Reinforcement · The hub
Reinforcement
Half of operant conditioning is about making behavior more likely. This page defines reinforcement, sorts its two types and its kinds of reinforcers, and points you to the deep pages on each.
Definition
Reinforcement is the process in which a consequence that follows a behavior makes that behavior more likely in the future. The consequence is called a reinforcer. Whether something is a reinforcer is decided by its effect on behavior, never by how it looks or feels.[1]
A "reward" that does not increase the behavior it follows is not a reinforcer. A scolding that does increase it is.
In brief
- Reinforcement is any consequence that makes a behavior more likely; it is defined by that effect, never by how it looks or feels.
- Positive reinforcement adds a stimulus and negative reinforcement removes one; both increase behavior, and the signs are arithmetic, not judgments.
- Most failures of reinforcement are failures of implementation: late, non-contingent, too small, delivered to a satiated person, or on the wrong schedule.
The two types of reinforcement
Both types make behavior more likely. They differ only in what happens to the stimulus: it is added (positive, +) or removed (negative, −). "Positive" and "negative" are arithmetic signs, not judgments.
Positive reinforcement
A stimulus is added after the behavior and the behavior increases. The dog sits and gets a treat; you finish a task and feel the satisfaction.
Definition, examples, how to use it ›Negative reinforcement
A stimulus is removed after the behavior and the behavior increases. You buckle up and the chime stops; you take an aspirin and the headache goes.
Why it is not punishment ›The two-question test
Ask, in this order: (1) Did the behavior become more or less likely? More likely means reinforcement. (2) Was something added or removed? Added means positive; removed means negative. Try it on any scenario with the quadrant checker ›
Kinds of reinforcers
| Kind | What it is | Examples |
|---|---|---|
| Primary (unconditioned) | Reinforcing without any learning, because of biology | Food, water, warmth, sleep, sex, relief from pain |
| Secondary (conditioned) | Acquires its power by being paired with other reinforcers | Praise, grades, a clicker, a "like," a check mark |
| Generalized | A conditioned reinforcer paired with many others, so it works almost regardless of the person's state | Money, tokens, attention, approval |
| Activity | The opportunity to do something more probable than the target behavior | Play after homework, the walk after the coffee — the Premack principle |
| Social | Delivered by other people | Smiles, thanks, being listened to |
| Automatic | Produced by the behavior itself, with no one delivering it | The feel of scratching an itch, the sound of your own humming |
What makes reinforcement work
- Immediacy. Reinforcers that arrive within seconds teach; delayed ones mostly strengthen whatever happened just before they arrived. Bridge a delay with a conditioned reinforcer (a word, a click, a check-off).[2]
- Contingency. The reinforcer has to depend on the behavior. Reinforcement that arrives anyway teaches nothing — or teaches superstition.[3]
- Magnitude and quality. Bigger and better reinforcers work better, with diminishing returns; a cut in magnitude is felt as a loss.[4]
- Motivating operations. Deprivation makes a reinforcer stronger; satiation makes it weaker. Food does not reinforce a full rat.
- Schedule. Reinforce every occurrence while a behavior is being learned, then thin to an intermittent schedule to make it durable. Schedules of reinforcement ›
- The alternatives. A behavior's strength depends on what everything else pays. Enrich the alternatives and the behavior weakens without any punishment. The matching law ›
Reinforcement is not bribery, and not a reward
A bribe is offered before a behavior to induce it; a reinforcer follows the behavior. A reward is something a person gives because it seems nice; a reinforcer is defined afterward, by the fact that the behavior went up. Most failures of "positive reinforcement" in classrooms and homes are failures of one of the five factors above — the reinforcer was late, non-contingent, too small, delivered to a satiated person, or on the wrong schedule — rather than failures of the principle.
Key takeaways
- A reinforcer is defined afterward, by the fact that the behavior went up. A "reward" that does not increase the behavior is not a reinforcer; a scolding that does increase it is.
