Shaping · Atlas

1. Introduction

Version followed: a composite, since no single manual exists. Operant shaping follows Skinner (1953) and the standard behaviour-modification procedure in Miltenberger (2018, shaping chapter). Graded task assignment (GTA) follows Beck, Rush, Shaw & Emery (1979) and the Beck Institute staircase worksheet (J. S. Beck). Rehabilitation shaping follows the CI (constraint-induced) therapy protocol of Morris, Taub & Mark (2006). Where these differ, the guide says which one is being followed.

Shaping is a procedure, not a stand-alone therapy. It slots into ABA programmes, CBT, behavioural activation (BA) and rehabilitation. The core idea is differential reinforcement of successive approximations. The clinician:

starts from something the person already does;
reinforces versions that come a little closer to the target;
stops reinforcing the older, cruder versions.

This moves the person's behaviour step by step toward a response that was rare or absent at the start. Skinner described it as "reinforcing a series of successive approximations" (quoted in Wikipedia, citing Skinner 1953, pp. 92–93). GTA is the cognitive-therapy version. An overwhelming task is broken into a staircase of small, doable steps. Each completed step:

produces direct evidence of competence, which disconfirms beliefs like "I can't do anything";
restores contact with reinforcement (mastery);
builds momentum toward the larger goal.

[source: Wikipedia, Shaping (psychology); Simply Psychology; Beck et al. 1979, described from secondary sources]

2. Framework

Shaping has no fixed session count. It is a run of steps inside a wider treatment. The phases below combine Miltenberger's procedural steps with Pryor's planning rules. [source: Miltenberger 2018 via chapter flashcards; Pryor, Ten Laws of Shaping; synthesis for the phase grouping]

Phase A: Target and suitability

Goal: Define the terminal behaviour in observable terms, covering its form plus the dimension that matters (frequency, duration, latency, force, distance or accuracy).
Is shaping the right tool? Decide whether shaping is actually needed. It is not needed if instructions, modelling or prompting can already produce the target. It is also not needed if the target already occurs often enough to reinforce directly. [source: Miltenberger via flashcards]
Move on when: The target is written so that two observers would agree on whether it happened [source: behavioristbookclub, citing Parnell et al. 2017].

Phase B: Starting point and baseline

Goal: Identify a starting behaviour that the person already does at least occasionally and that is related to the target. Measure the person's current best level.
Format: Observation or self-monitoring over a few days. In CBT/BA this is the activity and mood diary, usually from session 1–2 [source: Emory BA protocol; Simply Psychology].
Move on when: The starting level is known and a first criterion can be set just beyond it, so that reinforcement is reachable within the first few trials [source: behavioristbookclub].

Phase C: Plan the steps and the reinforcer

Goal: Write the full sequence of approximations in advance, so the next criterion is ready if the person jumps ahead (Pryor's "plan ahead").
Choose a reinforcer: It must be deliverable immediately and in small amounts that will not cause satiation. Conditioned reinforcers such as a marker word, a click, tokens or specific praise help with both.
In GTA: The project is defined jointly and steps are ordered from simple to complex [source: Miltenberger via flashcards; Pryor; Beck et al. 1979].
Move on when: There is a written step list, a reinforcer and a rule for when to advance.

Phase D: Shaping proper (the advancing loop)

Goal: Reinforce the current approximation until it is reliable, then raise the criterion and stop reinforcing the previous level.
Format:
- ABA: Discrete trials across sessions.
- CBT/BA: A step is agreed in session, done as homework and reviewed next session.
- CI therapy: Sets of ten 30-second trials per task, with feedback on every trial, inside an intensive daily programme [source: Morris, Taub & Mark 2006]. The usual CI schedule is 2 consecutive weeks of several hours per weekday [uncertain: hours vary across CI protocol versions].
Advance criteria: Miltenberger requires that each step is mastered before moving on. One ABA secondary source suggests advancing after two consecutive sessions at the current criterion, and moving back if the earlier approximation re-emerges [source: behavioristbookclub, citing Ghaemmaghami et al. 2018]. In GTA the person moves up when the step was done and the person has acknowledged doing it [source: Beck et al. 1979].
- Practical default [synthesis]: Advance when the person meets the criterion on most opportunities across two consecutive occasions.
- Drop back [synthesis]: Go back one step after two consecutive failures.

