HCD.3 — Diagnose Whether a Capability, Misconception, or Behaviour Limits Work
Type: Human Capability Development practitioner diagnostic pattern Status: Stable
HCD.3:0 - Use This When
Use this pattern when a named person’s later-Work demand is recoverable, performance is inadequate or unstable, and distinguishing plausible explanations could change the finding or subsequent work.
The first useful result is either a qualified intervention target, the smallest changeable human limitation supported by current evidence, or a non-training return explaining an external constraint, identifying missing evidence or returning a specialist question to the practice that can answer it when known. For a target, state the grounds, any plausible rivals, the conditions protected during observation and what would reopen the finding. A non-training return explains what the evidence settles. For a conclusion that remains blocked, identify the missing input and what would permit reconsideration.
Use the diagnosis to decide whether human development work is warranted. Use HCD.1 when the later-Work demand itself is unclear. Use HCD.4 after a compatible demand and claim-level evidence exist and simultaneous capability claims must be compared. A separate programme comparison (HCD.2) may use a target or another qualified basis; HCD.3 stops at the target or non-training return.
Clinical diagnosis, choice of intervention, claims of development, surveillance, assignment, safety, employment and provider decisions each require evidence and authority suited to that use.
HCD.3:0.1 - Working Terms
| Name in this pattern | What it denotes |
|---|---|
| diagnostic question | A bounded question about which live explanation best accounts for an observed later-Work gap and would affect the warranted conclusion or the next Work. Clinical diagnosis belongs to the clinical practice. |
| rival branch | A materially different explanation whose plausibility matters to the diagnosis, choice of observation, responsible source or development question. |
| human limitation | A difficulty this person has in making a required contribution under the named later-Work conditions. Current evidence supports treating that difficulty as changeable and relevant to the decision. |
| qualified intervention target | The smallest supported human limitation, with evidence, any still-plausible rivals, conditions, intended observation, specialist boundary, and reopen condition. “Intervention target” names what a later choice might try to change; it selects no intervention. |
| non-training return | A finding that identifies an external constraint, a question outside the demand envelope or missing evidence for a human target. If another practice must supply a result, return the question there when known. |
| protected condition | A condition held stable or bounded so that an observation can discriminate rivals without quietly changing the holder, task, tool, authority, or support arrangement. |
| observed target change | A difference in the selected target between observations under stated conditions. It can inform the next development decision before its cause is established. |
| target engagement | A change in the selected target attributable to an intervention. Evidence must concern that target and justify the attribution; the label assigned to an activity cannot show that the target changed. |
The diagnosis concerns the person’s difficulty making the required contribution under the named conditions. A score, result, behaviour description or inferred state may identify a difficulty worth examining. Test what could account for it before identifying a change the person needs to make.
HCD.3:1 - Problem Frame
When Work goes badly, the person performing it is easy to see and may be easier to ask for a change than those who provide the tools, access or assignment. Yet the same observation can arise because the demand was wrong, the Method or source was missing, a tool or cue changed, the assignment or authority failed, workload or recovery was inadequate, a specialist return was absent, the response was not recognized as applicable, or the person’s capability really is limiting.
These branches have different owners and remedies. More practice can be useless when the task lies outside the claimed envelope, unsafe when a clinical or safety condition is current, and misleading when assistance or prompt design controls the result. Conversely, an endless list of possible causes is not a diagnosis. Compare the plausible explanations whose difference matters to the finding or subsequent work.
HCD.3:2 - Problem
The common shortcut is to transform “the result was poor” into “the person needs training.” It hides the evidence chain, turns a familiar intervention into a cause, and exposes the person to monitoring or employment consequences without a justified target.
The opposite shortcut refuses any bounded target while uncertainty remains. That makes practical diagnosis impossible even when controlled observations have eliminated the decision-changing rivals. The practitioner needs a conclusion that identifies any supported target, the plausible alternatives and the questions that need another contributor. Even if the evidence rules out the alternatives examined, state the conditions and limits of the finding.
HCD.3:3 - Forces
- A familiar human explanation is actionable, while Method, tool, authority, organization, environment, or specialist constraints may be more decisive.
