Error correction

The procedure begins with a clear error definition

For one task, an error might be selecting a different symbol within five seconds. For another, it might be omitting a safety step in a multistep routine. A request for help, a pause, device failure, or assent withdrawal may have a different classification.

Define the instruction, response form, response window, correct-response criteria, ordinary supports, and invalid-opportunity rules before teaching. Clear definitions protect the learner and improve the usefulness of data.

Correction should be immediate, neutral, and teachable

A simple sequence might include a neutral acknowledgment, a model of the correct response, a prompted practice response, and another opportunity later. Another plan may use a controlling prompt immediately after the error and fade that help across future opportunities.

The correction should avoid ridicule, raised volume, prolonged repetition, or public attention. Tone, pace, and response effort matter. An efficient procedure helps the learner contact the relevant cue and correct response while keeping the task manageable.

Several design choices change the procedure

The clinician decides:

• whether to model, prompt, guide, or repeat
• whether the correction trial receives reinforcement
• how many practice responses occur
• when the original item returns
• how prompts fade
• what ends teaching for that period

These choices affect errors, practice, duration, and motivation. Saying “use error correction” leaves the teaching plan incomplete.

Correction and punishment are different concepts

An instructional correction is designed to teach a response after an error. Punishment is defined by a decrease in future behavior because of a consequence. A correction can still feel aversive, produce escape, or increase distress, so the label never settles the ethical question.

Monitor the learner’s communication, willingness, latency, affect, error pattern, and later independent responding. A rising error rate may signal a weak cue, poor prompt timing, inaccessible materials, task difficulty, health concern, or a goal that needs revision.

Errorless teaching is a related option

Errorless teaching arranges prompts or stimuli to reduce mistakes during acquisition. Error correction allows an error to occur and then follows the planned response. A program may use both across different phases or skills.

Selection depends on assessment, evidence, client preference, response cost, prompt history, discrimination difficulty, and context. The number of errors is one measure among many; independent learning and generalization also matter.

Research shows that details matter

In a randomized study of 28 autistic participants learning tact relations, Leaf and colleagues compared an errorless procedure with an error-correction procedure. Both were effective and efficient under the study conditions. The error-correction condition produced more independent correct responses and more independent errors. The narrow skill, sample, and procedures limit generalization.

Another small study involving four autistic children examined reinforcement after correction. Its results reinforce the need to specify the correction and consequence arrangement. Four participants cannot supply a universal reinforcement rule.

Clinical authority and safeguards stay visible

The BACB BCBA Test Content Outline, 6th edition covers prompting, prompt fading, teaching procedures, observable goals, contextual fit, procedural integrity, and evaluation. It is examination content, not a protocol.

For covered behavior analysts, the current BACB Ethics Code addresses competence, consent and assent when applicable, client involvement, assessment, risk, data, and continual evaluation. A qualified clinician chooses and revises the procedure within scope. Staff follow the written plan and route concerns.

Communication, AAC, food, water, bathroom access, mobility, prescribed care, breaks, and emergency help remain available. The learner needs an accessible way to request help, pause, or withdraw when that right applies.

A fictional correction record

Sora practices identifying ten unfamiliar pantry labels. Across ten defined errors, the teacher completes the planned two-step correction in 8 of 10. Two corrections omit the wait period and receive coaching. In a separate unprompted probe the next day, Sora responds correctly in 6 of 10 eligible opportunities.

Procedure completion is 8/10; later probe performance is 6/10. The denominators answer different questions. The figures do not prove that error correction caused the probe result. The team also records prompt level, error type, session duration, pauses, and Sora’s feedback.

Useful review separates teaching and testing

Coached correction trials, ordinary teaching trials, and unprompted probes should have separate labels. Report independent correct responses divided by eligible opportunities, corrections implemented as written divided by corrections due, prompt levels, repeated errors, duration, and transfer performance.

Define exclusions before data collection and keep invalid opportunities visible by reason. Improvement in one narrow set of items may coexist with weak everyday use.