Phase E: Maintenance and transfer

Goal: Make the target hold under natural conditions:
- thin the extra reinforcement;
- hand reinforcement over to natural consequences, which include mastery and the person's own recognition of what they achieved;
- add home practice and self-monitoring.
CI therapy formalises this as the "transfer package" [source: Morris, Taub & Mark 2006].
End when: The target occurs reliably in the person's everyday settings without therapist-delivered reinforcement [synthesis].

Typical placement in CBT for depression: Behavioural work (activity monitoring, scheduling, graded tasks) comes in roughly the first four sessions of a 16–20 session course, and earlier and more heavily for more severely depressed clients [source: Simply Psychology; Beck et al. 1979]. In the Emory BA protocol, GTA is a dedicated module (session 10) [source: Emory BA protocol].

3. Guiding principles

Start where the person is. [source: Miltenberger; Pryor; Morris et al. 2006]
- The first step must be something they already do.
- In practice, the first step should feel almost too easy.
Small increments. [source: Pryor; Miltenberger; Morris et al. 2006]
- Raise the criterion only as much as keeps reinforcement frequent.
- If success rates drop sharply, the step was too big.
Make each step a real approximation of the target. [source: Miltenberger]
- Each step must sit on the path to the goal.
- Do not reinforce side-steps just because they are easier.
Differential reinforcement, consistently applied. [source: Miltenberger; behavioristbookclub]
- Reinforce the current step.
- Stop reinforcing the previous one once the criterion has moved.
- Mixing the two stalls progress.
Immediacy. [source: Miltenberger; Morris et al. 2006]
- Reinforcement or feedback comes right after the behaviour.
- In session, mark the moment.
- For homework, review it at the start of the next session.
Don't linger, don't rush. [source: Miltenberger]
- Staying too long on one step locks it in.
- Advancing before mastery causes breakdown.
- Use an explicit advance rule.
One dimension at a time. [source: Pryor]
- Raise either difficulty or duration or speed, never two at once.
- When you add a new criterion, relax the others for a while (Pryor's "ease off the old").
Plan the whole route in advance. [source: Pryor]
- Have the next few steps written so that a sudden leap forward can be rewarded.
Consistency of shaper. [source: Pryor]
- One person sets the criteria for a given behaviour.
- With co-therapists, parents or nurses, agree the current criterion in writing.
Change method if stuck; return to basics if performance slips. [source: Pryor]
- If progress stalls, try a different route rather than pushing harder.
- After a lapse, go back to an easy step and rebuild.
End on success. [source: Pryor]
- Close a session, or set homework, at a level the person has just achieved.
Collaboration and ownership (GTA). [source: Beck et al. 1979; Beck Institute worksheet]
- The client co-designs the steps.
- The client credits their own effort for success.
- Watch for the client minimising what they achieved.
Action precedes motivation (BA/CT). [source: Simply Psychology; Emory BA protocol]
- Do not wait for the client to feel ready.
- Schedule the step and let mood follow.
Function and biology set limits. [source: Wikipedia, citing Breland & Breland]
- Shaping works poorly against strongly prepared or species-typical responses, or against behaviour maintained by stronger competing reinforcers.
- Check what currently maintains the behaviour before you start.
Shaping can build problems too. [source: Miltenberger via flashcards]
- Inadvertent shaping happens; for example, escalating attention for escalating distress.
- Look for it in families and wards.
Measure every step. [source: behavioristbookclub]
- Record the level achieved, not just whether it was done.
- Decisions about advancing rest on data.

4. Intervention inventory

The inventory has three groups:

A. Core operant shaping: Skinner and the behaviour-modification tradition.
B. Graded task assignment and related CBT/BA procedures.
C. Rehabilitation shaping: CI therapy.

All example language is illustrative unless it is quoted and cited.