- One observation is cheap, but competing explanations may still matter to the diagnosis.
- A controlled probe improves discrimination, while the probe can teach, prime, fatigue, expose, or otherwise change the person.
- Assisted performance reveals a joint result, while the human limitation may concern applicability, access, adaptation, enactment, or independent challenge.
- Motivation language is easy to apply, while change in the proposed target needs evidence beyond a theory label or assigned technique.
- A narrow target supports later design, while a diagnosis that is too broad becomes a trait label or a curriculum in disguise.
- Privacy, consent, worker governance, clinical safety and employment authority can constrain how observations may be collected or used.
- A specialist boundary protects the person, while HCD must still preserve any legitimate non-clinical Work question.
HCD.3:4 - Solution
Compare plausible explanations in the detail needed to distinguish the findings or next actions they would support. Use existing evidence where it suffices. If further observation is needed, seek a safe, attainable observation that can distinguish the explanations. State the smallest supported human limitation, explain an established external constraint or identify the evidence still missing. Give a question that requires another practice to the practice that can answer it, when known.
HCD.3:4.1 - Bind the Diagnostic Use
Name:
- the person whose contribution is in question and the basis for their participation;
- the HCD.1 demand account or equivalent qualified later-Work input, including configuration, conditions, horizon, and evidence window;
- the observed gap and the finding or development decision it could affect;
- the evidence collector, declared use, access conditions, privacy or worker-governance boundary, and the responsible parties where these conditions require their action;
- the protected conditions and safe probe limit; and
- the point at which a clinical, safety, legal, employment, target-domain, or other specialist owner must take over.
If the demand is absent or incompatible, stop the diagnosis and return the missing scope or evidence needed to consider this person’s difficulty. A role label, population profile or imagined task does not establish the demand.
HCD.3:4.2 - Compare Decision-Changing Rivals
Consider explanations that are plausible for this gap and could affect the diagnosis or the next Work. The table gives examples.
| Rival branch | Question that keeps it live |
|---|---|
| demand or envelope | Is the demand specified correctly and inside the capability envelope being relied upon? |
| applicability, access, expression, adaptation, or enactment | Is the response unavailable, or does a cue, context, interference, configuration, or performer arrangement prevent selection or use? |
| Method or description | Is the target Method missing, stale, disputed, or inapplicable? |
| tools and supports | Are AI, information, cues, access, environment, workload, recovery, or another support condition limiting? |
| assignment and organization | Is assignment, permission, authority, coordination, organization, provider access, or a required specialist result missing? |
| specialist condition | Does health, disorder, safety, ethics, law, or another specialist condition require its direct owner? |
| human change target | Does the evidence support a particular capability, knowledge, skill, discrimination, misconception, behaviour, self-regulation, or commitment as limiting and changeable? |
| measurement artefact | Could a scoring, recording, or interpretation error explain the poor indicator without the corresponding failure in Work? |
Use evidence about an actual condition or a consequential gap to decide which explanations warrant comparison.
Too few or poorly linked observations, coupled conditions, contamination, or unsafe collection can prevent this comparison from distinguishing the live branches. Narrow the conclusion or give a non-training return that names the evidence limit, what observation could distinguish the remaining explanations and any conditions for obtaining it safely.
Look for a missing intermediate performance, not just an absent basic skill or an unknown whole procedure. FPF B.1.5.EW can help identify how the necessary parts work together. In a stipulated dance case, the person has sufficient strength, holds the axis when stationary and recalls the figure, but loses balance during its rotation. Those observations locate a coordination question between the muscular actions and the figure; they do not establish a general lack of strength or memory. If existing observations already distinguish the decision-changing rivals, use them. Otherwise choose a suitable comparison of the rotation in isolation and in the figure, under domain-qualified conditions. Return the supported limitation or the unresolved alternative rather than prescribing more of whichever exercise is easiest to name.
HCD.3:4.3 - Probe Before Declaring Capability Loss
When a prior qualified capability or result exists and the differential can change the next Work, use an applicable E.23.CAE disposition before declaring loss or prescribing more practice. Use an applicable reference observation when it establishes the contrast. If a further probe is needed and can be made safely, return to the reference conditions without intervening development, vary the smallest consequential condition, and distinguish applicability, access, adaptation, enactment and result.
