HCD.13 — Test Human Capability Retention and Dependence on Tools or Other Support

Type: Human Capability Development practitioner method pattern Status: Stable

HCD.13:1 - Problem Frame

Use this when earlier success is being relied upon after an interval, or when the person now performs differently with changed tools or support. Begin with a qualified earlier observation of the same person’s contribution and the conditions under which it succeeded. Then separate the delay question from the support question.

The first useful result is a retention and support-dependence account: what remains demonstrable after which interval and practice history, under which support, and what any apparent loss does or does not show. The practical gain is to avoid both assuming that earlier success lasts indefinitely and prescribing relearning for an access or configuration failure.

Retention concerns an already demonstrated contribution. If that earlier performance was never established, assess the current contribution through HCD.11. Use HCD.12 when the central question is unfamiliar work. If a known missing source or failed tool already explains the current inability to perform, return that arrangement problem to its owner before diagnosing loss.

Source-dependent work can remain well performed with its sources. An independently available core is a separately justified demand, not the person’s only “real” capability. Stronger safety, credentialing, employment or deployment reliance requires the relevant domain evaluation and assurance.

HCD.13:2 - Problem

A delayed failure can reflect many changes besides loss of capability: no recent practice, different tasks, inaccessible sources, unfamiliar tools, missing cues, fatigue, or a changed criterion. Testing only after a delay while also removing support confounds the two questions.

A successful delayed test can also be overread. Continued use may have maintained the contribution; a warm-up or earlier probe may have taught or primed the answer; a short core may cover only a few contributions. None of those observations establishes retention without practice, whole-course mastery or transfer to other work.

HCD.13:3 - Forces

HCD.13:4 - Solution

HCD.13:4.1 - Recover the earlier contribution and the intended interval

Recover the person, contribution, criterion, earlier observations, relevant task variation, allowed means and evidence window. A remembered impression that “the person could do it” is not a qualified baseline. If the earlier evidence cannot support the intended comparison, assess now and establish an honest current starting point.

Choose the interval from the work: how long might this contribution be unused, what support will remain available, and what consequence would a failure have? There is no universal retention interval. An infrequent high-consequence task may need a different observation and assurance basis from a routine task.

Recover intervening use, practice, instruction, source changes, tool use and relevant uncertainty. Report continuing practice as continuing practice. An uncertain history narrows the conclusion rather than becoming an assumed period of non-use.

HCD.13:4.2 - Compare delay under reference conditions

Specify a delayed task that elicits the same contribution at a comparable demand and criterion. Use the earlier qualified support conditions, or explain the deliberate change and its effect on the inference. Preserve task access and distinguish material differences from harmless surface variation.

Observe the delayed attempt before remedial teaching when the retention question needs that observation. Keep help, incomplete work, material errors and corrections separate. If the criterion or required source has changed, assess the affected new requirement rather than interpreting every difference as human decline.

The direct comparison is earlier versus delayed performance under the stated reference conditions. It can support retention in that tested envelope and history. It does not isolate the cause of the result from every intervening influence.

HCD.13:4.3 - Test support dependence as a separate question

State which support matters and why. A tool contrast is useful when the person must sometimes contribute without that tool, when a fallback is required, or when the assessor needs to distinguish the person’s contribution from assisted output.

For a reduced-support core, specify its own contribution, data, allowed means, time and criteria. Preserve required accessibility support. Do not remove a current technical source merely to make a source-dependent task appear independent.

A minimum interpretable comparison distinguishes:

Control order, cueing, rehearsal and fatigue as far as the receiving use requires. Use fresh comparable material or separate occasions when an earlier task could supply the answer. Record unavoidable order effects; repeated testing is not inert. A positive independent core supports only its stated contribution, not all work previously performed with a tool.

HCD.13:4.4 - Investigate apparent loss only when the distinction changes the next action

First check that the present demand is inside the earlier claim. Return an outside-envelope demand or known configuration failure directly.

When at least two live explanations would lead to different development or support work, and probing is safe and informative, use a qualified E.23.CAE differential. Recover a previously successful or separately qualified reference condition without further teaching. Vary the smallest condition that distinguishes the rivals. A reference–changed-condition–reference comparison can be useful when order or transition history matters.

Observe whether the person recognizes the applicable response, can access it, adapts it, enacts it and achieves the required result only where the evidence distinguishes those contributions. Do not infer every earlier failure from an incorrect final product.