A1. Target definition and shaping suitability check

Purpose: Decide what you are building and whether shaping is the right procedure.
When to use: Before any shaping programme.
Steps:
1. Write the terminal behaviour as an observable action with its key dimension. For example: "initiates a conversation with a colleague" or "sits through a full 45-minute class".
2. Ask whether the person can already do it when given instructions, a demonstration or a prompt. If so, use those methods instead of shaping.
3. Ask whether the behaviour already occurs often enough to reinforce directly. If so, reinforce it directly.
4. Ask what currently maintains the competing behaviour. If a strong competing reinforcer is present, plan to reduce it as well.
Example language (illustrative): "Let's get really specific. If I were watching a video of you on a good day, what exactly would I see you doing?"
Watch for:
- Vague targets such as "be more confident" or "stop procrastinating". These must become observable actions.
- A target is well defined when the client and therapist would score it the same way.
[source: Miltenberger via flashcards; behavioristbookclub]

A2. Identifying the starting behaviour and baseline

Purpose: Find the first approximation, which must be something the person already does and which is related to the target.
When to use: After A1.
Steps:
1. Observe the person, or have them self-monitor, for several days. Record the best current level of the target dimension (how long, how often, how close, how fast).
2. Choose the nearest related behaviour that already occurs at least occasionally. Examples:
  - sitting up in bed, as a step toward getting up;
  - opening an email, as a step toward replying;
  - a single vowel sound, as a step toward a word.
3. Set the first criterion at or just beyond the current best, so that reinforcement will come within the first few tries.
Example language (illustrative): "Over the next week, just notice and jot down the longest stretch you managed each day. No need to change anything yet."
Watch for:
- A starting point that is too ambitious. Signs are early non-completion and the client feeling ashamed.
- Baseline measures that are only pass/fail. Record the level reached instead.
[source: Miltenberger via flashcards; behavioristbookclub]

A3. Differential reinforcement of successive approximations (core procedure)

Purpose: Build a rare or new behaviour, bring back a lost one, or change a dimension (such as duration or force) of an existing one.
When to use: The target is absent or rare, and it cannot be prompted into existence directly.
Steps:
1. Write the step sequence from the starting behaviour to the target. Each step must be closer to the target than the last.
2. Reinforce the starting behaviour every time it occurs, immediately, until it occurs reliably.
3. Raise the criterion to the next step. From now on, reinforce only the new step. Stop reinforcing the previous one; this is extinction of the old approximation.
4. Expect variability, often a short burst of varied responses, when reinforcement for the old level stops. Reinforce whichever response comes closest to the new criterion.
5. Repeat until the terminal behaviour occurs reliably.
6. Then reinforce the terminal behaviour intermittently and move to maintenance (Phase E).
Example language (illustrative):
- In social-skills work with a withdrawn adolescent, the step sequence might be: looks up when greeted, then says hello, then adds one question.
- Specific praise only when the current step occurs: "You looked up and said hi to me when I came in. That's exactly it."
- Once "hi" is reliable, give warm but neutral acknowledgement for "hi" alone, and enthusiastic praise only when a question follows.
Watch for:
- Steady movement through the steps with high rates of reinforcement shows it is working.
- Frustration, stopping responding or rising problem behaviour usually mean the step was too large or the reinforcer has weakened.
- Getting stuck repeating one step means you lingered too long.
Historical illustration: Isaacs, Thomas & Goldiamond (1960) shaped speech in a mute inpatient diagnosed with schizophrenia. They reinforced eye movement toward a stick of gum, then lip movement, then sounds, then words [source: Isaacs et al. 1960; summarised from secondary knowledge, uncertain].
[source: Skinner 1953 via Wikipedia; Miltenberger via flashcards; Wikipedia procedure section]

A4. Conditioned reinforcer or marker (bridging signal)

Purpose: Deliver reinforcement at the exact moment of the approximation, without satiation or interruption.
When to use: Fast or brief responses, physical or motor work, children, and groups.
Steps:
1. Choose a brief, distinct signal. Options include a click, a word such as "yes" or "there", a token, or a specific short phrase of praise.
2. Pair it repeatedly with a back-up reinforcer so that it acquires value. Back-up reinforcers include tokens exchanged later, preferred activities, or tangible items for children.
3. Deliver the marker at the instant the criterion is met. Deliver the back-up reinforcer afterwards.
4. With adults in talk therapy, the marker is usually specific verbal acknowledgement. It must be contingent and must not be scattered.
Example language (illustrative): "Every time you hear me say 'there', it means that last bit was exactly the step we're working on."
Watch for:
- The marker losing value through over-use without back-up.
- Tokens working only in the clinic. Plan to hand over to natural reinforcers.
- Satiation, where too much of the same reinforcer is given.
[source: Miltenberger via flashcards (clicker, tokens, praise as conditioned reinforcers); Pryor; Peterson 2004 on Skinner's use of a conditioned reinforcer, uncertain]