A recovered response under the reference conditions weakens the claim of loss across all conditions. Transfer requires evidence in the target conditions. Failure across reference and target conditions may warrant capability-claim revision only after demand, configuration, support, and measurement rivals have been addressed. If the probe teaches, primes, fatigues, changes exposure, or cannot be run safely, narrow the conclusion or return an unresolved differential that states the observation limit and its effect on the conclusion.
For a new observation, identify who will collect and interpret it, how the person will participate, its declared use, protected conditions and applicable access and privacy limits. Treat an assisting System’s inference as decision input. Acting on it requires separately established authority for the particular action.
HCD.3:4.4 - Record the Result
A qualified intervention target contains:
| Result position | Required content |
|---|---|
| use boundary | Human holder, later-Work demand, configuration, horizon, evidence window, receiving question or decision, participation basis, and governance limits. |
| observed gap | Claim-sized Work observation, intended result or critical-error boundary, and why the gap matters. |
| target | The person’s difficulty with the required contribution and the smallest supported human change to target under the named conditions. A later decision selects any practice or other intervention. |
| evidence | The observations, work products or contrasts used, how they support the target, and any uncertainty, contamination or measurement limits. |
| rival disposition | The explanations considered, any still-plausible alternatives, and their effects on the target and next observation. When none remain, keep the evidence limits and reopen condition explicit. |
| protected conditions | Method, task, tools, prompts, information, time, environment, assignment, authority, supports, and specialist routes held or bounded. |
| intended observation | The target change to observe, its conditions, and the claim that would depend on it. Explaining a result as an effect of the intervention through that target requires evidence for each link. |
| boundaries and continuation | Direct specialist or owner boundary, unsupported uses, and the condition that reopens the target. |
Observed performance may be reported with its conditions before a causal explanation is established. A better score alone establishes neither a change in the proposed target nor the intervention’s effect through that target.
A non-training return contains:
| Result position | Required content |
|---|---|
| actual owner or missing result | The established external constraint or the specific input still needed, with the party responsible for any further action where known. |
| conclusion and limits | What the external condition establishes and which human-target conclusion or development decision, if any, remains unsupported. |
| evidence and rival disposition | What is known, what remains unresolved, and why a human target is not supported. |
| safe interim boundary, when needed | While a conclusion is unresolved, state which assignments, inferences, prescriptions or other uses the available evidence does not warrant. |
| reconsideration | If a conclusion remains blocked, name the result, observation, decision or evidence that would permit a retry. For a settled finding, identify a change that would matter if it were revisited. |
HCD.3:4.5 - What Changes in Practice
Ask which explanation affects the warranted conclusion or the next Work, which evidence distinguishes it and whether the evidence supports a human change. If an external condition explains the difficulty, state its effect. For a missing input or specialist decision, say what is needed and return the question to the practice able to address it when known.
HCD.3:5 - Archetypal Grounding
HCD.3:5.1 - Constructed Positive Case: ImpactReview-17
ImpactReview-17 is a constructed case. ImpactReview-17-Domain-Oracle-v1 supplies engineering correctness and criticality criteria for PumpWorks-Config-Baseline-42 and AI-Assist-M4. The baseline observations and neutral-probe results below are stipulated case data.
For human holder Engineer-P14, the constructed HCD.1 account demands independent review of AI-proposed change-impact claims under a forty-five-minute limit with current sources, configuration trace, auditable evidence, and bounded safety and maintenance specialist returns. The baseline stipulates correct source/configuration trace and evidence return in eight of eight cases, correct specialist routing in two of two relevant cases, and rejection of two of four plausible wrong AI claims. Both misses occur in unfamiliar variants and are labelled safety-critical only by the constructed oracle.
The applicable MethodDescription, source access, tool configuration, time, assignment, authority, and specialist route are present in the case, so those branches do not explain the two misses. An E.23.CAE-shaped neutral probe preserves those conditions. Engineer-P14 recognizes that independent checking is required but does not generate and test a competing configuration hypothesis on either unfamiliar miss.
