Behavior Acquisition
Domain C is the largest single block on the exam — 19 of the 75 scored questions (25%) across eleven tasks, C.1 through C.11 — and it is the teaching half of your job: everything you do to build a skill the client does not yet have. It trips people up because the procedures overlap in practice. A single teaching trial can involve reinforcement, a prompt, a fading step and a token, and the exam will show you that trial and ask which procedure the RBT is implementing. The way through is to learn each procedure by what makes it different from its nearest neighbor, not by its definition in isolation.
What this domain asks you to do (C.1–C.11)
Eleven tasks, and ten of them start with "implement." The exception, C.9, asks you to distinguish maintenance from acquisition procedures, which tells you something about how the domain is tested: mostly by showing you a procedure being run and asking you to name it, correct it, or continue it. Note that the 3rd-edition outline split reinforcement into two tasks — C.1 for positive and negative reinforcement along dimensions such as magnitude and variety, C.2 for establishing and using conditioned reinforcers — where the 2nd edition had one. It also folded the old standalone stimulus-control-transfer task into C.7, which now names errorless teaching, least-to-most prompting, stimulus fading and time delay explicitly. Expect the exam to use those specific names rather than generic categories.
Reinforcement: positive, negative, and everything that makes it work (C.1)
Reinforcement is defined by its effect: a consequence that follows a behavior and makes that behavior more likely in future. Positive reinforcement adds something after the behavior — praise, a token, a preferred item, a high five. Negative reinforcement removes, reduces or postpones something after the behavior — the demand is withdrawn, the loud noise stops, the last two problems are canceled. Both increase behavior. That sentence is the most commonly missed fact in the whole exam, because "negative" sounds like a punishment word and is not; the plus and minus signs describe whether something was added or taken away, never whether it was pleasant. Four delivery variables decide whether reinforcement actually works. Immediacy: deliver within a second or two, because whatever the client is doing when the reinforcer arrives is what gets strengthened. Contingency: deliver it only after the target response, never as consolation for an unsuccessful trial. Magnitude, quality and variety: bigger and better reinforcers do more work, and rotating them prevents satiation — the outline names this "along a continuum of dimensions." Establishing operations matter too: a snack reinforces a hungry client and does nothing for one who just ate lunch, which is why the same item can look powerful in the morning and useless after break.
Schedules of reinforcement
The schedule is the rule that says which responses get reinforced. Continuous reinforcement (CRF) reinforces every correct response, and it is what you use while a skill is being acquired, because it builds the behavior fastest and gives the clearest feedback. Intermittent schedules reinforce some responses and not others, and they are what you move to once the skill is established, because intermittent reinforcement makes behavior far more durable — more resistant to extinction — and is closer to how the natural world works. Four basic intermittent schedules appear in plans. Fixed ratio (FR-5: every fifth response) and variable ratio (VR-5: on average every fifth response, unpredictably) are response-based. Fixed interval (FI-2 min: the first response after two minutes) and variable interval (VI-2 min: the first response after a varying average of two minutes) are time-based. Variable schedules produce steadier responding than fixed ones, because the client cannot predict which response will pay. The exam's practical version of all this is the thinning question: a client has just mastered a target, so what changes? You do not stop reinforcing; you move from continuous to intermittent, gradually, in the way the plan specifies.
Conditioned reinforcers and pairing (C.2)
Unconditioned reinforcers work without any learning history — food, water, warmth, sleep, relief from pain. Conditioned reinforcers acquire their power by being paired with reinforcers that already work: praise, tokens, stickers, a thumbs-up, the sound of a timer that signals break. Almost everything you use in a session is conditioned, which is why establishing them matters. Pairing is the procedure: you deliver the neutral stimulus together with, or immediately before, an established reinforcer, repeatedly, until the neutral stimulus alone functions as a reinforcer. Saying "nice work!" as you hand over the preferred snack is pairing praise; handing over a token immediately before the client trades it for a preferred item is pairing the token. The same principle explains why you spend the first sessions with a new client pairing yourself with reinforcement rather than placing demands — you are becoming a conditioned reinforcer, so that your presence and attention will later be worth working for. Two exam-relevant details: a conditioned reinforcer that is never again paired with anything will lose its effect over time, and a generalized conditioned reinforcer is one that has been paired with many different backup reinforcers, which is exactly what makes tokens and money so durable.