A5. Shaping a dimension: duration, latency, rate, intensity or distance

Purpose: Change how long, how fast, how often or how forcefully an existing behaviour occurs, rather than its form. Examples:
- minutes on a task;
- time before starting work;
- voice volume;
- extending the interval between bathroom visits in bladder retraining;
- distance walked with a walker.
When to use: The behaviour already exists in some form but falls short on a dimension.
Steps:
1. Measure the current typical and best values.
2. Set the first criterion slightly above the typical value.
3. Reinforce any response that meets or exceeds the criterion.
4. Raise the criterion in small increments only after consistent success. Shape only this one dimension; hold others (for example accuracy) loose.
5. After a lapse, go back one step.
Example language (illustrative): "This week, the aim is just to stay with the reading until the timer goes, at the same length you already manage most days. Once that's easy we'll add a few minutes."
Watch for:
- Other dimensions deteriorating, for example quality falling as duration rises. That is expected for a while (Pryor: ease off old criteria).
- Re-tighten the other dimensions once the new level is stable.
Constraint: For eating, weight or exercise targets, shape the behaviour (for example tolerating a new food on the plate, or attending an agreed session). Do not shape toward numeric intake or weight targets.
[source: Wikipedia applications section; Pryor; Miltenberger; synthesis for the eating constraint]

A6. Percentile schedule (self-adjusting criterion)

Purpose: Set the criterion from the person's own recent performance instead of a fixed step list. This keeps the step size right automatically.
When to use:
- Degree-based targets such as on-task duration, work rate or response force.
- A fixed step list has stalled.
- Several staff need an objective rule.
Steps:
1. Keep a running record of the last several responses (for example, the last 10 durations).
2. Choose a reinforcement density. Galbicka's framework sets the criterion so that a chosen proportion of responses will qualify [uncertain: formula].
3. Reinforce a response only if it beats a set rank within that recent window. For example, longer than about 7 of the last 10.
4. Update the window after every response, so the criterion rises automatically as performance improves.
Example language (illustrative): "Your target today is simply to beat most of your own last few attempts. Not anyone else's, just yours."
Watch for:
- Steady upward drift in the running window shows it is working.
- It needs reliable measurement and is fiddly without a spreadsheet or app.
[source: behavioristbookclub, citing Kwak et al. 2022; Galbicka 1994, uncertain]

A7. Changing-criterion programme (stepwise targets)

Purpose: Move a behaviour that is already measurable through explicit, announced criterion levels. Each level holds until it is stable. This is often used for self-managed habits, such as reducing a behaviour or increasing study time.
When to use: The client can self-monitor and is motivated by visible targets. It is also a way to show experimentally that the contingency is doing the work.
Steps:
1. Baseline.
2. Set criterion 1 just beyond baseline, with a planned reinforcer for meeting it.
3. Hold the criterion until performance is stable at it, then announce the next level.
4. Vary the size of the steps and how long each phase lasts. If behaviour tracks each change, the programme is controlling it.
Example language (illustrative): "For the next two weeks the bar is here. If you hit it on most days, you earn the reward we agreed. Then we move the bar together."
Watch for:
- Behaviour that overshoots criteria. That suggests the steps are too small, so raise them faster.
- Repeated misses. That suggests the steps are too large.
[uncertain: Hartmann & Hall 1976 is the usual primary citation for the changing-criterion design. It is a research design that clinicians use as a stepwise shaping frame.]