HCD.3 therefore returns this qualified intervention target:
Generate and test an independent configuration counter-hypothesis before accepting plausible AI output in unfamiliar variants.
The stipulated review permits an ordinary reminder to construct and test a competing hypothesis, but the engineer must supply the case-specific rival and its test. In the proposed comparison, the neutral prompt is “Review this AI claim against the current configuration.” The other prompt adds “Before accepting it, construct and test a competing configuration hypothesis.” Neither supplies the alternative or tells the engineer which claim to reject.
The evidence is narrow: two unfamiliar misses inside eight constructed cases. A prompt-design rival remains live because the neutral prompt may have suppressed the counter-hypothesis move. The protected conditions are the same source, configuration, AI assistant, time limit, authority, and specialist routes. The intended observation is whether Engineer-P14 produces and tests a competing configuration hypothesis and rejects the planted wrong claim in each of two further unfamiliar cases: one first attempt with a neutral prompt, and one first attempt on the other case with an explicit counter-hypothesis prompt. Success with the reminder supports performance under that permitted condition and may justify returning prompt design to the Method/tool owner; it does not establish self-initiated use of the operation. Different cases leave case-difference uncertainty; the comparison informs the next decision without isolating a causal prompt effect. Reopen when the prompt, configuration, case family, source access, tool posture, time, observation count, or target-domain oracle changes.
The constructed result identifies a target for a later development decision. Operational use needs qualified current domain criteria and observations for the actual person and Work, including evidence that distinguishes the live rivals.
HCD.3:5.2 - Missing Case: ReleaseWork-R9
Engineer-E27 has only a prospective HCD.1 demand hypothesis for ReleaseWork-R9. The actual Work occurrence, occurrence-produced result claim, occurrence-bound result record, representative observations, and compatible actual SYSE.21 supplier result required by XRI-21 are not recoverable.
HCD.3 returns a non-training return with diagnostic-evidence-missing. The blocked decision is whether a changeable human limitation accounts for the difficulty. A MethodDescription, plan, target repertoire, curriculum, constructed demonstration, course completion, or platform score cannot discriminate the rival branches.
Do not base training assignment, an inference of human deficiency, employment consequences, or a transformation claim on the unsupported HCD inference. Separate decisions require their own sufficient evidence and appropriate authority. Retry after a compatible HCD.1 demand basis and holder-specific evidence distinguish the live demand/envelope, Method, tool/support, assignment/authority, specialist, and human-target branches under the intended configuration and evidence window. Recovering a Work occurrence alone does not complete the diagnosis; the observations must be decision-discriminating.
HCD.3:5.3 - Unlike Case: Social-Anxiety Clinical Route
When an adult seeks treatment for social anxiety, return that question to qualified clinical care. A diagnostic label, symptoms, avoidance, or treatment request alone does not establish a target for changing a separately specified contribution in later Work. The cited NICE social-anxiety guideline (BEH-05) places recognition, assessment, treatment choice, provider competence, safety, shared decision-making, and outcome monitoring with qualified clinical care.
HCD.3 returns a non-training clinical route whenever diagnosis, treatment choice, exposure or CBT content, clinical safety, consent, provider competence, or clinical outcome interpretation is current. A separately posed, non-clinical question about later participation can remain. Report what participation was observed and with what support. A claim about participation in another setting or a later outcome needs evidence from that use. Retry HCD only when the person and clinical owner supply a bounded non-clinical Work question whose evidence can be used safely without taking clinical authority.
