Learning

This is Unit Five of Psychology, Part Two: consequences, and watching. Part One covered learning by association. This part covers learning from what happens afterwards — and it has the most misused vocabulary in the whole course sitting right at the front of it. Four consequences, from two directions crossed with two operations. Add something good. Remove something bad. Add something bad. Remove something good. The first two both increase behavior — and one of them is named negative, which is why an entire error entry exists to say it is not punishment. Then schedules, which is the part with the most practical use in it. Four ways of spacing out reinforcement, each producing a different pattern of behavior and a different resistance to stopping — and the unit gives you a method for picking one from the outcome you actually want. There is also a trap worth knowing about. When a reinforcer stops, the behavior gets worse before it stops — and that spike is the exact moment people give in. And the part closes on learning by watching, in four steps, with four entries naming what it looks like when each step is the one that failed. By the end you will be able to do five things. One: place any consequence in the right cell of four. Two: say why negative reinforcement is not punishment. Three: choose a schedule from the behavior you want. Four: say what an extinction burst is and why it is dangerous. Five: name the four steps of modeling and locate a failure.

Two entries opening the part. Behavior followed by consequences, and the consequences change how often the behavior happens — and a law stated over a century ago that says it in one sentence.

The mechanism.

Psych 0537 Operant Conditioning

Learning in which the consequences of a voluntary behavior change how often that behavior occurs. Voluntary is what separates it from Part One, where the response was automatic.

The law.

Psych 0539 The Law of Effect

Behaviors followed by satisfying consequences become more likely, and those followed by unpleasant ones less likely. The whole mechanism, before any of the vocabulary was invented.

Seven entries, and this is where the vocabulary problem lives. Positive and negative here mean adding and removing. They do not mean good and bad — and because everybody hears them as good and bad, one of the four cells gets misread constantly.

The first direction.

Psych 0541 Reinforcement

Any consequence that increases the frequency of the behavior it follows. Defined by its effect, which means you find out what it was afterwards.

The second.

Psych 0542 Punishment

Any consequence that decreases the frequency of the behavior it follows. Also defined by effect — something intended as punishment that increases a behavior was not punishment.

The first cell.

Psych 0544 Positive Reinforcement

Adding a desirable consequence after a behavior, increasing that behavior's frequency. The cell everybody understands correctly.

The second.

Psych 0546 Negative Reinforcement

Removing an unpleasant condition after a behavior, increasing that behavior's frequency. Increasing — negative describes the removal, not the direction of the effect.

And the error.

Psych 0547 The Negative Reinforcement Error

Treating negative reinforcement as a form of punishment. It is reinforcement — it increases the behavior — and the word negative refers to something being taken away, which is exactly the wrong thing for the word to suggest.

The third cell.

Psych 0549 Positive Punishment

Adding an unpleasant consequence after a behavior, decreasing its frequency. The one most people mean by the word punishment.

The fourth.

Psych 0551 Negative Punishment

Removing a desirable condition after a behavior, decreasing its frequency. Also punishment, and it works by subtraction rather than addition.

Two entries. Two kinds of reinforcer — one that works because of biology and one that works because of association — and then a procedure for producing a behavior that does not exist yet, so cannot be reinforced when it happens.

Two kinds.

Psych 0553 Primary and Secondary Reinforcers

A primary reinforcer satisfies a biological need directly; a secondary one acquires its power by association with a primary. The second is Part One's mechanism, doing work inside this one.

Watch a behavior get built.

Psych 0555 Shaping

Reinforcing successive approximations of a target behavior until the full behavior occurs. The solution to reinforcing something that has never happened — reinforce what is nearest to it, repeatedly.

Five entries, and the four partial schedules are a two-by-two. Ratio or interval — counting responses or counting time. Fixed or variable — predictable or not. Every schedule is one choice on each dimension.

The baseline.

Psych 0557 Continuous Reinforcement

Delivering a reinforcer after every occurrence of the behavior. The fastest way to establish something, and the least durable once it stops.

Counting, predictably.

Psych 0559 The Fixed-Ratio Schedule

Reinforcing after a set number of responses. The count is known, so the pattern of effort follows the count.

Counting, unpredictably.

Psych 0561 The Variable-Ratio Schedule

Reinforcing after an unpredictable number of responses, averaging some value. Unknown count, steady high responding — and the hardest to extinguish.

Timing, predictably.

Psych 0563 The Fixed-Interval Schedule

Reinforcing the first response after a set amount of time has elapsed. Responding drops after each reinforcement and climbs again before the next.

Timing, unpredictably.

Psych 0565 The Variable-Interval Schedule

Reinforcing the first response after an unpredictable amount of time. Unknown timing, so responding stays steady rather than cycling.

Five entries, and this is where the topic becomes usable. A method for selecting a schedule from what you want, and then four goals worked through — teaching something new, keeping output steady, getting a burst before a deadline, and building a habit that survives a lapse.

The method.

Psych 0567 Choosing a Reinforcement Schedule

Matching the response pattern you want and the extinction resistance you want to the schedule producing both. Two requirements, and they can pull in different directions.