Discrete-trial teaching (C.3)
Discrete-trial teaching (DTT) breaks a skill into small units and teaches them in structured, repeated trials, usually adult-led and often at a table. Each trial has the same anatomy: the discriminative stimulus (the instruction or materials that signal the opportunity), a prompt if the plan calls for one, the client's response, the consequence (reinforcement for a correct response, or the error-correction procedure for an incorrect one), and then a short inter-trial interval before the next trial begins. What makes DTT powerful is trial density — you can run many more learning opportunities per minute than the natural environment offers — and consistency, because every trial looks the same. What makes it limited is that the skill is being practiced in a contrived arrangement with contrived reinforcers, so it does not automatically show up in the client's real day. That is why generalization has to be planned rather than hoped for, and why plans typically pair DTT with naturalistic teaching rather than choosing one forever.
Naturalistic teaching, and how to tell it from DTT (C.4)
Naturalistic teaching — incidental teaching, natural environment training (NET) — teaches the same targets inside the client's ongoing activity. The client's own interest sets the occasion: they reach for the bubbles, so the bubble jar becomes the teaching opportunity and a request is prompted; they hand you a closed snack container, so "open" is the target. The reinforcer is the natural one — the bubbles, the snack, the toy itself — rather than a token or an unrelated edible. That is the single cleanest test on the exam. Ask two questions of the scenario: who chose the activity, and what was the reinforcer? Adult chose it and the reinforcer was unrelated to the task, you are looking at DTT. The client's interest chose it and the reinforcer was the natural product of the response, you are looking at naturalistic teaching. Naturalistic teaching generalizes better and feels less like work, but it yields fewer trials per minute and depends on the client's motivation appearing, which is why you sometimes contrive it — putting a preferred item in view but out of reach so the opportunity to request arises. Neither approach is universally correct, and an exam option claiming one is "always better" is wrong on its face; the plan specifies which is used for which target.
Task analysis and chaining (C.5)
A behavior chain is a sequence of steps that together produce an outcome — handwashing, making a sandwich, getting dressed. The task analysis is the written breakdown of that sequence into steps small enough for this client. Three chaining procedures teach it. Forward chaining teaches step one to criterion while you prompt the client through the rest, then adds step two, and so on in order. Backward chaining teaches the last step first, with the client prompted through all earlier steps, then the second-to-last, working backwards; its advantage is that the client independently performs the step that immediately produces the natural reinforcer — the clean hands, the finished sandwich — on every single trial, which is why it is often chosen for clients who need early contact with success. Total-task chaining runs every step on every trial, with you prompting whichever steps the client cannot yet do independently; it suits clients who already perform many of the steps and only need help with a few. In all three, the completion of one step is the cue for the next — each step's product acts as the discriminative stimulus for the following one. If a scenario says the client is being taught the whole sequence each session with prompting as needed, that is total task; if it says one step at a time starting from the end, that is backward.
Shaping, and how it differs from chaining (C.10)
Shaping builds one new behavior by reinforcing successive approximations to it. You reinforce what the client can currently do, then require slightly more before reinforcing again, and keep raising the bar until the terminal behavior appears. Teaching a vocal approximation is the standard example: at first any vocalisation is reinforced, then only sounds beginning with the right consonant, then a closer approximation, then the word. Two rules keep shaping honest. Move the criterion up only when the current approximation is occurring reliably, and stop reinforcing the earlier approximation once the new one is established — otherwise the client has no reason to advance. If progress stalls, the usual cause is that the step you asked for was too big; the fix is a smaller step, and if the plan does not authorize you to adjust the criterion, that is a report to your supervisor. The exam pairs shaping with chaining constantly, and the difference is clean: shaping changes one behavior gradually until it becomes the target behavior, while chaining links several distinct behaviors into a sequence. If the client is producing successively better versions of the same response, it is shaping. If the client is producing step one, then step two, then step three, it is chaining.
