Backward chaining

Backward chaining starts from a task analysis

The team first describes the skill as an ordered set of observable actions. Each step needs a clear start, completion criterion, ordinary materials, relevant safety conditions, and accessible response form. The analysis should match how the person wants to do the task.

For making tea, steps might include choosing a cup, adding a tea bag, obtaining prepared hot water with the agreed safety support, pouring, steeping, removing the bag, and carrying the drink to a chosen place. Age, mobility, vision, motor access, health, and preference can change that sequence.

The learner completes the last unmastered step

During early teaching, the support person completes all earlier steps and pauses at the final step selected for instruction. The learner receives the planned cue, time to respond, and any needed prompt. A successful completion reaches the task’s natural outcome, such as drinking the tea, starting the song, or using the finished item.

When data meet the plan’s advancement criterion, the next earlier step joins the learner’s part of the chain. The already learned later step follows it. This sequence can make the outcome occur soon after the learner’s response.

Prompts and completion support need a plan

Prompt type, timing, and fading are separate design choices. The clinician may select modeling, gesture, visual support, verbal instruction, or physical guidance when appropriate and consented to. The plan should define how a prompt is delivered, how help is requested or declined, and what causes the team to fade, change, or stop it.

Early steps still get completed somehow. Teacher completion, manual guidance, and learner completion of untrained steps create different practice, session-length, and independence experiences. Describe which approach is in use rather than treating “backward chaining” as the whole procedure.

Procedure choice depends on the person and task

Backward chaining can fit a task whose final step has a clear, valued result, or a learner who benefits from contacting that result quickly. Forward chaining may fit when the natural cue at the beginning is especially useful. Total-task presentation offers practice on every step each time.

Choice should reflect the person’s goals, current skills, preferences, stamina, motor access, safety, prompt history, and opportunities outside teaching. A task can also be redesigned so fewer steps or better tools reduce effort.

Research illustrates variation

In a small study, Jerome, Frantino, and Sturmey taught three adults with developmental disabilities a 13-step internet task using backward chaining with a most-to-least, errorless prompt sequence. Independent steps increased and performance generalized to another computer. Because the intervention combined several components and involved three people, it cannot isolate backward chaining or predict another learner’s outcome.

Bancroft and colleagues compared three ways of completing untrained steps during forward chaining with seven autistic children. All produced learning, while efficiency varied across participants and measures. The study concerns forward chaining, yet it underscores a practical point that applies here: completion procedures and the learner’s response matter.

Ethics and access stay active

The BACB BCBA Test Content Outline, 6th edition includes chaining, prompting, prompt fading, observable goals, contextual fit, integrity, and evaluation. It is examination content, not a clinical protocol or authority to practice.

For covered behavior analysts, the current BACB Ethics Code addresses competence, client and stakeholder involvement, consent and assent when applicable, assessment-based intervention, risk, data, and ongoing evaluation. Preserve AAC, breaks, mobility, essential care, and an accessible way to pause or withdraw.

A fictional backward chain

Imani chooses a seven-step routine for preparing an art playlist before drawing. At a review point, Imani completes the last three steps independently across the day’s one eligible opportunity. Earlier steps receive the planned support.

Current independent-chain coverage is 3 of 7 steps. That ratio describes which steps met the operational definition in that opportunity. It does not show that backward chaining caused the performance, that the full routine is mastered, or that the skill transfers to another device. The team keeps step-level prompts, errors, pauses, and Imani’s feedback visible.

Measure learning and fit separately

Useful records include independent steps divided by eligible steps, prompt level by step, complete chains divided by eligible opportunities, duration from the defined start to finish, and performance in uncoached probes. Report skipped opportunities, assent withdrawal, equipment failures, safety stops, and teacher completion separately.

Ask whether the person values the completed skill, whether effort is reasonable, and whether the routine works with familiar people, places, and materials. Independence can include self-selected tools and ordinary supports.