Teaching something new.

Psych 0571 Schedule for Teaching a New Skill

A new skill is taught fastest on continuous reinforcement, with every correct response reinforced. Speed of acquisition is the priority, and durability comes later.

Steady output.

Psych 0573 Schedule for Steady Output

Steady, uninterrupted responding comes from a variable-interval schedule, where reinforcement arrives after unpredictable amounts of time. Nothing to time, so nothing to pause for.

A burst before a deadline.

Psych 0575 Schedule for a Burst Before a Deadline

A burst of effort before a fixed point comes from a fixed-interval schedule. The behavior everybody recognises from deadlines is a schedule effect, not a character flaw.

A habit that lasts.

Psych 0577 Schedule for a Habit That Survives Lapses

A habit surviving interruption is built on a variable-ratio schedule, which produces the greatest resistance to extinction. The same property that makes some habits very hard to break.

Four entries. What happens when the reinforcer stops — and then the specific trap, which is that the behavior gets worse first. The entry after it names the mistake, and it is the most practically useful entry in the unit.

Stopping.

Psych 0569 Extinction in Operant Conditioning

The decline of a behavior when its reinforcer no longer follows it. The operant version of Part One's extinction, on a different mechanism.

The spike.

Psych 0579 The Extinction Burst

A temporary increase in the frequency and intensity of a behavior when its reinforcer first stops. Things get louder before they get quieter.

And the trap.

Psych 0580 The Danger in an Extinction Burst

The burst is when reinforcement is most often resumed by accident — which teaches the intensified version of the behavior. Giving in at the worst moment does not restore the old behavior; it trains the louder one.

A system built from this.

Psych 0582 The Token Economy

Secondary reinforcers earned for target behaviors and later exchanged for primary ones. The whole topic, assembled into a working arrangement.

Two entries that neither mechanism accounts for. Learning happening with no reinforcement at all and staying invisible until there is a reason to show it — and an internal representation that supports routes nobody has ever taken.

Learning, unshown.

Psych 0584 Latent Learning

Learning occurring without reinforcement and not shown in behavior until there is reason to show it. Which separates learning from performance — a distinction neither previous mechanism needed.

The representation.

Psych 0586 The Cognitive Map

An internal representation of spatial layout supporting navigation by routes never taken before. Something was built that no sequence of reinforced responses explains.

Four entries on the third mechanism. Learning by watching somebody else — with four requirements, a canonical study where children reproduced behavior nobody had reinforced them for, and reinforcement received by proxy.

The mechanism.

Psych 0588 Observational Learning

Learning by watching another person's behavior and its consequences. Both halves — the behavior, and what happened to them for it.

Four requirements.

Psych 0590 The Four Steps of Modeling

Attention to the behavior, retention of it, the motor capacity to reproduce it, and motivation to perform it. All four, and failing any one means nothing appears.

The study.

Psych 0592 The Bobo Doll Study

Children who watched an adult act aggressively toward an inflatable doll reproduced that aggression without being reinforced for it. No consequence to them, and the behavior appeared anyway.

Reinforcement by proxy.

Psych 0594 Vicarious Reinforcement

An increase in a behavior after observing someone else be reinforced for it. The consequence landed on another person and changed this one's behavior.

Four entries, one per step, and this is a diagnostic set. Each describes a learner who got further than the last one and stopped somewhere specific — and each failure calls for a different fix.

The first step.

Psych 0596 Modeling Step One: Attention

Failing at attention when the learner watches the behavior without registering its sequence. They were present and they did not take it in.

The second.

Psych 0598 Modeling Step Two: Retention

Failing at retention when the learner attends fully and cannot hold the sequence long enough to use it. Attention succeeded and memory did not.

The third.

Psych 0600 Modeling Step Three: Reproduction

Failing at reproduction when the learner remembers the sequence and lacks the physical capacity to perform it. Nothing about this is a learning problem.

And the fourth.

Psych 0602 Modeling Step Four: Motivation

Failing at motivation when the learner can perform the behavior and has no reason to. Everything was learned, and none of it appears — which is Topic Seven's distinction, arriving here again.

One capstone, and it makes the unit's three mechanisms compete. One target behavior, and a plan for teaching it by association, by consequence, and by watching — with each choice argued rather than assumed.

The capstone. One behavior, three routes.

Psych 0604 Teaching One Behavior Two Ways

Selecting between classical, operant and observational methods for one target behavior, and defending each choice. Both parts of this unit, put in competition on a single problem.

Five things you can now do, and Unit Five is complete. You can place any consequence in one of four cells, remembering that positive and negative mean adding and removing rather than good and bad. You can say why negative reinforcement is not punishment: it increases the behavior, and negative refers only to something being taken away. You can choose a schedule from the behavior you want — continuous to teach, variable-interval for steady output, fixed-interval for a burst before a deadline, variable-ratio for a habit that survives lapses. You can say what an extinction burst is and why it is dangerous: the behavior intensifies first, and giving in then trains the louder version. And you can name the four steps of modeling and locate a failure at the right one — watched without registering, held too briefly, remembered without the capacity, or performable with no reason to.