Reinforcement, Punishment, Automatic, and Socially Mediated Contingencies

The Two Functional Definitions

Reinforcement is a contingent consequence relation in which behavior increases in future frequency, duration, magnitude, or another dimension. Punishment is a contingent consequence relation in which behavior decreases in the future. Both are functional definitions — they are confirmed only by an observed change in behavior, never by whether the consequence seems nice or harsh.

Positive means a stimulus is presented or added after the response. Negative means a stimulus is removed, reduced, postponed, or avoided after the response. The words positive and negative describe the direction of the stimulus change, like plus and minus signs in math — they carry no value judgment.

Crossing those two dimensions yields four contingencies. To classify any scenario, answer two questions in order: (1) Did future behavior increase or decrease? That settles reinforcement vs. punishment. (2) Was a stimulus added or removed? That settles positive vs. negative. Never classify from staff intent alone.

The Four-Quadrant Fast Sort

• Positive reinforcement: a child completes work, earns a token, and work completion increases.
• Negative reinforcement: a child screams, the demand is withdrawn, and screaming increases. Two subtypes: escape (terminating an ongoing aversive) and avoidance (preventing an aversive that has not yet started).
• Positive punishment: a child runs in the hall, gets a reprimand, and running decreases.
• Negative punishment: a child swears, loses 5 earned tokens (response cost) or access to a preferred activity (time-out), and swearing decreases.

The deadliest trap on the exam is confusing negative reinforcement with punishment. Negative reinforcement always increases behavior; both kinds of punishment decrease it. If behavior goes up, it cannot be punishment no matter how aversive the situation sounds.

Automatic Versus Socially Mediated

The source of the consequence creates a second cut across all four contingencies.

In an automatic contingency, the response itself directly produces the consequence — no other person is required.

• Automatic positive reinforcement: rocking that produces vestibular stimulation; hand-flapping that produces visual feedback.
• Automatic negative reinforcement: scratching that reduces an itch; covering ears to dampen a loud noise.

In a socially mediated contingency, another person delivers the consequence.

• Socially mediated positive reinforcement: a child signs "more" and a caregiver provides juice (or attention).
• Socially mediated negative reinforcement: a student hits, and the teacher removes the worksheet.

This source dimension is what functional assessment ultimately maps onto: attention and tangible access are socially mediated positive reinforcement, escape is (usually socially mediated) negative reinforcement, and sensory/automatic maintenance is produced by the behavior itself. The exam will hand you a vignette and ask both the four-quadrant label AND whether it is automatic or socially mediated.

Unconditioned, Conditioned, and Generalized

Reinforcers and punishers are also sorted by their learning history.

• Unconditioned (primary) reinforcers/punishers function without any prior pairing because of phylogeny: food, water, warmth, oxygen, and sleep act as unconditioned reinforcers; pain, extreme cold/heat, and loud noise act as unconditioned punishers.
• Conditioned (secondary) reinforcers/punishers acquire their function by being paired with existing reinforcers or punishers. A token, a buzzer, or the word "no" gains function through pairing.
• Generalized conditioned reinforcers are paired with many different backup reinforcers, so they work across motivational states and are relatively independent of any single MO. Money, tokens, and praise are the textbook examples — a token economy works precisely because tokens are exchangeable for varied backups.

Two high-yield cautions: Preference does not equal reinforcement. A preferred item from a preference assessment is only a reinforcer if its contingent delivery actually increases behavior. And a reprimand is not automatically a punisher — for attention-maintained behavior, a reprimand often functions as positive reinforcement and makes the behavior worse. Always confirm function with data, not with how the consequence looks.

Two-Step Decision Procedure with Worked Cases

The single most reliable way to avoid the classic distractors is to run every scenario through the same two questions, in this order.

1. Did future behavior increase or decrease? Increase → reinforcement. Decrease → punishment. (If the stem gives no future-effect data, the contingency cannot be named yet.)
2. Was a stimulus added or removed? Added → positive. Removed/reduced/postponed → negative.

Work these cases:

• Seatbelt buzzer. The buzzer sounds; you buckle; the buzzer stops; you buckle faster on future trips. Behavior increased, a stimulus was removed → negative reinforcement (escape).
• Speeding ticket. You speed; you get a ticket; you speed less afterward. Behavior decreased, a stimulus was added → positive punishment.
• Losing recess. A student talks out; the teacher removes 5 minutes of recess; talking-out decreases. Behavior decreased, a stimulus was removed → negative punishment (response cost).
• Sunscreen/avoidance. You apply sunscreen; you prevent a future sunburn that has not yet started; applying increases. This is avoidance, a subtype of negative reinforcement — the aversive was prevented, not merely terminated.

Notice that 'aversive-sounding' does not equal punishment, and 'pleasant-sounding' does not equal reinforcement. The hallway, the buzzer, and a reprimand can all increase behavior depending on function.

Finally, layer the source question on top: was the consequence delivered by another person (socially mediated) or produced directly by the response (automatic)? A complete classification names the quadrant AND the source — for example, 'socially mediated negative reinforcement' for escape that a teacher grants, versus 'automatic negative reinforcement' for scratching that relieves an itch with no one else involved.