The ABC Model: Antecedent, Behavior, Consequence

How the ABC model of behavior works

The word doing the work is contingency. The consequence happens because the behavior happened, and the behavior is emitted in the presence of the antecedent because that is where the consequence has followed before. A rat whose lever-presses produce food only while a light is on comes to press when the light is on and to ignore the lever when it is off.^[2] Nothing about the light forces the press; it matters only because of the history attached to it. The antecedent is the front of the loop, but the consequence is what loads it.

What is an antecedent?

An antecedent is anything in place before the behavior: the immediate stimuli, the wider context, and the state of the organism. Behavior analysts distinguish several kinds, because each is changed in a different way.

The discriminative stimulus (S^D) and S-delta

A discriminative stimulus, written S^D, is a stimulus in whose presence a behavior has been reinforced. Its counterpart, S-delta (S^Δ), is one in whose presence the same behavior has gone unreinforced. The lit "Open" sign is an S^D for pulling the door; the dark sign is an S^Δ. A friend's grin is an S^D for a crude joke; your grandmother's presence is an S^Δ for the same joke.

The S^D does not elicit behavior the way a puff of air elicits a blink. It evokes it — raises its probability — and only because of what has followed the behavior there before. Pavlov's stimulus produces the response; Skinner's merely signals that a response will pay. Operant vs. classical conditioning ›

Stimulus control, discrimination, and generalization

When a behavior's probability depends on whether a stimulus is present, the behavior is under stimulus control. It gets there through discrimination training — reinforcement in the presence of the S^D, extinction in the presence of the S^Δ — and its mirror image is generalization, responding to stimuli that resemble the S^D: pigeons reinforced for pecking a key lit with 550 nm light pecked less and less as the color moved away from it, tracing a smooth generalization gradient.^[3] Discrimination training also reshapes that gradient (peak shift), can be arranged so the learner never makes an error (the basis of prompting and fading), and extends to the context itself, which is why an extinguished behavior can renew when the setting changes back. Stimulus control in depth ›

Motivating operations

Jack Michael pointed out in 1982 that two different antecedent functions were being lumped together. A stimulus can signal that a reinforcer is available (the discriminative function) or change how much the reinforcer is worth (the motivational function).^[4] He called the second an establishing operation; the field later adopted motivating operation (MO) as the umbrella term, with an establishing operation increasing a reinforcer's value and an abolishing operation decreasing it.^[5][6]

Every MO has two effects at once: it alters the reinforcer's value and it alters behavior, evoking responses that have produced that reinforcer or abating them. Eight hours without food makes food a stronger reinforcer and makes you open the fridge; a large lunch does the opposite. A child who slept badly finds escape from demands more valuable, so escape behavior climbs. The vending machine says chips are available; hunger says they are worth having. MOs are the closest thing to "motivation" in the model, and unlike the folk concept they are manipulable: train the dog before dinner, not after.

Setting events

Setting events are more distant conditions — illness, poor sleep, an argument at breakfast — that change the strength of the immediate A-B-C relation without being part of it.^[7] Most can be re-described as motivating operations, but the term survives in schools because it reminds observers to look further back than the last thirty seconds.

Prompts

A prompt is a supplementary antecedent — an instruction, a gesture, a demonstration, physical guidance — added when the natural S^D does not yet evoke the behavior. "What do you say?" prompts "thank you" until the gift itself does the job. Prompts must be faded so control transfers to the natural stimulus; prompting without fading produces a learner who waits to be told.^[9]

What counts as a behavior?

Operational definitions and the dead-man test

A behavior is something the organism does: observable, measurable, and defined so that two observers would agree whether it occurred — objective, clear, and complete, in the textbook formula.^[9] The quickest screen is Ogden Lindsley's dead-man test from 1965: if a dead man can do it, it isn't behavior; if a dead man can't, it is.^[8] "Sit still," "don't interrupt," and "stop snacking" all fail; a corpse manages every one. They name the absence of behavior, and an absence cannot be reinforced. Rewrite them as what you want to see: "completes the worksheet while seated," "raises a hand and waits," "eats an apple at 3 p.m."

Response classes and the dimensions of behavior

An operant is a response class: every response that produces the same consequence, whatever its form. Pressing the lever with the left paw, the right paw, or the nose is one operant; asking, pointing, and grabbing are one operant if they all get the cookie. Behavior is defined by function, not by which muscles moved. It also has measurable dimensions — frequency, duration, latency (how long after the antecedent it begins), and intensity — and the problem decides which you track. A tantrum is a duration problem; a morning run is a latency problem, the minutes between alarm and front door.

