Psychology › Section F · Learning › Topic F.3
Operant conditioning
Why do people keep checking their phones, and why does a child who is sometimes given sweets for whining keep whining? Operant conditioning explains how the consequences of a behaviour change how often it happens, and why the timing and pattern of those consequences matter so much.
Thorndike's law of effect
In the 1890s Edward Thorndike shut hungry cats in puzzle boxes that opened when the cat pressed a lever or pulled a loop. At first the cats escaped by chance; over many trials they got faster as useless actions dropped away. Thorndike summed this up as the law of effect: behaviour followed by a satisfying result becomes more likely, and behaviour followed by an unpleasant result becomes less likely. Unlike classical conditioning, which links two stimuli and involves automatic responses, this kind of learning links a voluntary behaviour with what follows it.
Skinner and the operant chamber
B. F. Skinner built on Thorndike's idea from the 1930s onwards and called it operant conditioning, because the animal operates on its environment. He studied rats and pigeons in an operant chamber, often called a Skinner box, containing a lever or key the animal could press or peck to receive food, with a device recording every response over time. Skinner's radical behaviourism explained behaviour by its history of consequences, without appeal to inner mental states, and his precise records let researchers compare patterns of reward.
Reinforcement and punishment
Two pairs of words do the work, and both are used in a technical sense. Reinforcement is any consequence that makes a behaviour more likely; punishment is any consequence that makes it less likely. Positive means something is added; negative means something is taken away. Neither word means good or bad. Combining them gives four cases: positive reinforcement (add something pleasant, such as praise or pay), negative reinforcement (remove something unpleasant, such as nagging or pain), positive punishment (add something unpleasant, such as a fine or a scolding) and negative punishment (remove something pleasant, such as phone time or a privilege).
Worked example: four quadrants
Ask first whether the behaviour became more or less likely, then whether something was added or removed. 1. A waiter is tipped well after recommending a dish and recommends it more often: more likely, added, so positive reinforcement. 2. Leah finishes her essay early so her parents stop reminding her, and starts earlier next time: more likely, unpleasant thing removed, so negative reinforcement. 3. A driver is fined for speeding and slows down: less likely, added, so positive punishment. 4. A teenager loses gaming time for swearing and swears less: less likely, pleasant thing removed, so negative punishment. 5. Check: negative reinforcement still strengthens behaviour, so it is never a kind of punishment.
Primary and secondary reinforcers
A primary reinforcer satisfies a biological need and works without learning: food, water, warmth, sleep and relief from pain. A secondary, or conditioned, reinforcer has no value of its own but gains power through association with primary reinforcers or other rewards: money, grades, praise, stickers and points. A token economy builds a system on secondary reinforcers: people earn tokens for target behaviours and exchange them later for rewards. Token systems have been used with some success in classrooms, residential care and psychiatric wards, though gains often fade when the tokens stop.
The limits of punishment
Punishment can stop a behaviour quickly, especially when immediate and consistent, but it has drawbacks. It shows what not to do without teaching what to do; it can make a child fear the person who punishes; and the behaviour often returns when the punisher is absent. Physical punishment carries particular risks. Elizabeth Gershoff's 2002 review of decades of studies found spanking associated with more aggression and poorer parent-child relationships, though most of that evidence is correlational. Many professional bodies, including the American Academy of Pediatrics, now advise against it and favour reinforcing the behaviour that is wanted.
Shaping
Reinforcement can only strengthen a behaviour that already happens, so how does an animal learn something it would never do by chance? By shaping: reinforcing successive approximations of the target. To teach a dog to fetch a ball into a basket, a trainer might first reward looking at the ball, then picking it up, then carrying it towards the basket, and finally dropping it in, rewarding only the closer version at each stage. Skinner shaped pigeons to turn in circles and even to knock a ball back and forth. Teachers shape skills such as handwriting in the same stepwise way, and applied behaviour analysis, used in some autism therapies, builds everyday skills on these principles.
Schedules of reinforcement
Continuous reinforcement rewards every response; it builds a new behaviour fastest but the behaviour also stops quickly once rewards end. Partial (intermittent) reinforcement rewards only some responses, and behaviour learned this way resists extinction far longer. Partial schedules are either ratio (based on the number of responses) or interval (based on time since the last reward), and either fixed or variable. Fixed-ratio: a reward after a set number of responses, like pay per finished item. Variable-ratio: after an unpredictable number, like a slot machine. Fixed-interval: the first response after a set time, like checking for a weekly pocket-money payment. Variable-interval: the first response after an unpredictable time, like checking for a message. Apps that deliver likes and notifications unpredictably exploit these variable schedules, which helps explain frequent phone checking.
Worked example: naming the schedule
1. A fruit picker earns $1 for every crate filled: a set count of responses, so fixed-ratio, giving fast work with a short pause after each reward. 2. A fisher lands a fish after an unpredictable number of casts: variable-ratio, giving high, steady responding that is hardest to extinguish. 3. A pupil checks for exam results that will appear at some unknown point in the week: variable-interval, giving steady, moderate checking. 4. A student with a quiz every Friday studies little on Monday and hard on Thursday: fixed-interval, the typical "scallop" of a lull after each reward and a burst before the next. 5. Check: ratio means count, interval means time.
Learning without reward: Tolman's rats
Strict behaviourists held that learning needed reinforcement; Edward Tolman disagreed. In a 1930 study with C. H. Honzik, rats that explored a maze for about ten days with no food in it were then given food at the goal box; they quickly ran the maze about as well as rats that had been rewarded all along. Tolman argued that they had built a cognitive map, a mental representation of the layout, and that their learning had been latent: present but not shown until there was a reason. The finding brought expectation and mental representation back into the study of learning.
Common mistakes to avoid
First, treating negative reinforcement as punishment: removing something unpleasant strengthens the behaviour. Second, reading positive and negative as good and bad, rather than added and removed. Third, assuming punishment teaches the desired behaviour: it only suppresses the unwanted one. Fourth, confusing ratio with interval schedules: ratio depends on the number of responses, interval on time. Fifth, expecting continuous reinforcement to produce the most persistent behaviour: partial reinforcement, especially variable-ratio, resists extinction best. Sixth, giving in occasionally to a tantrum, which places it on a variable-ratio schedule.
Key takeaways
Operant conditioning links a voluntary behaviour with its consequences: Thorndike's law of effect, developed by Skinner with the operant chamber. Reinforcement increases behaviour and punishment decreases it; positive adds and negative removes. Primary reinforcers meet biological needs, secondary reinforcers such as money are learned, and token economies use them. Punishment suppresses but does not teach, and physical punishment is linked with harmful outcomes. Shaping rewards successive approximations. Continuous reinforcement teaches fastest; partial schedules (fixed or variable, ratio or interval) last longer, with variable-ratio the most persistent. Tolman's latent learning showed learning can occur without reward.
Flashcards
Flashcard
Satisfying results make behaviour more likely; unpleasant ones make it less likely
Test your understanding
Question 1 of 5
Omar's headache goes away after he takes a painkiller, and he becomes more likely to take one next time. Which process is this?
Question 2 of 5
A teacher takes away five minutes of break time each time a pupil talks over others, and the talking decreases. This is:
Question 3 of 5
Which reinforcement schedule usually produces the behaviour that is hardest to extinguish?
Question 4 of 5
A trainer first rewards a parrot for turning its head towards a bell, then only for moving towards it, and finally only for ringing it. This technique is called:
Question 5 of 5
In Tolman's maze studies, rats that explored without food quickly matched rewarded rats once food was added. What did Tolman conclude?
