Behavioral contingency

Contingency refers to a relation, not a reward chart

In ordinary conversation, “contingency” can mean a backup plan. Behavior-analytic sources also use the word inconsistently: for a general relation, a response-consequence rule, a probability contrast, or temporal pairing. In clinical writing, name the relation instead. For a programmed rule, state when the consequence is delivered and when it is not. For descriptive data, state the events, observation window, and comparison probabilities.

Poling and Li's conceptual critique documents these conflicting uses and argues that the term can obscure the relation being described. That is a critique of terminology, not a BACB rule or evidence that every use of the word is wrong.

The three-term contingency adds context

A common operant shorthand is:

Discriminative stimulus (SD): response → consequence

This notation represents a conditional relation. It is not merely a sequence of three events and does not by itself demonstrate causation. In a reinforcement contingency, an SD is an antecedent stimulus correlated with greater reinforcement availability for a particular response than a relevant comparison condition, due to a history of differential reinforcement. The SD can change the current probability of responding; it does not force a response. For clinical measurement, define the response in observable or otherwise validly measurable terms. Classify a consequence as reinforcing or punishing only from its measured effect on later responding, not from its label or an adult's intention.

Suppose a powered automatic door has a working accessibility button. In that condition, pressing the button opens the door. When the power is off, pressing does not open it. The context, response, and response-produced outcome form a three-term relation. Whether door opening reinforces button pressing requires evidence that pressing is strengthened or maintained in comparable future opportunities.

The Association for Behavior Analysis International basic-principles page lists contingencies, stimulus control, reinforcement, punishment, and motivating operations as higher-education content. The BACB BCBA Test Content Outline, 6th edition covers reinforcement and punishment contingencies, automatic and socially mediated contingencies, stimulus control, motivating operations, measurement, and experimental design. These training and examination documents do not validate one person's clinical plan.

A fourth term must be named

Some clinicians expand the shorthand to include a motivating operation:

Motivating operation → discriminative stimulus → response → consequence

Here, the arrows are shorthand, not a claim that four discrete events always occur in a fixed timeline. The motivating operation may already be in effect. For a reinforcer, it changes the event's current effectiveness and momentarily evokes or abates behavior that has previously produced it. An SD concerns differential availability of reinforcement, rather than its current effectiveness. The functions are distinct even when both operate together. Because authors use the fourth term differently, state the notation and definition whenever “four-term contingency” appears. Setting events, health variables, cultural context, rules, and broader environmental conditions can also matter without being forced into one universal diagram.

An ABC observation is a record format, while a contingency is a proposed or demonstrated relation. Writing that a request occurred, a person communicated “break,” and the task paused describes a sequence. It cannot by itself show which event influenced later responding, whether the pause was response-dependent, or whether another variable changed at the same time.

Contiguity and dependency answer different questions

Contiguity concerns closeness in time or sequence. A programmed response-consequence dependency is a delivery rule, while conditional probabilities from descriptive observation estimate association. McComas and colleagues use dependency for the limiting case in which a consequence follows at least some responses and is never observed without the response. Other sources use the word more broadly, so writers should name their convention. A consequence can follow quickly under a response-independent schedule, while a delayed consequence can still be delivered according to a response-dependent rule.

Descriptive data can identify sequential patterns, but unequal conditional probabilities do not establish why the pattern occurred. Stronger causal inference requires experimental manipulation and a design suited to the question.

For a simple trial-based programmed rule, teams can report:

• programmed consequences delivered during the defined post-response window divided by all response windows opened
• consequences delivered during a matched no-response window divided by all no-response windows opened
• timing from response to consequence, with start and stop events defined
• later response levels under comparable conditions, if claiming reinforcement or punishment

The response and no-response comparisons need the same observation unit, eligibility rules, and exposure-window length. The opportunity categories should be mutually exclusive and exhaustive, with raw counts and relevant base rates reported. Prompts, ordinary supports, exclusions, competing responses, and implementation integrity also remain visible. Delivery percentages describe implementation, not whether the consequence changed behavior or the arrangement was useful and ethical.

A fictional example shows the calculation

Sam is a fictional teenager who chooses a goal of using speech or AAC to request one of two songs during a recreation group. An eligible opportunity begins when both songs are available, AAC is accessible, and the group can start either song. A response window lasts up to ten seconds and ends when Sam requests or ten seconds elapse. A ten-second delivery window begins immediately afterward. Following a request, the partner should start the selected song during that window. Following no request, any song start during the matched window is recorded as a response-independent delivery.

Across ten eligible opportunities, Sam requests in six. The selected song starts during the delivery window in 5 of 6 response opportunities. Sam does not request in four; a song starts during the matched window in 1 of 4 no-response opportunities. The two denominators are mutually exclusive and account for all ten opportunities. The team missed a programmed delivery in one response opportunity, and a song started independently in one no-response opportunity. Those are implementation findings, not mistakes attributed to Sam.

These ten descriptive trials show how closely implementation matched the programmed rule. They do not establish a stable contingency or show that song access reinforced requesting. That claim needs repeated data and a credible comparison showing that later requesting strengthened or maintained in relation to the consequence. The team also asks Sam whether the arrangement is enjoyable and worth continuing, while keeping communication, breaks, social access, water, bathroom use, and other essential support available outside the contingency.

Clinical contingencies need ethical boundaries

Contingencies occur naturally and can also be programmed. A clinical arrangement should serve a client-informed goal, offer an accessible response, fit the setting and culture, and avoid coercion. Do not make food needed for adequate nutrition, hydration, bathroom access, communication, pain care, prescribed treatment, safety, or human dignity conditional on performance.

For BCBA and BCaBA certificants and people who have completed an application for either credential, the current BACB Ethics Code addresses competence, client and stakeholder involvement, informed consent and assent when applicable, assessment, client-informed goals, risk, documentation, and evaluation. A qualified clinician should monitor both the selected outcome and unwanted effects, then modify or stop an arrangement that lacks benefit or creates harm.