Why "be more productive" is not a behavior

"Be more productive," "eat healthier," and "be a better listener" are labels for outcomes. They cannot be observed at a moment in time or counted, so nothing can be made contingent on them. The translation is always the same: find a specific, countable act that would, repeated, produce the outcome. "Open the document and write one sentence before checking email" has a start and an end, either happened or didn't, and can be followed by a consequence within seconds. Start with the smallest version, because that is the one that actually gets emitted — the logic of shaping and of every effective habit protocol.

What is a consequence?

A consequence is a stimulus change that follows a behavior. Its effect on the future frequency of the behavior — not its appearance or the intent of whoever delivered it — determines what it is. Consequences sort into four quadrants plus one non-event:

Immediacy and contingency

Two properties decide how much of the loop closes. Immediacy: a reinforcer's effect falls steeply as the delay grows from seconds to minutes, which is why trainers use a click and why a monthly paycheck reinforces almost nothing in particular.^[10] Contingency: the consequence must depend on the behavior. Pigeons fed on a timer regardless of what they did developed "superstitious" rituals from accidental pairings, and adding free reinforcers that arrive whether or not the animal responds weakens the behavior.^[11][12]

The four-term contingency

Because a reinforcer only reinforces when its value is high, a complete account adds the motivating operation: MO → S^D → behavior → consequence, the four-term contingency.^[9] It is 6 p.m. and you last ate at noon (MO); you walk into the kitchen (S^D); you open the fridge (B); you eat (C). Remove the MO with a late lunch and the same kitchen evokes nothing. Practitioners reach for all four terms because the fourth is often the easiest to change.

Changing the antecedent is the underrated lever

When most people say "behavior change" they mean changing the consequence. That works, but it acts at the hardest moment — after the old habit has already run. The antecedent acts before the moment of weakness. Skinner listed "changing the stimulus" among the basic techniques of self-control: remove the cue for an unwanted behavior, or put the cue for a wanted one where you cannot miss it.^[1] Charge the phone in the kitchen. Put the guitar on a stand, not in a case.

Two research programs outside behavior analysis reach the same conclusion. Peter Gollwitzer's implementation intentions are plans of the form "when situation X arises, I will do Y." Forming one links a concrete cue to a concrete response in advance, so the cue evokes the behavior with little deliberation; a meta-analysis of 94 studies found a medium-to-large effect on goal attainment.^[13] An implementation intention is a discriminative stimulus installed verbally — antecedent control's cognitive cousin.

Wendy Wood's habit research shows the other side. Habits are cued by context and survive on context rather than intention: students who transferred universities kept their exercise, reading, and TV habits only when the new setting resembled the old one, and habitual cinema popcorn-eaters ate stale popcorn as readily as fresh in a cinema but not in a meeting room.^[14][15][16] To keep a behavior, keep its cue constant; to lose one, change the cue.

Why notifications are weak antecedents

Held against the three-term contingency, most attempts to build a habit are all B: a behavior with a checkbox, no antecedent, and a consequence that may or may not do anything. Where there is an antecedent it is usually a push notification, and notifications are weak antecedents for three reasons. Habituation: a stimulus repeated without any differential consequence loses its power to evoke a response, which is why the fortieth buzz of the day is background noise.^[17] Many people switch them off. And, least appreciated, a notification is a discriminative stimulus for picking up the phone — the behavior that reinforcement has actually followed — not for flossing. The fix is a cue that already occurs reliably in the world: after the coffee, when you sit in the car, when the meeting ends. Building habits with all three terms ›

Three worked ABC examples

The model's value is that it hands you three places to intervene. Each example is analyzed once, then attacked at A, at B, and at C.

A child's tantrum in the supermarket

An employee who keeps missing deadlines

Your own doom-scrolling

ABC data collection in practice

The ABC recording chart

ABC recording means writing down, for each occurrence of a target behavior, what came immediately before and after. Formalized for field research by Bijou, Peterson, and Ault in 1968, it is now the standard first step in schools and clinics.^[18]

Ten to twenty entries usually reveal a pattern — here, shouting reliably follows academic demands and is reliably followed by their removal. Write what you saw rather than what you inferred, and fill in the "possible function" column after the observations, not before.

