RBT Behavior Acquisition Study Guide

Reinforcement and schedules

Reinforcement is defined by its effect: if the behavior it follows increases, it reinforced. "Positive" and "negative" describe whether something was added or removed — never whether it was pleasant. Adding a sticker that increases work completion is positive reinforcement; removing a demand in a way that increases break-requesting is negative reinforcement.

Unconditioned reinforcers (food, water, warmth) require no learning history. Conditioned reinforcers (tokens, praise, money) acquire their value through pairing. A token economy works entirely because the tokens are reliably exchangeable for back-up reinforcers the client actually wants.

Continuous reinforcement (CRF / FR1)

Every correct response reinforced. Fastest acquisition, least resistance to extinction. Use during acquisition, then thin.

Fixed ratio (FR)

Reinforcement after a set number of responses. Produces a post-reinforcement pause after each delivery.

Variable ratio (VR)

Reinforcement after an average number of responses. Highest, steadiest rate of responding and the greatest resistance to extinction.

Fixed interval (FI)

Reinforces the first response after a set amount of time. Produces a scalloped pattern — little responding early, a burst near the end.

Variable interval (VI)

Reinforces the first response after an average amount of time. Produces steady, moderate responding.

Discrete-trial teaching and naturalistic teaching

A discrete trial runs: SD → (prompt) → response → consequence → inter-trial interval. Deliver the SD once, clearly, then wait the specified interval. Repeating the instruction teaches the client to wait for the second or third delivery. When an error occurs, run the error-correction procedure in the plan — usually re-presenting with a prompt that guarantees a correct response — so the client rehearses the correct response rather than the error.

Naturalistic teaching (incidental teaching, NET) captures the client’s own motivation inside an ongoing activity, and uses the natural reinforcer. DTT gives you many more trials per minute; naturalistic teaching gives you far better generalization, because the teaching conditions resemble the conditions where the skill has to work.

Prompting and transfer of stimulus control

Prompts are temporary. The goal is always transfer of stimulus control: the natural SD alone eventually evokes the response. When prompts are delivered reliably and never faded, the prompt itself becomes the SD and the client learns to wait for help — prompt dependence.

Most-to-least

Start with the most intrusive prompt that guarantees success, then reduce intrusiveness. Minimises errors, so it suits brand-new skills.

Least-to-most

Start with the least intrusive prompt and add assistance only after an error. Maximises independent opportunities, so it suits partially acquired skills.

Constant time delay

A fixed pause between the SD and the prompt on every trial.

Progressive time delay

The pause lengthens across successive trials — one second, then two, then three.

Prompt hierarchy (least → most intrusive)

Gestural → verbal → model → partial physical → full physical.

Shaping, chaining and discrimination

Shaping is differential reinforcement of successive approximations toward a behavior the client cannot yet perform at all. If the client can already perform the response but needs help to do it, you need prompting, not shaping. Raising the shaping criterion too far too fast makes reinforcement unavailable and responding breaks down.

A task analysis breaks a multi-step skill into teachable steps. Forward chaining teaches step one first; backward chaining teaches the final step first, so the client always finishes the chain and contacts the natural reinforcer; total-task chaining runs the whole sequence every session with prompting as needed.

Discrimination training reinforces responding in the presence of the SD and withholds it in the presence of the S-delta. Watch for irrelevant stimulus control: if the correct card is always larger, or always on the left, or the RBT always glances at it first, the client can respond correctly without ever attending to the intended feature.

Verbal operants

Four verbal operants appear repeatedly, and the exam distinguishes them by what controls the response, not by what the words are. The same word "cookie" can be any of the four.

Mand

A request. Controlled by a motivating operation, reinforced by receiving the specific thing requested. To create mand opportunities, place a preferred item in view but out of reach.

Tact

Labelling something present. Controlled by a non-verbal stimulus, reinforced socially.

Echoic

Repeating what a speaker just said. Point-to-point correspondence with a vocal-verbal stimulus.

Intraverbal

Verbal behavior controlled by someone else’s verbal behavior without point-to-point correspondence — answering questions, conversation, fill-in-the-blank.

Generalization and maintenance

Generalization rarely happens by itself; it has to be programmed. Multiple-exemplar training — varying people, settings and materials during teaching — is the most direct method. Stimulus generalization is the skill occurring under new conditions; response generalization is new but functionally equivalent responses emerging.

Maintenance is performance after teaching ends. The usual cause of a maintenance failure is that reinforcement stopped entirely rather than thinning to a natural level. Periodic probes plus intermittent reinforcement keep a mastered skill alive.