Learning

Learning is inferred from changes in behavior

People cannot directly observe another person's learning as an internal object. They can observe and measure behavior before, during, and after relevant experiences. A conclusion about learning should name the response, conditions, evidence, and limits.

For example, “Kai learned communication” is too broad to evaluate. A more useful statement is: “Across five snack routines this week, Kai independently selected an available AAC symbol for help in four routines, compared with zero of five comparable routines before teaching.” That statement documents change in the sampled conditions. It does not by itself prove that teaching caused the change, and it leaves maintenance and generalization open.

Learning includes much more than following adult directions. It can involve communicating a preference or refusal, using a chosen safety response, using mobility or communication supports, asking a healthcare question, reporting pain or discomfort, solving a problem, participating in a valued routine, or adults changing their own behavior to make the environment more accessible.

Learning, performance, mastery, maintenance, and generalization differ

These labels answer separate questions:

A person can perform accurately in one session because the materials are familiar, a prompt is present, or the setting is unusually supportive. A low score can also reflect pain, fatigue, inaccessible materials, absent AAC, unclear instructions, a changed preference, or withdrawn assent. The score needs context before anyone interprets it as learning or lack of learning.

“Mastered” is a program decision label, not a permanent property of a person. A descriptive analysis by McDougale, Richling, Longino, and O’Rourke reviewed 473 articles from three behavior-analytic journals published from 2015 through 2017 and included 157 skill-acquisition articles. Of those 157 articles, 26% did not report a specific mastery-criterion method, and more than half did not include maintenance probes. The paper supports stating the actual criterion and checking later performance. No universal mastery rule follows from it.

How learning is measured in ABA

The measure should match the response and the decision. Accuracy may fit a choice among defined options. Frequency may fit countable communication attempts. Duration may fit time engaged in a chosen activity. Latency may fit the time between a clear opportunity and a response. A permanent product may fit a completed schedule or prepared item.

Every percentage needs a denominator and a predeclared eligibility rule. Suppose nine opportunities are planned. An opportunity is eligible only when the cue and materials are present, ordinary communication and accessibility supports are available, the setting is safe, and Noor assents. Adults provide those conditions in eight; required AAC is unavailable in one. Report adult access implementation as 8 of 9, or 89% when rounded, and record the invalid opportunity separately. If Noor independently uses a help message in six of eight eligible opportunities, report 6 of 8, or 75%. Do not create exclusions after seeing a low score.

Useful evidence can include:

• baseline performance before teaching
• performance during teaching, with prompt levels reported
• independent probes with examples reserved from direct instruction
• maintenance checks after a stated interval
• generalization checks across relevant people, materials, or settings
• the person's experience, assent, dissent, and preferences
• adult and environmental actions that affect access

The BACB BCBA Test Content Outline (6th ed.) identifies respondent and operant conditioning, observational learning, measurement, experimental design, generalization, maintenance, and data-based decisions as examination content. The outline describes knowledge and skills expected for entry-level BCBAs. The document cannot determine whether a specific person learned a specific skill.

A fictional example

Amara is a fictional fourteen-year-old who wants a reliable way to ask for a quieter space during an after-school program. She chooses a short AAC message and a hand signal as portable options.

Before measurement, the team defines an eligible opportunity as naturally occurring noise within a predeclared range or low-risk practice Amara agreed to, with both communication options, ordinary supports, and a trained responder available. Noise is never raised solely to provoke distress, and Amara may stop at any time. An independent response is either chosen form within the agreed interval without an adult prompt. Leaving, turning away, or another established dissent signal is honored, recorded separately, and never blocked or scored as a skill error. Access to a quieter space does not depend on the score.

Across five eligible baseline opportunities, Amara uses a chosen form independently in one, or 20%. After teaching and environmental changes, ten eligible opportunities occur across three sessions: eight include an independent chosen form, one includes a prompted form, and one includes no chosen form. Any other established communication or dissent in that opportunity is recorded and honored separately. Adults honor seven of the eight independent messages, or 88% when rounded, and document the remaining event as an adult implementation failure that requires repair.

The pattern supports learning under the measured conditions. It does not by itself establish mastery unless the predeclared criterion was met, maintenance, or generalization. The team schedules a later check and samples an untrained activity leader or setting before making those claims. If Amara decides the goal or message no longer fits, her preference triggers revision rather than a failure label.

Teaching should support useful, chosen change

The Association for Behavior Analysis International basic-principles curriculum places respondent and operant processes, consequences, stimulus control, and related concepts among foundational training topics. Those principles can help explain changes in behavior. They do not supply a person's goals or authorize a procedure.

The public CASP Version 3.0 summary describes standards of care intended for planning, implementing, and evaluating ABA assessment and treatment for autism. Access to the full guideline requires a licensing agreement. The current BACB Ethics Code for Behavior Analysts applies to BCBA and BCaBA certificants and applicants. It addresses appropriate client and stakeholder involvement, required informed consent and applicable assent, assessment and intervention design, client and stakeholder preferences, risk minimization, correct data collection and use, and continual evaluation.

A technically measurable change can still be a poor goal. Teams should ask whether the learning expands communication, access, safety, autonomy, participation, relationships, or another outcome the person values. They should also ask whether changing the environment or teaching partners would be more helpful than placing every demand on the learner.

Questions to ask about a learning claim

• What exact response changed, and under which conditions?
• What baseline or comparison supports the conclusion?
• Are prompts and ordinary access supports reported separately?
• What are the numerator, denominator, dates, settings, and exclusions?
• Could health, access, context, or partner behavior explain the score?
• Has the response been checked for maintenance and generalization?
• Does the person find the goal and teaching process useful and acceptable?
• What evidence would cause the team to revise its conclusion or plan?