If the response reappears without relearning under the qualified reference condition, a simple global-loss explanation is weakened in those conditions. That recovery does not prove unfamiliar transfer or identify a memory mechanism. If the probe teaches the answer, support changes cannot be separated, or safety prevents the contrast, return the unresolved distinction.

A warranted claim of changed capability requires holder-specific evidence after the relevant envelope and configuration rivals have been addressed. A single unsuccessful return probe is not enough by itself.

HCD.13:4.5 - Return separate conclusions and a practical next move

Keep the following separately reportable: delayed performance under reference support; performance in any independent core; observed support dependence; a supported access, adaptation or enactment difficulty; possible loss or warranted revision of a capability claim; and unresolved or omitted tests.

State the interval, practice history, conditions, observations, inference and limits. The next useful action may be ordinary continued use, a targeted observation, access repair, a revision hypothesis through HCD.14 or a direct specialist return. The evidence account does not choose every later action on the person’s behalf.

Stop at the smallest supported conclusion. For stronger reliance, obtain the relevant baseline quality, task comparability, observation coverage, independent judgement and protection against probe-induced learning. Extra follow-up time, access and assessor effort need an actual arrangement; specifying an eight-week check does not obtain them.

Reopen only the affected claim when delay, use history, source, tool, task, criterion or intended reliance changes.

HCD.13:5 - Archetypal Grounding

HCD.13:5.1 - A matched eight-week comparison with an independent core

Learner-L17 and all observations in this example are constructed. The example stipulates a separately resourced follow-up with an assessor and source access. It neither schedules an actual programme visit nor supplies observations of Engineer-E27 or a learner cohort.

At reference time T0, a qualified assessment records three selected contributions associated with F.1, C.16 and C.11. The engineering tasks require HW3/FW8. A supplier report for FW7 cannot justify FW8; a complete HW3/FW8/SCHED2 report can support the stated configuration. SCHED2 requires two hours of checks per change and one hour of preparation per homogeneous batch of up to four. Ten changes require 23 hours, twelve require 27; at 24 available hours, a ten-change proposal leaves two of twelve new changes in the weekly queue. At 22 hours, nine changes require 21 and ten cannot be promised.

The existing alternatives are A, a second stand still requiring a shared-power check; B, the supplier option with the FW7 report; and C, the compatible SCHED2 arrangement with the stated capacity limit. Supplied financial assumptions give A/B/C approximate NPVs of 48.9/51.6/44.6 thousand. B’s financial rank cannot settle its compatibility.

Two stipulated weekly logs each record ten accepted changes, all required checks completed and twelve arrivals; together they show four additional queued changes. Those observed counts and periods remain distinct from the modelled 23-hour requirement. The source set contains these logs, the exact configuration reports and the calculation assumptions. The release manager is the chooser for release/use; L17 prepares a bounded recommendation. These are local exercise facts.

The delayed assessment is eight weeks after T0. The constructed history records two intervening source-selection uses and one comparison/decision-support exercise, with no rehearsal of the exact follow-up cases. Thus the question is retention under intermittent practice, not eight weeks of non-use.

These are selected contributions, not a demonstration of the whole F.1, C.16 or C.11 Method envelope.

At both T0 and follow-up, a separate fifteen-minute core uses supplied tables and a calculator but no AI. It asks L17 to recover the required configuration, notice an unsupported report, and reconstruct a feasible time-limited proposal. In both constructed observations, L17 distinguishes the configuration and time constraints and explains why ten cannot be promised at 22 hours.

The core runs before the source-dependent task on different material, with no corrective teaching until both are complete. The report retains that order. It supports the narrow core under those conditions and the later source-dependent performance after that core; it is not evidence that task order never matters. A receiving use requiring an entirely unwarmed first attempt would need that observation.

The result is retention of the compared contributions under the stated eight-week history and support conditions, plus the separately tested core. It supports neither untested contributions nor retention without practice, and does not establish an actual learner result or a causal effect of the development arrangement.

HCD.13:5.2 - A failed source link changes the arrangement result

During an additional constructed source-dependent task, the required technical source is inaccessible. The earlier qualified contribution depended on using that source. The assessor can verify the access failure directly.

Return source access to the responsible provider and leave the dependent performance judgement incomplete. There is no reason to run a memory-loss diagnosis to explain a known missing input. If a later restored-access attempt is observed, report it under the restored condition without retroactively completing the interrupted attempt.