- Ask two questions, in order: did the behavior become more or less likely, and was something added or removed? More likely means reinforcement; added means positive and removed means negative.
- Reinforcers come in kinds: primary (biological), conditioned (learned by pairing), generalized (money, tokens, approval), activity (a more probable behavior), social, and automatic (produced by the behavior itself).
- Reinforcement depends on immediacy, contingency, magnitude, motivating operations, and schedule. Reinforce every occurrence while a behavior is being learned, then thin to an intermittent schedule to make it durable.
- A bribe is offered before a behavior; a reinforcer follows it. A behavior's strength also depends on what everything else pays, so enriching the alternatives weakens it without any punishment.
Check yourself
A teacher scolds a student every time he calls out, and calling out becomes more frequent. Was the scolding a punisher?
No. A consequence is classified by its effect on behavior, never by how it looks or feels. The behavior became more likely, so the scolding was a reinforcer: attention added after the behavior, which makes it positive reinforcement.
Food is delivered to a pigeon on a timer, regardless of what the pigeon does. Why might it end up repeating some odd movement?
Because reinforcement that arrives anyway strengthens whatever happened just before it arrived. Without contingency, the reinforcer does not teach the intended behavior; it teaches superstition.
You take an aspirin and your headache fades, and you reach for aspirin sooner next time. Is this reinforcement or punishment, positive or negative?
Reinforcement, because the behavior became more likely; negative, because a stimulus (the headache) was removed. Negative reinforcement is not punishment: the sign only says whether something was added or taken away.
A trainer wants to reinforce a dog's recall, but the treats are in a bag across the yard. What should happen the instant the dog arrives?
A conditioned reinforcer, such as a word or a click, should mark the behavior immediately. Reinforcers that arrive within seconds teach, while delayed ones mostly strengthen whatever happened just before they arrived; the conditioned reinforcer bridges the delay to the treat.
Explain it to a friend. Explain what makes a consequence a reinforcer, using one example involving a person and one involving an animal.
Go deeper
Schedules
Fixed and variable, ratio and interval — with a live simulator.
Read ›Shaping
Building behavior that does not yet exist by reinforcing approximations.
Read ›Premack principle
When a behavior is the reinforcer.
Read ›Avoidance learning
Negative reinforcement's strangest and most persistent form.
Read ›The matching law
How reinforcement divides behavior between options.
Read ›Examples
Fifty-plus scenarios sorted by quadrant.
Read ›Frequently asked questions
What is reinforcement in psychology?
Any consequence that makes the behavior it follows more likely in the future. It is defined by its effect: if the behavior does not increase, no reinforcement occurred, whatever the consequence looked like.
What is the difference between positive and negative reinforcement?
Both increase behavior. Positive reinforcement adds a stimulus (a treat, praise); negative reinforcement removes one (a chime stops, a headache ends). "Negative" does not mean bad, and negative reinforcement is not punishment.
What are the types of reinforcers?
Primary (biological: food, water), secondary or conditioned (learned: praise, money, a clicker), generalized (paired with many reinforcers: money, tokens), activity (a preferred behavior), social, and automatic (produced by the behavior itself).
Why does reinforcement sometimes fail?
Usually because of timing (too late), contingency (delivered regardless of behavior), magnitude (too small), motivating operations (the person is satiated), or schedule (thinned too fast). The principle rarely fails; its implementation often does.
References
- Skinner, B. F. (1953). Science and Human Behavior. Macmillan.
- Cooper, J. O., Heron, T. E., & Heward, W. L. (2020). Applied Behavior Analysis (3rd ed.). Pearson.
- Skinner, B. F. (1948). "Superstition" in the pigeon. Journal of Experimental Psychology, 38(2), 168–172.
- Hutt, P. J. (1954). Rate of bar pressing as a function of quality and quantity of food reward. Journal of Comparative and Physiological Psychology, 47(3), 235–239.