Discrimination training and stimulus control (C.6)
Discrimination training teaches the client to respond in the presence of one stimulus and not in the presence of another. The discriminative stimulus (SD) is the one in whose presence the response is reinforced; the S-delta is the one in whose presence it is not. Reinforce responding to the SD, withhold reinforcement for responding to the S-delta, and over trials the response comes under stimulus control — the client responds when the SD is present and not otherwise. This is what every receptive-identification program is doing: "touch cat" is reinforced when the cat card is present and not when it is not. Practical points the exam likes: rotate the position of materials so the client learns the picture rather than the location on the table, vary the order of targets so responding is not simply a sequence, and watch for unintended stimulus control, where the client is responding to something other than the intended cue — your gaze, your slight lean, the fact that the correct card is always the newest one. If you notice a client is scoring perfectly but appears to be watching you rather than the materials, that is unplanned stimulus control and it belongs in your session note and your next supervision contact.
Prompts and fading: most-to-least versus least-to-most (C.7)
A prompt is help added to make the correct response happen; fading is the systematic removal of that help so the natural cue takes over. Response prompts act on the client's behavior and run roughly from most to least intrusive: full physical, partial physical, model, gestural, verbal (with position on the list varying by program and client). Stimulus prompts act on the materials instead — making the correct card larger or closer (position and salience prompts), then fading that alteration out until the materials look normal. That last procedure is stimulus fading, and the outline names it explicitly. Now the discrimination the exam cares about. Most-to-least starts with the level of help that guarantees success and fades toward independence. It is indicated when the skill is brand new, when errors are costly or dangerous, when the response is complex or physical, or when the client has a history of practicing errors — it minimizes mistakes but risks prompt dependence if you fade too slowly. Least-to-most starts with the least help — an opportunity to respond independently — and steps up only as needed. It is indicated when the skill is partly in the repertoire and you want to give independence a chance every trial; it produces more errors and takes more time. Errorless teaching is the most-to-least family taken to its conclusion: prompt immediately, so the client never has the chance to make the mistake, then fade the prompt in planned steps. Time delay is the fading mechanism that inserts a gap between the instruction and the prompt — constant time delay uses the same gap every trial (say two seconds), progressive time delay gradually lengthens it — giving the client room to respond independently before help arrives. Whichever hierarchy the plan uses, the exam's recurring failure mode is the same: a technician who keeps the prompt in place because the session goes better with it. That is prompt dependence, and the answer is to fade as the plan directs and to report the difficulty, not to invent a new fading schedule.
Error correction
Error correction is what you do when the response is wrong, and its purpose is to end the trial with the correct response having been performed, without delivering reinforcement for the error. Programs vary in their exact procedure — a common shape is to remove the materials briefly, re-present the instruction with a prompt sufficient to produce the correct response, deliver only neutral acknowledgment rather than full reinforcement for that prompted response, and then re-present the trial without a prompt to check for independence. Whatever your program specifies, run it exactly as written and identically every time, because inconsistent error correction is one of the fastest ways to stall an acquisition target. Two things never belong in error correction: reinforcement for the incorrect response, and any reaction that punishes the attempt. Saying "no, that's wrong" with a sigh teaches the client that trying is risky; a flat, neutral re-presentation teaches them what to do instead.
