Generalization Probe¶
Probe — instantiates Negative-Mere-Exposure Reversal for Disliked Targets
Tests whether reduced aversion survives outside the practice setting by deliberately measuring the response in a fresh, untrained context.
Generalization Probe asks the question the training setting cannot answer itself: does the softened attitude travel? Its defining move is a deliberate change of context — it takes the response that has improved inside the practiced, protected exposure conditions and measures it again in a new, untrained setting to see whether the gain is real or merely local. It is not the ongoing in-setting log and not the first taste; it is a purpose-built transfer test, run at the point where a bounded success has been achieved and the open question is whether that success is context-bound. The premise is that liking acquired in one artificial arrangement often evaporates the moment the arrangement changes, so an intervention that never probes outside its own conditions cannot know if it has actually moved anything.
Example¶
A driving school has spent weeks getting an anxious learner comfortable behind the wheel — but all of it on the same quiet residential loop at the same off-peak hour, with the same calm instructor beside her. Inside those conditions her white-knuckle grip is gone and she reports feeling fine. The instructor does not mistake that for success. He runs a generalization probe: an unannounced-in-advance session on an unfamiliar arterial road, at a busier hour, with a different instructor in the passenger seat. He measures the same things he has been measuring all along — self-reported anxiety, physical tension, willingness to continue — but now in the untrained context, and compares them to her comfortable in-setting readings. If the calm holds, the reduced aversion has genuinely generalized and she is closer to ready. If it collapses back to white knuckles the instant the road and the instructor change, then what she acquired was comfort with that loop, not with driving — and the training has to add varied contexts before it can claim to have worked. The probe is what separates those two conclusions.
How it works¶
The probe deliberately varies the exposure context along the dimensions most likely to matter — setting, time, people, stakes, or format — while holding the target itself constant, and re-measures the response there. It reuses the program's affect indicators but takes the reading outside the trained conditions, then compares the fresh-context response to the in-setting response. A holding response means the gain has transferred; a collapse means it was context-bound. What distinguishes it from its siblings is where and why it measures: not to track the trajectory within the practice setting over time, but to test, at a chosen moment, whether that trajectory means anything beyond it.
Tuning parameters¶
- Context distance — how far the probe setting is from the training one. A distant context is a stringent test of real transfer but risks overwhelming a fragile gain; a near context is gentler but proves less.
- Number of probe contexts — one novel setting or several. Multiple probes map how broadly the gain generalizes but cost effort and exposure opportunities.
- Timing — how early the probe is run. Early probes catch context-bound liking before it is mistaken for success; too early, and they test a gain that has not consolidated yet.
- Announced vs. unannounced — whether the participant knows a new context is coming. Unannounced probes read a more natural response; announced ones preserve agency and trust at some cost to naturalness.
- Indicator continuity — how closely the probe reuses the in-setting measures. High continuity makes the comparison clean; changing measures confounds context effects with instrument effects.
When it helps, and when it misleads¶
Its strength is that it exposes context-bound liking — the comfortable-only-here result that looks like success from inside the training setting and fails everywhere else. It operationalizes transfer of learning: whether a capacity or attitude built in one setting carries to different, untrained conditions, which is never guaranteed and must be tested rather than assumed.[1]
Its failure mode cuts two ways. A probe pitched into too harsh a context can reactivate the aversion it was meant to measure — the test becomes an over-dose that undoes the gain, especially if run before the softening has consolidated. Read the other way, a single passed probe can be over-generalized into "it transfers everywhere," when it has only shown transfer to one nearby context. The classic misuse is skipping the probe entirely and declaring victory on in-setting comfort alone. The discipline that keeps it honest is to start with a near context and widen only on success, to run the probe as a measurement rather than a challenge, and to treat a collapse as a prompt to add context variety to the exposure — not as the participant's failure.
How it implements the components¶
Generalization Probe fills the transfer slots — the move to a new context and the reading taken there, not the sequence or the starting record:
generalization_pathway— it is the deliberate step into varied, untrained contexts that tests whether familiarity built in one setting extends beyond it.aversion_and_satiation_monitor— it measures and classifies the affective response, but takes that reading in the fresh context to gauge transfer.
Its nearest twin, Valence Tracking Log, takes the same kind of reading inside the training setting over time and owns the starting record (baseline_valence_profile) — the Log watches the trajectory where practice happens, the Probe measures once where it does not. It does not build or dose the sequence (graduated_exposure_ladder, frequency_intensity_spacing_calibration, that's Graded Exposure Plan), and it does not scope the first small sample (disliked_target_definition, that's Familiarity Sampling Trial).
Related¶
- Instantiates: Negative-Mere-Exposure Reversal for Disliked Targets — the Probe is the transfer test that tells whether a bounded success is real or context-bound.
- Consumes: Valence Tracking Log supplies the indicator set and the in-setting readings the probe's fresh-context result is compared against.
- Sibling mechanisms: Graded Exposure Plan · Valence Tracking Log · Neutral/Positive Pairing Schedule · Voluntary Contact Session · Familiarity Sampling Trial · Backlash Stop Rule · Exposure Dose Curve
Editorial Notes¶
Form Classification¶
Form family: Experiment, Test & Rehearsal
Rationale: Generalization Probe operates as a bounded trial, probe, simulation, or rehearsal that generates evidence from performance because it tests whether reduced aversion survives outside the practice setting by deliberately measuring the response in a fresh, untrained context.
Independent corroboration: The frozen evidence defines Generalization Probe as 'Tests whether reduced aversion survives outside the practice setting by deliberately measuring the response in a fresh, untrained context', so its operative form is Experiment, Test & Rehearsal.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Psychology
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Stimulus generalization and transfer-of-learning experiments directly motivate testing behavior in a novel context.
Related originating lineages:
- Education & Pedagogy — Learning-science transfer research materially operationalized near and far transfer probes.
- Statistics & Experimental Design — Experimental design supplies held-out contexts and comparative measurement.
Review resolution: Both reviewers agree that psychology is primary: Stimulus generalization and transfer-of-learning experiments directly motivate testing behavior in a novel context. I retain education_pedagogy, statistics_experimental_design only as formative lineage, not as a list of later applications. I resolve origin_mode as cross_disciplinary_synthesis because the artifact joins distinct disciplinary contributions. I resolve domain_reach as multi_domain because it transfers across several fields but is not a domain-free primitive. Encyclopedia synthesis is false because the exact generalized packaging is already established enough that encyclopedia-specific synthesis is not required.
Review outcome: Reconciled after independent review; high confidence.
References¶
[1] Barnett, S. M., & Ceci, S. J. "When and Where Do We Apply What We Learn? A Taxonomy for Far Transfer". Psychological Bulletin 128(4), 612–637 (2002). Treats transfer as application of learning across changed contexts, with success conditional and empirically assessable. registry ↩