Positive Reinforcement

How positive reinforcement works

The sequence is always the same: an antecedent sets the occasion, a behavior occurs, and immediately afterward a stimulus appears that was not there before. If the behavior then happens more often under similar conditions, positive reinforcement has taken place.

• A rat presses a lever → a food pellet drops → lever-pressing increases.
• A child says "thank you" → a parent smiles and says "you're welcome" → "thank you" increases.
• You post a photo → likes appear → posting increases.

Notice that in each case the consequence is delivered because of the behavior (contingency) and right after it (contiguity). Remove either and the effect weakens sharply. A bonus paid in December for effort in March reinforces very little of that effort; the pellet that arrives ten seconds after the press teaches the rat almost nothing about pressing.^[2]

Check any scenario in two questions

Not sure whether a consequence is this quadrant or a neighbor? Answer the two questions and the checker names it.

Examples of positive reinforcement

Not every example of "adding something nice" is positive reinforcement. If a parent gives a child candy to stop a tantrum in the store, the candy positively reinforces the tantrum (and the end of the tantrum negatively reinforces the parent's candy-giving). Both people learned something; neither learned what they intended.

Types of positive reinforcers

Primary (unconditioned) reinforcers

Stimuli that reinforce without any learning history because they relate to biological needs: food, water, warmth, sexual contact, relief from pain, and — for social species — physical contact. Their effectiveness depends on deprivation: food is a powerful reinforcer for a hungry rat and a weak one for a full one.

Secondary (conditioned) reinforcers

Neutral stimuli that acquire reinforcing power by being paired with existing reinforcers. Money is the classic example; so are grades, praise, a clicker's sound, and a green checkmark. Conditioned reinforcers are what make delayed real-world consequences workable: the click bridges the gap between the dog's sit and the treat that follows a second later.^[3]

Generalized reinforcers

Conditioned reinforcers paired with many different reinforcers, so they work regardless of the organism's current state. Money, tokens, and social approval are generalized reinforcers; that is why token economies work in classrooms and psychiatric wards.^[4]

Social, tangible, activity, and sensory reinforcers

Practitioners often sort reinforcers by category: social (attention, praise, a smile), tangible (a sticker, a toy, a bonus), activity (getting to play, choosing the music), and sensory (a pleasant sound or texture). The Premack principle is the rule for activity reinforcers: a higher-probability behavior can reinforce a lower-probability one. "You can play video games after you finish your homework" is Premack in action.^[5] The Premack principle in depth ›

What makes positive reinforcement effective

1. Immediacy. Deliver the reinforcer within seconds. If you can't, deliver a conditioned reinforcer (a click, a word, a checkmark) immediately and the real one later.
2. Contingency. The reinforcer follows the behavior and only the behavior. Non-contingent goodies are nice, but they don't teach.
3. Magnitude and quality. Bigger and better reinforcers work better up to a point: rats press faster for larger and for sweeter food rewards, with diminishing returns as amount grows.^[8] But the effect is relative. Crespi found that rats switched from a large reward to a small one ran slower than rats that had only ever had the small one — a negative contrast effect — and rats shifted upward briefly ran faster than rats always given the large amount.^[9] Cutting a reinforcer is felt as a loss, not as a smaller gain. And small, immediate reinforcers usually beat large, delayed ones.
4. Motivating operations. Deprivation makes a reinforcer stronger; satiation makes it weaker. Training a dog after dinner with kibble is a losing game.
5. Schedule. Reinforce every occurrence while the behavior is being learned, then shift to an intermittent schedule so it persists. Schedules of reinforcement ›
6. Individualization. Reinforcers are personal. What works is discovered by preference assessment and by watching the data, not assumed.

Positive reinforcement vs. negative reinforcement vs. bribery

Three things get confused constantly:

The difference between reinforcement and bribery is timing and target. Reinforcement follows the desired behavior. Bribery precedes it and is typically triggered by an undesired behavior, so the undesired behavior is what gets strengthened. More on negative reinforcement ›

Common mistakes

• Reinforcing too late. "Great job on the presentation" three days later is pleasant, not reinforcing.
• Reinforcing the wrong behavior. Attention given during a tantrum reinforces the tantrum. Reinforcement must be contingent on the behavior you want.
• Assuming the reinforcer. Stickers do not reinforce every child; public praise does not reinforce every employee.
• Reinforcing every time forever. Continuous reinforcement builds behavior quickly but leaves it fragile. Thin the schedule.
• Reinforcing outcomes you can't control. "Lose weight" can't be reinforced in the moment; "walked after lunch" can.
• Pairing praise with criticism. "Good, but…" turns the praise into a warning signal.

