B.F. Skinner, The Behavior of Organisms (1938), Science and Human Behavior (1953)

- rejects implicit "S-O-R" psychology of Hull and the classical behaviorists: no appeals to implicit unobservable physiological responses inside the organism, or to underlying neural connections in organism; rejects "S-C-R" psychology of Tolman: no appeals to intervening cognitive phenomena or mental states; accepts only observable "S" and "R" events, moreso than any other behaviorist: "empty organism" view

- view of science: no appeal to any kind of theory proposing an underlying cause of behavior - method is to just catalog and summarize observations about behavior; similar to Hume's view of causation: no knowable "cause-effect" relation, just observation of certain events reliably following others

Skinner vs. Thorndike on operant conditioning: (1) Skinner assumes no neural model or brain states explaining S-R connections; (2) Skinner does not believe reinforcement strengthens an S-R connection - responses are not caused by stimuli, but rather are selected and produced for their reinforcing consequences

Skinner's operant conditioning:

1) goal is perfect prediction / control of behavior; emphasis on technology / engineering of behavior (practical) rather than on science / explanation of behavior (theoretical); lawfulness must be found in the individual, not in groups of subjects; individual differences left in behavior must be due to differences in reinforcement histories

2) cumulative record: a learning curve plotting cumulative number of responses against time (so it can only go up or stay flat) - slope is "response rate", the main Skinnerian dependent variable; emphasizes maintenance of behavior: the end product of learning rather than the actual process of learning

- "Skinner box" captures lots of behavior in little time with little fatigue; response is bar press for rats, key peck for pigeons

3) reinforcement increases rate of responding; positive = delivering a stimulus the animal "wants" (e.g., food), negative = taking away a stimulus the animal "doesn't want" (e.g., shock)

- punishment decreases rate of responding; positive = delivering a stimulus the animal "doesn't want" (e.g., shock), negative = taking away a stimulus the animal "wants" (e.g., parental attention in "time-out" procedure)

- according to Skinner, punishment causes at best a temporary suppression of responding

- note: reinforcement and punishment are both defined solely in terms of their effect on behavior (and not in terms of "drive reduction", "goals", etc.): anything that increases the rate of responding is considered a reinforcer, anything that decreases the rate of responding is considered a punishment

4) response - molar: an "operant" is a class of behaviors which includes any response that is controlled by the reinforcement, i.e., any response that brings about a given consequence; same behavior may be instance of different operants in different contexts

- "superstition" in pigeons develops when some behavior is accidentally reinforced and then controlled by its apparent consequences

5) stimulus - event correlated with the production of a response; stimulus is occasion for, not cause of, response

- "stimulus control": discriminative stimulus SD indicates response will be reinforced; S∆ is the stimulus indicating the response will not be reinforced; example: light turned on or off in Skinner box - bar press only reinforced when light is on

6) conditioned reinforcement ("secondary reinforcement" for Hull) - a stimulus associated with reinforcement eventually becomes reinforcing itself; works like higher order conditioning: must be backed up with primary reinforcement or extinction will result

- in a Skinner box, "magazine training" is the first step - click of "magazine" (food delivery mechanism) when food is delivered acts as SD for response of approaching the food tray; because the click always accompanies food delivery, it becomes a conditioned reinforcer; food may not be consumed immediately, but click does follow response immediately (improves reinforcement timing)

- generalized reinforcer - a stimulus associated with many primary reinforcers, not tied to any particular motivational state - for example, money, social approval, etc.

- chaining: note that all SD are conditioned reinforcers because responding in their presence always leads to reinforcement!; thus a complex behavior pattern can be conceived as a chain of simple responses: each response "link" reveals a new SD which indicates that the next response will be reinforced, while at the same time acting as an SR (reinforcement) for the previous response

7) shaping - method for producing new responses in an animal, consisting of differential reinforcement of successive approximations to a desired response; using shaping techniques, pigeons have been taught to play ping-pong, a response that is obviously not in their inital operant repertoire; whereas stimulus control is based on discrimination of stimuli, shaping is based on discrimination of responses

8) schedules of reinforcement - partial reinforcement effect says that response is stronger when animal is not reinforced on every trial; measuring strength of the response by its resistance to extinction, the basic schedules in order of increasing effectiveness are: CR (ex.: soda machine); FI (ex.: cramming for quizzes; "scallop" occurs because time is an SD); VI (ex.: checking e-mail; no scallop); FR (ex.: piecework); VR (ex.: slot machine)

9) other learning phenomena as treated by Skinner:

- motivation: no "drive-reduction" or other theoretical entity is hypothesized - there is just an empirical observation that food-deprived rats respond at a higher rate for food reinforcement

- extinction: not necessarily the disappearance of a response, but rather a return to the response's "operant level" (the rate at which the response appears without any reinforcement)

- spontaneous recovery: recognized as an empirical phenomenon without much explanation of its mechanism

- generalization: when a stimulus complex sets the occasion for a response, the response also occurs when the animal encounters stimulus complexes which share elements with the original stimulus or are otherwise similar

- inhibition: the unobserved theoretical entity / intervening variable employed by Pavlov and Hull plays no role for Skinner - there is only a tendency to produce one response or another, depending on reinforcement history