Generalization (C.8) and maintenance versus acquisition (C.9)
A skill that only appears in the therapy room with one technician and one set of cards has not been learned in any useful sense. Stimulus generalization is the skill occurring across new settings, people and materials; response generalization is the client producing untaught variations of the response — asking for juice using a phrase you never explicitly taught. Generalization is programmed, not assumed. The practical procedures are to teach with multiple exemplars rather than one set of materials, to teach across several people and settings, to teach loosely by varying the wording of instructions and the irrelevant details, to include stimuli from the natural environment in teaching, and to move toward reinforcers the natural world actually delivers. Maintenance is a different problem: keeping a mastered skill available after teaching has moved on. The signature of maintenance procedures is that the schedule is thinned, the target is checked periodically rather than taught, and mastered items are interspersed among acquisition targets — which also keeps the session's success rate high and behavioral momentum going. Task C.9 asks you to tell the two apart, and the test is what is currently being built: acquisition procedures teach something not yet in the repertoire, maintenance procedures preserve something already there. A common exam scenario shows a client failing a target that was mastered three months ago; the correct framing is that this is a maintenance problem to be recorded and reported, not evidence that the original teaching failed.
Token economies (C.11)
A token economy is a reinforcement system with three parts: the tokens themselves, the backup reinforcers they can be exchanged for, and the rules governing how tokens are earned and exchanged. Tokens are generalized conditioned reinforcers — they have no value in themselves, they acquire it through pairing with the backups, and because they can be traded for many different things they keep working across the whole session even as the client's motivation shifts. Running one well comes down to a few habits. Deliver the token immediately and contingently, exactly as you would any reinforcer, and pair it with brief praise that names the behavior. Keep the exchange ratio at what the plan specifies, and keep the exchange itself reliable — a token system where the client sometimes does not get the exchange stops working almost immediately, and repairing it is much harder than protecting it. Make sure the backup reinforcers are actually preferred right now, which is where your preference assessment feeds in. Response cost — removing earned tokens after problem behavior — is a punishment procedure, and you implement it only if it is written into the plan and you have been trained on it; it is not something you add because a session is going badly.
Key facts and definitions for Domain C
The list below is what to have automatic before the exam. Because this domain is a quarter of the test, the fastest gains come from the discriminations rather than the individual definitions — for each pair, be able to say in one sentence what separates them.
Worked scenario: name the procedure, then fix the fade
An RBT is teaching Sam to put on his jacket. The written program lists six steps and directs that all six be run each session, with the technician providing the least amount of help each step requires and reducing help across sessions. In session one the RBT used hand-over-hand help for steps three and four and gestures for the others. By session eight Sam completes five steps independently and still needs a light touch at the elbow for step four. The RBT notices the sessions run faster with the touch prompt and keeps using it for three more weeks; step four does not become independent. Four options: this is backward chaining and the RBT should switch to forward chaining; this is total-task chaining with least-to-most prompting, and continuing the prompt has produced prompt dependence, so the RBT should follow the plan's fading steps and report the stall to the supervisor; this is shaping, and the RBT should reinforce a closer approximation to independent jacket-donning; this is total-task chaining, and the RBT should stop prompting step four entirely so Sam has to do it. Work it in two passes. First, name it: all six steps are run every session with help as needed, which is total task by definition, and the help starts minimal and increases only as needed, which is least-to-most. That eliminates the backward-chaining and shaping options — shaping would apply if a single response were being refined, but a jacket has distinct steps. Second, choose between the two total-task options. Removing the prompt abruptly is a plan change made by the technician, and it also sets Sam up to fail the step he cannot yet do. Following the written fading steps while telling the supervisor that step four has stalled keeps fidelity and puts the decision about a program change where it belongs. The correct answer is the second option, and notice why the fourth is tempting and wrong: it identifies the right problem — prompt dependence — and then solves it by improvising.
Exam traps in Domain C
Domain C traps are almost entirely nearest-neighbor confusions. In each pair below, the exam will give you a scenario in which both procedures are plausible and one detail decides it. Learn the deciding detail, not the definition.
Key takeaways
This is the largest domain on the exam, so these eight sentences are worth more raw points than any other list in this guide.
Practice stays free. The full RBT (Registered Behavior Technician) study guide is the material itself, taught start to finish — a downloadable PDF + EPUB you keep.