Does positive reinforcement undermine intrinsic motivation?

This is the most serious research-based objection, and the honest answer is "sometimes, under specific conditions." A meta-analysis of 128 studies found that expected, tangible rewards for doing an already-interesting activity reduced free-choice engagement with it afterward; verbal praise and unexpected rewards did not, and informational feedback generally increased intrinsic motivation.^[6] A competing meta-analysis found the negative effect small and limited to a narrow set of conditions.^[7] The practical guidance that follows from both:

• Use social and informational reinforcers (specific praise, feedback on progress) freely.
• Use tangible rewards for behaviors the person would not otherwise do, and fade them as the behavior contacts natural reinforcers.
• Avoid paying people for things they already love doing, especially with rewards contingent merely on engaging rather than on quality.

How to use positive reinforcement: a six-step protocol

1. Define the behavior so a camera could see it. "Raises her hand before speaking," not "is respectful." If you cannot count it, you cannot reinforce it.
2. Find a reinforcer that works for this individual. Watch what the person chooses when free, ask, or test. A reinforcer is proven by its effect, not by your intentions; a sticker that a teenager finds embarrassing is not one.
3. Deliver it immediately and every time — at first. Within seconds, and after every occurrence, until the behavior is reliable. If the real reinforcer must wait, mark the behavior instantly with a conditioned reinforcer: a word, a click, a check.
4. Make it contingent and only contingent. The reinforcer follows the behavior and nothing else. Free access to the same reinforcer at other times drains its power.
5. Thin the schedule. Once the behavior is steady, reinforce it only sometimes, unpredictably. Intermittent reinforcement is what makes the behavior survive days when no one is watching. How to thin a schedule ›
6. Measure, and fade to natural reinforcers. Count the behavior before and after. Then hand the job to consequences the world already provides — the finished essay, the dog's walk, the satisfaction of the check mark — so the behavior no longer depends on you.

Praise, done properly

Praise is the cheapest positive reinforcer available, and most of it is wasted. Jere Brophy's analysis of teacher praise found that it changed behavior only when it was contingent (delivered for the behavior, not as a reflex), specific (named what was done well), and credible (sincere, and not inflated).^[10] A later review added a fourth condition: praise for effort and strategy supports motivation, while praise for ability or for merely finishing can undermine it.^[11] "You kept the paragraph to one idea — that made it clear" reinforces; "great job" does not. Praise, token economies, and the Good Behavior Game in the classroom ›

A worked example

A second-grade teacher wants a student, Maya, to start her worksheet without a reminder. Behavior: pencil on paper within one minute of the worksheet landing on her desk. Reinforcer: the teacher notices that Maya lights up when asked to hand out materials, so "hand out the next set" becomes the consequence. Delivery: the moment the pencil moves, a quiet "you started on your own — you're handing out the readers at ten." Every time, for a week. Thinning: in week two, the job goes to two starts out of three, then to unpredictable ones. Result: starts rose from one in five worksheets to almost all of them, and by the end of the month the teacher's occasional nod was enough. Every step is on this page; none of it required a sticker chart.

Quick check

Positive reinforcement in practice

Positive reinforcement is the default procedure of applied behavior analysis, the core of reward-based dog training, the basis of classroom systems like PBIS and token economies, the engine of contingency management in addiction treatment, and the mechanism behind most successful habit-formation methods. The through-line: identify the behavior, find a reinforcer that actually works for that individual, deliver it immediately and contingently, and thin the schedule over time.

Key takeaways

• "Positive" means a stimulus is added, not that the outcome is pleasant. A reinforcer is not a reward: it is defined only by its effect on the behavior it follows.
• Reinforcers are personal and are discovered by watching the behavior, not assumed. Praise a teenager finds embarrassing is not a reinforcer; attention during a tantrum, even scolding, often is.
• Contingency and contiguity are both required: the reinforcer arrives because of the behavior and right after it. Bribery is offered before the behavior, usually to stop misbehavior, so the misbehavior is what gets strengthened.
• Reinforce every occurrence while the behavior is being learned, then thin to an intermittent schedule and fade to the natural reinforcers the world already provides.
• Expected, tangible rewards for an activity a person already enjoys can reduce intrinsic motivation. Specific praise, unexpected rewards, and informational feedback do not.

Explain it to a friend. Explain why a reinforcer is not the same thing as a reward, using an example from your own week and without using the word "reward" itself.