HCD.3:6 - Bias-Annotation
| Recurring bias | Likely drift | Repair |
|---|---|---|
| training-default bias | A poor result becomes a course or practice assignment. | Compare plausible explanations with different implications for the finding or subsequent Work and return the smallest supported target or a non-training result. |
| dispositional bias | A configuration-dependent observation becomes a stable trait, attitude, or motivation label. | Bind holder, Work, conditions, horizon, observations, and any still-plausible rivals. |
| tool-blame or human-blame bias | Responsibility is assigned to the AI tool or person before the joint arrangement is separated. | Distinguish the contributions of the person, tool, Method, prompt, authority and specialists. Vary a condition in an authorized comparison if that would provide needed evidence. |
| familiar-mechanism bias | “Forgot,” “misconception,” or “not motivated” substitutes for a controlled differential. | Use E.23.CAE where applicable and retain any still-plausible explanation that could change the next decision. |
| target-rhetoric bias | Naming autonomy, confidence, motivation, or practice is treated as evidence that it is limiting. | Require holder-specific evidence for the proposed target. A later claim that an intervention worked through that target also needs target-engagement evidence. |
| clinical appropriation bias | A disorder or treatment process is recast as development. | Return diagnosis, treatment, provider, safety, consent, and clinical outcomes to qualified clinical ownership. |
| surveillance bias | More data collection is assumed to be harmless and authorized. | Name declared use, participation, access, privacy or worker-governance conditions, and direct authority before collection. |
| intervention leakage | The target wording quietly names a programme, provider, schedule, or method. | Name the person’s difficulty with the required contribution, the smallest supported human change to target, and the intended observation; select any intervention separately. |
HCD.3:7 - Conformance Checklist
- The person whose work is in question and a compatible HCD.1 account or equivalent qualified later-Work input are recoverable.
- The observed gap, configuration, horizon, evidence window, receiving question or decision, and protected conditions are explicit.
- The comparison covers the credible explanations that could change the finding or the next Work.
-
An applicable
E.23.CAEdisposition is used before capability loss or more practice is declared when its differential can change the decision. - Method, tool, AI, cue, information, access, environment, workload, recovery, assignment, authority, organization, provider, specialist, and human-target branches keep their own kinds.
- Any qualified target names the person’s difficulty with the required contribution and the smallest supported human change under the named conditions.
- The result explains how the evidence supports its finding, which alternatives were considered, which remain plausible and which conditions were protected. It identifies any observation or specialist contribution still needed, the limits of HCD’s judgement and what would reopen the conclusion.
- A non-training return states the external constraint, specialist question or missing result and the limits of the finding. If a conclusion remains blocked, it names the interim boundary and the condition for retry.
- Evidence collection names collector, declared use, holder participation, access, privacy or worker-governance boundary, and required direct owners.
- The result authorizes no surveillance, assignment, employment action, specialist action, treatment, or development intervention.
- The result claims neither target engagement nor transformation before later evidence supports it.
- The receiver can identify any further observation needed, any applicable interim restriction and the responsible source of a returned question where known.
HCD.3:8 - Common Anti-Patterns and How to Avoid Them
| Anti-pattern | Why it fails | Repair |
|---|---|---|
| “The score is low, so train the skill” | The score does not identify the required contribution, where the person has difficulty, or the rival explanations. | Recover the later-Work account and compare decision-changing branches. |
| Prompted success is treated as evidence of required unaided use | The prompt may have supplied the selection the person was supposed to make. | Record what the prompted performance supports. Probe an unresolved contribution only when the receiving use requires it and the observation is safe and worthwhile. |
| “They missed twice, so capability was lost” | Small, configuration-bound evidence cannot eliminate support, prompt, envelope, or measurement rivals. | Preserve uncertainty and any still-plausible rival; narrow or return the target. |
| “Motivation is the cause” | Naming motivation supplies neither an observation of change nor a causal explanation. | Specify the proposed target and observe it. A claim that an intervention changed it and thereby changed a later result needs grounds for both links. |
| “The AI made the error” or “the human made the error” | The joint Method, prompt, source, review, and authority arrangement is hidden. | Separate the contributions using available evidence. If a consequential difference remains unresolved, use a safe, attainable comparison that can distinguish the contributions, or state what remains unresolved. |
| “A clinical technique is just another learning intervention” | HCD lacks diagnosis, treatment, provider, consent, safety, and clinical outcome authority. | Return the case to qualified clinical ownership and retain only a legitimate non-clinical Work question. |
| “The target is practice” | Practice is an activity intended to produce a change; it does not identify the person’s difficulty. | Name the difficulty making the required contribution, what must change, and the observation that would show that change. |
| “Diagnosis chooses the programme” | A target is evidence for a later decision, not a programme or authorization. | Keep programme comparison (HCD.2) separate; stop at the target or non-training return. |
HCD.3:9 - Consequences
A development decision can focus on one evidence-supported target instead of a generic deficit, or stop early when the real constraint belongs to another owner. Rival explanations and governance remain visible, reducing unnecessary practice, unsafe probing, and employment overreach.