A8. Shaping combined with prompting and fading

Purpose: Speed up shaping by adding instructions, modelling or physical or verbal prompts, then withdrawing them.
When to use: Shaping alone is slow, or the person can partly do the step with help.
Steps:
1. Prompt the current approximation, using the least intrusive prompt that works.
2. Reinforce the prompted response.
3. Fade the prompt over trials, while keeping the criterion where it is.
4. Raise the criterion only once the step occurs without the prompt.
Example language (illustrative): "Watch me first. I'll ask the shop assistant one question. Now you try, and I'll stand beside you. Next time I'll wait by the door."
Watch for:
- Prompt dependence, where the behaviour only happens when the prompt is given.
- Fade prompts before you raise the criterion.
[source: behavioristbookclub, citing Aravamudhan et al. 2020; Morris et al. 2006 (coaching and modelling within shaping)]

A9. Stepping back (regression to an earlier approximation)

Purpose: Recover after performance breaks down.
When to use:
- Two or more consecutive failures at a step.
- Return of an older approximation.
- After illness, holidays or a crisis.
Steps:
1. Drop to the last step that was reliably achieved.
2. Reinforce generously and make sure reinforcement is easy to earn.
3. Hold that step until it is stable again. One secondary ABA source suggests two consecutive sessions.
4. Re-advance, possibly with a smaller step than before.
Example language (illustrative): "Let's go back to the step that was going well. That's not losing ground, it's how this method works."
Watch for:
- Clients reading the step back as failure. Frame it in advance as a normal part of the method.
[source: Pryor (return to basics); behavioristbookclub]

A10. Therapist differential reinforcement of in-session behaviour

Purpose: Shape clinically relevant behaviour as it happens in the room. Examples include:
- more specific self-report;
- expressing a need;
- staying with a topic;
- concise speech.
When to use: The target behaviour, or a crude version of it, can occur in session.
Steps:
1. Name the target class privately, for example the client stating a preference.
2. Notice the crude approximations, such as a hedged hint.
3. Respond to them with genuine interest, warmth or compliance. Give less of that response to the alternative behaviour, without punishing it.
4. Gradually require clearer versions before responding as fully.
5. Optionally, make it explicit with the client.
Example language (illustrative): "You just told me directly that you'd rather stop there today. I really appreciate hearing it straight. Let's do that."
Watch for:
- Mechanical or manipulative-feeling responses. Reinforcement in the room must be natural and genuine.
- Shaping the wrong thing by accident, for example giving more attention when distress escalates.
Note: Functional Analytic Psychotherapy develops this into a full therapy. It is not covered here.
[synthesis: applies the core procedure to the session; Isaacs et al. 1960 is the classic instance]

B1. Graded task assignment (Beck et al. 1979)

Purpose: Overcome inertia and the belief "I can't" in depression. The client meets a goal they think is out of reach by doing a staircase of tasks, and gathers direct evidence of their own competence.
When to use:
- The client is overwhelmed by or avoiding a task.
- The client is immobilised by low mood.
- The client predicts failure.
- Most useful early in treatment and with more severe depression.
Steps: The components below follow the sequence usually attributed to Beck et al. (1979, ch. 7). [uncertain]
1. Define the problem. The problem is typically the client's belief that they cannot do something that matters to them.
2. Formulate a project together. Choose a concrete goal the client values.
3. Assign tasks in steps from simple to complex. The first step should be one the client is very likely to manage.
4. Make success directly observable. The client sees for themselves that they did it, for example by ticking it off or showing the result.
5. Elicit doubts and minimising. Draw out the client's doubts, cynicism and discounting ("anyone could do that") and say them out loud.
6. Encourage a realistic evaluation of performance. Compare what happened with what was predicted.
7. Stress that the client reached the step by their own effort and skill.
8. Jointly set the next, more demanding step.
Example language (illustrative):
- "Cleaning the whole flat feels impossible right now, so let's not aim for that. What's the smallest piece of it you'd be fairly sure you could do before Thursday?"
- When reviewing: "You predicted you wouldn't manage even that. What actually happened?" and "Who did that?"
Watch for:
- Discounting ("it was only the dishes").
- Perfectionist step-skipping.
- The therapist setting steps the client didn't really choose.
Signs it is working:
- The completion rate is rising.
- The client volunteers the next step.
- The client's predictions of failure become less extreme.
[source: Beck et al. 1979; Simply Psychology; Brainscape behavioural-techniques cards describing GTA as a shaping strategy]