HCD.13:5.3 - Recovery after a cue narrows an apparent-loss claim

In a separate constructed comparison, the person does not select a familiar Method for the current problem. Without teaching its content, the assessor supplies a previously qualified task cue; the person then produces the familiar response. If that cue is a legitimate reference condition and the contrast addresses the live rival, the result supports availability under that condition and a possible selection difficulty.

It does not show uncued selection, unfamiliar transfer or the mechanism of remembering. If supplying the cue also explains the solution, the no-relearning comparison fails and the loss-versus-access distinction remains unresolved.

HCD.13:6 - Bias-Annotation

An assessor may prefer a simple story of forgetting or resilience. Resist it when support, access, practice or task conditions changed. People with required assistive technology should not receive a lower capability judgement merely because the assessment removed it.

Repeated probes can favour people who learn from the test itself. Record relevant practice and order effects rather than interpreting every later improvement as stable retention.

HCD.13:7 - Conformance Checklist

• Earlier qualified evidence concerns the same person and contribution.
• The chosen interval and intervening practice history are explicit.
• Delay and support questions have separate interpretable comparisons, or the combined-condition limit is stated.
• A reduced-support core has its own justified target and preserves required access arrangements.
• Help, feedback, unfinished work and probe reactivity remain visible.
• Any apparent-loss differential uses a qualified reference and addresses live rivals without hidden teaching.
• Retention, support dependence, access difficulty and capability-loss claims remain distinct.
• The result names tested coverage, unresolved questions, resource needs, stop and reopen conditions.

HCD.13:8 - Common Anti-Patterns and How to Avoid Them

HCD.13:9 - Consequences

The practitioner can preserve supported performance, repair access and target a genuine uncertainty without restarting development indiscriminately. A short independent core may answer an important fallback question while leaving normal source-dependent work intact.

The cost is a recoverable baseline, practice history and a meaningful comparison. Some distinctions remain unresolved without another observation. A cleaner-looking “retention score” is worse if it hides tool withdrawal, teaching during the probe or a narrower tested contribution.

HCD.13:10 - Rationale

Elapsed time is a condition of the comparison, not proof of its cause. Support dependence is likewise not synonymous with a deficit: it may be appropriate to the required work. Separating the questions makes the next action local, while a qualified reference return prevents apparent loss from automatically becoming a relearning prescription.

HCD.13:11 - SoTA-Echoing

Working question: what can a delayed assessment show when practice and support can also change performance?

Selected line — adopt and adapt. Carpenter, Pan and Butler’s review supports spacing and retrieval as bounded learning approaches; the classroom-testing meta-analysis by Yang and colleagues reinforces that testing conditions can affect learning. HCD.13 therefore records intervening practice and probe reactivity in §4.1–§4.4 instead of treating a test as a passive reading.

Serious alternative: compare an earlier assisted score with one delayed unassisted score. It is inexpensive and can describe that combined condition, but it cannot distinguish delay from assistance withdrawal. The selected line keeps a reference-support comparison and adds only the support contrast needed for the decision. Its extra cost is explicit; an unaffordable contrast returns a narrower result rather than a false loss claim.

The Bastani and colleagues AI study motivates the support distinction with school-mathematics evidence, not adult occupational or long-retention evidence. The exact mutation is §4.3’s separate independent-core target and §5.1’s two reports. E.23.CAE supplies the reference-return discipline when an apparent-loss differential is actually needed.

No cited source supplies a universal interval, a local programme effect or the truth of this fictional case. Reopen the conclusion if task comparability, practice history, required access or teaching during the probe changes the inference.

HCD.13:12 - Relations

HCD.11 supplies or recovers the prior and current performance evidence. HCD.12 concerns meaningful unfamiliarity, which may coexist with delay but needs its own claim. HCD.14 can use a retention or support finding to revise the affected development assumption.

E.23.CAE governs a qualified apparent-loss differential; a known configuration failure returns to A.15.8 and the direct owner instead. E.23.CDI supplies the capability-development evidence distinction under its holder and Work-family conditions; a retained response does not by itself establish an achieved development intervention.

Target-domain criteria and specialist results remain necessary for correctness. The programme or instrument owner selects and resources particular visits and tasks. Measurement, causal and stronger assurance claims require their own applicable specifications and evidence; this Method creates no common resilience score.

HCD.13:End