Collecting and interpreting any additional observations costs time and effort. Some cases remain unresolved, and a narrow target may reopen when prompt, tool, task, support, evidence, or specialist information changes. Using a qualified target to design development still requires a separate intervention choice. A claim that the intervention worked through this target requires later target-engagement and result evidence.
HCD.3:10 - Rationale
A supported target can suffice for the next development decision while the broader causal explanation remains unresolved. State which explanations were considered, which remain plausible, the limits of the evidence and what would justify revisiting the conclusion.
A non-training return can explain the difficulty and spare the person misdirected practice. When another practice needs to act, give it the question and the evidence it can use. Clinical, safety, legal, target-domain, assignment, and employment questions retain their own authority even when human development is also relevant.
HCD.3:11 - SoTA-Echoing
How can a practitioner identify a development-relevant human target without treating every poor result as a training need or a medical diagnosis?
The selected current line joins three bounded contributions. First, Melanson and Fahmie’s Functional analysis of problem behavior: a 40-year review (2023; BEH-01) supports the warning to distinguish maintaining conditions before intervening, but its clinical populations, controlled procedure, causal functions, and specialist safety arrangements are not generalized into HCD. Second, E.23.CAE separates envelope, configuration, applicability, access, adaptation, enactment, and actual capability change through an observation-first differential. Third, Sheeran and colleagues’ meta-analyses Self-determination theory interventions for health behavior change (2020; MOT-01) and Does increasing autonomous motivation or perceived competence lead to health behavior change? (2021; MOT-02) support an intervention-to-target-to-result evidence chain when explaining an intervention’s result through its proposed motivational target. A theory label or assigned technique is not evidence that the target changed.
A serious alternative is the generic training-needs interview or competency-gap score. An interview or score can give people a quick account they can discuss. Use it as a starting point for comparing plausible explanations. Check what the measure shows about the person and what could instead reflect support or authority before naming a human target or a non-training result.
Human mechanism and intervention claims still require a direct source matched to the population, target, comparison, measure, horizon, support, and limit. Reopen when the holder, demand, configuration, prompt, tool, support, Method, assignment, authority, specialist result, evidence window, observation, declared use, or governance condition changes.
HCD.3:12 - Relations
HCD.1: supplies a same-holder demand account and evidence needs. HCD.3 may instead use an equivalent qualified later-Work input.E.23.CAE: supplies differential dispositions when apparent loss, access, expression, adaptation, or enactment would change the next Work. Its result is evidence, not an intervention choice.E.23.CDI: governs development-intervention and transfer evidence only after development is separately selected; HCD.3 does not perform that selection.- Melanson and Fahmie’s review (2023) (
BEH-01): supplies only the bounded “differentiate before intervening” warning; its clinical procedure and population remain outside HCD. - Meta-analyses by Sheeran and colleagues (2020, 2021;
MOT-01,MOT-02): supply target-engagement and intervention-to-target-to-result evidence boundaries; they do not make motivation the default cause. - The NICE social-anxiety guideline (
BEH-05) and direct clinical sources: provide criteria for diagnosis, treatment, provider competence, safety, consent, and outcome interpretation. Clinical specialists exercise professional judgement; the person and other authorized actors make the decisions that belong to them. HCD.3 supplies the stop-and-route boundary. - Target-domain Methods and specialist results: provide criteria and evidence for correctness, quality, criticality, hazard, legality, and other domain questions in the diagnostic comparison. Qualified specialists make the professional judgements, and authorized actors decide the corresponding actions.
HCD.4: can use a qualified target and its remaining plausible alternatives when comparing profile claims.- Programme comparison (
HCD.2): can use a qualified target if the task calls for programme comparison. - Employment, assignment, safety, legal, provider, organization, platform, and tool owners: retain their direct decisions and authority.