B2. Graded task staircase diagram (Beck Institute worksheet)

Purpose: A visual that turns an abstract goal into a sequence of steps. It cuts down the feeling of being overwhelmed by the end goal.
When to use: Alongside B1, especially with clients who keep focusing on how far away the end goal is.
Steps:
1. Draw a staircase. Write the goal on the top step.
2. Work backwards and forwards with the client to fill in the intermediate steps. The worksheet asks at minimum for the first step, the step before last and the last step.
3. Mark where the client is now.
4. Agree only the next step as homework.
5. Each session, move the marker up and remind the client that being partway up the staircase is progress.
Example: The Beck Institute worksheet's flat-hunting example runs from searching online, through working out affordable rent and phoning to book a viewing, to visiting several flats. These are paraphrased.
Example language (illustrative): "When you look only at the top step it seems miles away. Let's look at where you're standing and just the next stair."
Watch for:
- Steps that are still too big. If a step needs its own staircase, split it.
[source: Beck Institute GTA worksheet (Arabic translation, KSU-hosted); J. S. Beck, Cognitive Behavior Therapy: Basics and Beyond]

B3. Cognitive rehearsal of the steps

Purpose: Find obstacles in advance and improve follow-through. The client imagines doing each step of the assignment in sequence.
When to use:
- Before a graded task that the client has failed before.
- When poor concentration or forgetting are likely to derail the task.
Steps:
1. Ask the client to picture, step by step, doing the agreed task: when, where, what first, what next.
2. At each step ask what might get in the way, and agree a response to it.
3. Adjust the size of the steps if the rehearsal shows a gap.
Example language (illustrative): "Walk me through Thursday morning. You've just woken up. What happens next? What might stop you there?"
Watch for:
- Rehearsal that turns into rumination about failure. Keep it concrete and procedural.
[uncertain: cognitive rehearsal is described in Beck et al. 1979 as a behavioural technique for depression]

B4. Mastery and pleasure monitoring of graded steps

Purpose: Make progress visible and test discounting with data. This turns the client's own ratings into the reinforcer.
When to use: Throughout B1–B3, and alongside activity scheduling.
Steps:
1. The client logs each scheduled step and what was actually done.
2. The client rates each one for mastery (sense of accomplishment) and pleasure, each on a 0–10 scale. Remind them that mastery is judged relative to their current state, not their best ever.
3. Optionally, the client predicts the ratings beforehand and compares them afterwards.
4. Review in session: which steps raised mastery or mood, and which predictions were too pessimistic.
Example language (illustrative): "Given how low your energy was on Tuesday, how much of an achievement was that, out of ten?"
Watch for:
- Mastery ratings of 0 for completed tasks. That points to discounting, so return to B1 steps 5–7.
- Note that a 1–2 point mood rise counts.
[source: Simply Psychology; Emory BA protocol (accomplishment/pleasure 0–10); Brainscape cards (pleasure predicting, attributed to Burns)]

B5. Breaking tasks into concrete steps in behavioural activation

Purpose: Turn vague, avoided tasks into specific, scheduled small actions, and gradually approach what is being avoided.
When to use:
- Avoidance is maintaining depression, for example unopened bills or postponed calls.
- A scheduled activity keeps not happening.
Steps:
1. Identify an avoided task, and map its avoidance pattern: trigger, emotional response, avoidance behaviour.
2. Rewrite the task as a sequence of concrete actions. The Emory example breaks a debt to an insurer into: gather the bills, call about a payment plan, review the budget, calculate the payments, set up automatic payment.
3. Rate how difficult or distressing each step is. Pick a manageable first step.
4. Schedule it in a specific slot, with an alternative coping response ready for when the trigger comes.
5. Review the outcome, and problem-solve so the next step is realistic and small enough.
Example language (illustrative): "'Sort out the debt' is too big to put in a diary. What's the first physical thing you'd do: find the letters? Which drawer are they in?"
Watch for:
- Steps defined as outcomes ("feel better about money") instead of actions.
- Repeated non-completion of the same step means the step size needs fixing, not that the client should try harder.
[source: Emory BA protocol, TRAP/TRAC and GTA modules; Martell, Dimidjian & Herman-Dunn 2010]

