerade as reopened plasticity: additional practice, compensation and scaffolding, and expectancy. It is also the natural guard against reading an ordinary up-and-down fluctuation as an effect — regression to the mean flatters any single arm measured from a low baseline, and only a matched comparator removes it.^[1]

It misleads when the match is weak or the arm is under-powered: a comparator that differs from the reopening arm in some uncontrolled way, or that is too small to detect a real difference, manufactures a false positive as readily as no control at all. And it is easily run backwards — a deliberately feeble control chosen to make the trigger look good. The discipline is to pre-specify the comparator, match it on the dimension the trigger is suspected to move, blind where possible, and power it to detect the difference that would matter.

How it implements the components¶

ordinary_input_and_compensation_comparator — it is this component: the matched ordinary-input arm, including the supported-versus-unsupported compensation check, that turns the reopening claim into a testable difference.

It does not establish that a capacity constraint exists in the first place (closed_state_and_capacity_baseline — that's the Closed-State Capacity Challenge Panel) or verify that the induction changed the update state (Reopening-Signal Verification Panel); it isolates whether the change exceeds ordinary input, not whether the window opened.

Instantiates: Reopened Malleability Window — it provides the counterfactual that keeps the reopening claim falsifiable.
Sibling mechanisms: Reopening-Signal Verification Panel · Closed-State Capacity Challenge Panel · Trigger-Specificity and Dose-Escalation Trial · Selective Re-stabilization Challenge · Delayed Retention, Transfer, and Interference Battery

Notes¶

This protocol and the Reopening-Signal Verification Panel answer different questions and are complementary: the comparator asks "did the result beat ordinary input?" (an outcome contrast), while the verification panel asks "did the update state actually change?" (a mechanism check). A protocol can pass one and fail the other — a real state change that yields no advantage over practice, or an outcome edge with no evidence the window ever opened — which is why both belong in the appraisal.

References¶

[1] Regression to the mean — extreme measurements tend to be followed by ones closer to average, with no intervention at all. It is a standard reason a single treated group appears to improve, and a matched comparator is the standard corrective. ↩