Functional behavior assessment

A functional behavior assessment (FBA) combines indirect methods (interviews, rating scales), descriptive methods (ABC recording), and sometimes experimental methods to reach a hypothesis about what a problem behavior gets or avoids; U.S. special-education law has required one in certain disciplinary situations since 1997. The output is a behavior intervention plan that changes the antecedents, teaches a replacement behavior, and rearranges the consequences — all three terms, deliberately.

Functional analysis and the four functions of behavior

ABC recording shows correlation; a functional analysis tests causation. In the landmark 1982 study, Brian Iwata and colleagues exposed nine children who injured themselves to systematically arranged conditions — adult attention contingent on self-injury, escape from demands contingent on it, being alone, and a play control — and measured self-injury in each.^[19] The same behavior was maintained by attention in one child, escape in another, and its own sensory consequences in a third. A later summary of 152 such analyses found escape the most common function, followed by social-positive reinforcement (attention or tangibles) and automatic reinforcement.^[20]

Practitioners therefore speak of four common functions — attention, escape, access to tangibles, and automatic (sensory) reinforcement — and the function dictates the treatment. Escape-maintained shouting is treated by teaching the child to ask for a break and honoring the request, not by sending him to the hallway, which is the very consequence keeping the behavior alive.^[21] How applied behavior analysis uses this ›

Common misconceptions about the ABC model

• It is not the ABC model of cognitive-behavioral therapy. Albert Ellis's ABC, from rational emotive behavior therapy, runs Activating event → Belief → emotional Consequence: the B is a thought and the C is a feeling.^[22] In the behavioral ABC, the B is an observable act and the C is an environmental event that follows it. Both date from the 1950s; they describe different things.
• The antecedent does not cause the behavior. It evokes it, and only because of a consequence history. Change the consequences and the same antecedent stops working.
• ABC data does not tell you the function. Descriptive records suggest hypotheses; only a functional analysis tests them.
• "Consequence" does not mean punishment. It includes reinforcement, punishment, and the absence of either.
• A → B is not classical conditioning. A conditioned stimulus elicits a reflex regardless of what the organism does; a discriminative stimulus sets the occasion for a behavior whose fate is decided by what follows.

Key takeaways

• The three-term contingency is antecedent, behavior, consequence. The consequence happens because the behavior happened, and the behavior is emitted in the presence of the antecedent because that is where the consequence has followed before.
• Antecedents come in kinds that are changed in different ways: a discriminative stimulus signals that a reinforcer is available, a motivating operation changes how much it is worth, and a prompt is a supplementary cue that must be faded. Adding the motivating operation gives the four-term contingency.
• A behavior is something observable and countable that a dead man could not do, defined by its function rather than its form. "Be more productive" is an outcome label; "write one sentence before checking email" is a behavior.
• A consequence is sorted by its effect on future frequency, not by its appearance or the deliverer's intent: reinforcement increases behavior, punishment decreases it, and extinction is the usual reinforcer no longer arriving. Immediacy and contingency decide how much of the loop closes.
• Changing the antecedent is the underrated lever, because it acts before the moment of weakness. ABC recording reveals patterns, but only a functional analysis tests what a behavior gets or avoids, and the function dictates the treatment.

Check yourself

A teacher sends Jamal to the hallway every time he shouts during math. Over the month, shouting increases. Is the hallway a punishment?

No. A consequence is defined by its effect on future behavior, not by how it looks, and shouting went up, so the hallway is reinforcing it: shouting produces escape from the math demand. The fix is to teach Jamal to ask for a break and honor the request, not to keep delivering the very consequence that maintains the behavior.

Your phone buzzes and you pick it up. Does the buzz cause the behavior the way a puff of air causes a blink?

No. The buzz is a discriminative stimulus: it evokes picking up the phone only because that behavior has been reinforced after buzzes before, whereas the air puff elicits a reflex regardless of any history. Change the consequences and the same buzz stops working.

It is 6 p.m., you last ate at noon, you walk into the kitchen, and you open the fridge. Which part is the discriminative stimulus and which is the motivating operation?

The kitchen is the discriminative stimulus, because it signals that food is available; the six hours without food are the motivating operation, because they make food worth more and evoke the behavior that has produced it. Remove the motivating operation with a late lunch and the same kitchen evokes nothing.

A parent's goal for their child is "stop interrupting." Is that a behavior you can reinforce?

No. A dead man can manage not interrupting, so it fails the dead-man test: it names the absence of behavior, and an absence cannot be reinforced. Rewrite it as something you want to see, such as "raises a hand and waits," and reinforce that.

Explain it to a friend. Explain why the antecedent does not cause the behavior, using something you did today as the example and without using the words reinforcement or punishment.