C1. CI therapy shaping trials (rehabilitation)

Purpose: Get concentrated, repeated use of the more-affected limb after stroke or brain injury. This overcomes learned non-use. Skill on the task itself is a by-product.
When to use:
- Upper-limb hemiparesis with some active movement.
- Delivered by trained OT or PT staff. A psychotherapist's role is usually supporting adherence and the transfer package, not delivering the motor training [synthesis].
Steps: [source: Morris, Taub & Mark 2006]
1. Select tasks that target:
  - the joint movements with the most pronounced deficits;
  - the movements with the greatest potential for improvement;
  - then the patient's preference among tasks with similar potential. Pitch them at the upper end of current ability, without excessive effort.
2. Run each task as a set of ten 30-second trials, with rests.
3. After each trial, give immediate, specific, quantitative knowledge of results, such as the count or the time. Emphasise the positive.
4. Coach (cue or prompt specific movement improvements) and model the task at the start and between trials as needed. Encourage throughout.
5. Raise difficulty one parameter at a time. The parameters are target distance, target height, object size and speed.
6. Record trials and progress. The source gives no fixed rule for when to advance [uncertain].
Example language (illustrative): "That was 9 blocks in 30 seconds, up from 7. This time, try to straighten your elbow a little more on the reach."
Watch for:
- Fatigue.
- Frustration when tasks are set too hard.
- Compensatory trunk movement.
- Signs it is working: counts rising across trials, and more spontaneous use of the limb reported in the home diary.

C2. Transfer package (moving shaped behaviour into daily life)

Purpose: Move gains from the clinic into real life, and keep up adherence.
When to use: Alongside any intensive shaping programme. Adapting it beyond CI therapy is [synthesis].
Steps: [source: Morris, Taub & Mark 2006]
1. Monitoring: A structured interview on how much and how well the target is used in daily activities, plus a daily home diary.
2. Problem-solving: Review diary entries for times the target behaviour was not used. Find practical fixes, such as altering a task or adapting equipment.
3. Behavioural contract: A written agreement listing daily activities and the agreed way of doing each one. Signed by the patient, the therapist and a witness; revised as skills grow. A parallel caregiver contract sets out the carer's support.
4. Home skill assignment: The patient picks a set of daily activities, mixing easier and harder ones, to practise at home before the next session. The list is reviewed and renewed each session.
Example language (illustrative): "Let's look at the diary. Where did the plan slip, and what would make that one doable tomorrow?"
Watch for:
- Diaries filled in retrospectively.
- Contracts set at a level above what was achieved in clinic. Keep contract items at or just above the current step.
For psychotherapy [synthesis]: The same four elements adapt directly as a bridge from homework to life for GTA and BA.

5. Handling problems

Items marked [synthesis] are my own guidance and are not taken from the sources.

Stuck on a step (plateau).
- Likely causes: You stayed on an approximation too long. Reinforcement for the old level is still being given. Or the reinforcer has lost value.
- What to do:
  - Check that the old level now gets no reinforcement.
  - Raise the criterion by a smaller amount.
  - Refresh or vary the reinforcer.
  - If none of this works, change method, for example by adding prompting or switching to a percentile criterion.
- [source: Miltenberger; Pryor law 7; behavioristbookclub]
Breakdown after advancing.
- Meaning: The step was too big, or it was taken before mastery.
- What to do: Go back to the last step the client held reliably, re-stabilise it, and re-advance with a smaller increment (A9).
- [source: Miltenberger; Pryor law 9]
Client in CBT/BA does not do the homework step.
- What to do:
  - Treat non-completion as data, not defiance.
  - Ask what happened at the moment of starting.
  - Shrink the step until the client rates it as very likely to happen.
  - Rehearse it in imagination (B3).
  - Fix the time and place.
  - Consider doing the first step in the session.
- [source: Emory BA protocol (realistic, small enough); Brainscape cards (start small, frame as experiment); synthesis for the sequence]
Discounting success ("anyone could do that", "it doesn't count").
- What to do:
  - Name the discounting.
  - Ask the client to compare the step with what they did last week, not with how they functioned before.
  - Ask who did it.
  - Record mastery ratings (B4).
- [source: Beck et al. 1979, uncertain; synthesis for wording]
Overwhelm at the size of the goal.
- What to do:
  - Narrow attention to the next stair only (B2).
  - Set the long-term goal aside in writing so it is not lost but is not looked at every day.
- [source: Beck Institute worksheet rationale; synthesis]
Perfectionist step-skipping. The client jumps ahead and then crashes.
- What to do:
  - Agree in advance that doing more than the step does not change the next step.
  - Reinforce doing the step as agreed.
- [synthesis, consistent with Pryor laws 1–2]
Resistance to the rationale ("this is like training a dog", "I should be able to just do it").
- What to do:
  - Explain that all skill learning works in increments, as with physiotherapy or learning an instrument.
  - Point out that the client sets the target.
  - Make the reinforcers ones the client chooses or values, such as mastery, time or agreed rewards, rather than therapist approval alone.
- [synthesis]
Rupture, where the client feels managed, manipulated or patronised by praise.
- What to do:
  - Acknowledge it openly.
  - Make the contingencies transparent and joint.
  - Use specific, factual feedback ("you stayed 10 minutes") rather than effusive praise.
  - Ask the client what kind of acknowledgement feels genuine.
- [synthesis; Morris et al. 2006 favour quantitative knowledge of results]
Inadvertent shaping of problem behaviour. Examples are escalating self-harm threats, tantrums or symptom talk that bring escalating attention from staff or family.
- What to do:
  - Map who reinforces which level of the behaviour.
  - Give attention to calmer approximations of communicating the need.
  - Do not withdraw safety responses.
  - Any risk behaviour must still trigger standard risk assessment and a safety plan. Planned ignoring is never used for self-harm or suicidal communication.
- [source: Miltenberger (shaping can strengthen problem behaviour); synthesis for the risk rule]
Severe depression and suicidality during GTA.
- What to do:
  - Assess risk at the start and whenever mood drops. Use a collaborative safety plan, including lethal-means counselling (safe storage and removal).
  - Keep graded steps that involve isolating activities brief, and pair them with contact.
  - Failure on a step can feed hopelessness. Pre-empt this by making the first steps near-certain and by framing a missed step as a step-size error.
- [synthesis following standard risk practice]
Fear-driven avoidance on the staircase.
- What it is: If a step is avoided because of anxiety rather than inertia, it is really an exposure step.
- What to do:
  - Get informed consent to approach the feared situation.
  - Screen for contraindications: acute psychosis, unmanaged medical risk, active substance intoxication, current unsafe circumstances.
  - Agree that the client keeps control over pace, and plan the step with the client.
  - Consider a formal graded exposure protocol instead of GTA.
- [synthesis]
Competing reinforcers and biological limits.
- What it is: Shaping fails when the alternative behaviour is maintained by a stronger or more immediate reinforcer, or is strongly prepared. Breland and Breland's animals drifted back to food-getting behaviour.
- What to do:
  - Do a functional assessment.
  - Reduce access to the competing reinforcer, or make the target compete for the same function.
- [source: Wikipedia, citing Breland & Breland; synthesis for clinical translation]
Inconsistent shapers (parents, ward staff, co-therapists).
- What to do:
  - Keep one written current criterion.
  - Name one person who decides when to advance.
  - Hold brief handovers.
  - Use data sheets that record the level achieved.
- [source: Pryor law 6; behavioristbookclub]
Satiation or reinforcer fatigue.
- What to do:
  - Use smaller amounts, varied reinforcers and conditioned markers.
  - Shift toward natural reinforcers and intermittent delivery once the step is stable.
- [source: Miltenberger; Pryor law 3]
Interruptions and long gaps between sessions.
- What to do:
  - Expect slippage after gaps.
  - Start the next session one step back and with easy wins.
  - Avoid needless interruptions in the middle of a session.
- [source: Pryor laws 8–9]
Eating, weight and exercise targets.
- What to do:
  - Shape behaviours such as tolerating, tasting, attending or completing an agreed routine.
  - Do not set numeric intake, weight or exercise-load targets.
  - Coordinate with medical and dietetic care.
- [synthesis]

Sources

Sources: