L5 - Problems with the S-R theory, then associative learning Flashcards
Critisism of Hull-Spence S-R theory
1: Tolman - Learning without reinforcement (latent learning)
b: Brogden - Sensory preconditioning
2: Harlow: Reinforcement without drive reduction
- monkey solving problem with no reward
3: Extinction, punishment and avoidance learning
Tolman experiment
Rat in maze find their way to the end under different reward conditions.
Partial reinforcement learnt the most, being reinforced from day 11-17, by 12 they are performing at least as well as the ones that are being rewarded, they have learnt without incentive, once providing incentive they will show they have learnt.
Brogden (1939)
Sensory preconditioning
Stage 1 : light - noise
Stage 2: noise - shock
Stage 3: light - lead to shock without reinforcement
when two stimuli AB are paired without reinforcement, and then A is subsequently rewarded, B elicits expectation of reward, showing that an AB association has been learnt without reinforcement
fear response to light implies animal has learned light noise association without reinforcement
Harlow
Reinforcements without drive reductions
Studies of learning rhesus monkeys, they would learn to solve problems without intrinsic rewards.
Monkey could learn a rule (pick the stimulus that is different) without reinforcement
He then showed that monkeys would learn to pull a lever in order to watch a model train through a window, a stimulus that wouldn’t reduce any of the drives thought to be tied to physiological states.
Why has sensory preconditioning attracted more attention than latent learning?
Hull argued that simply removing the rat from the maze could be rewarding. If very small rewards are allowed, latent learning is difficult to disentangle from instrumental learning. Sensory preconditioning does not have this drawback.
How did Amsel’s frustration theory explain the partial reinforcement extinction effect?
In the course of training, the subject experiences frustration before a reward. Over time it comes to associate the frustration with a future reward. During extinction training, the frustration is ineffective in extinguishing the behaviour, because frustration is now a CS for reward.
What was Skinner’s underlying philosophical approach to science?
You don’t need theories of learning, instread: Radical empiricism.
How was Skinner’s radical behaviourism different from early models?
It focused purely on measurable behaviour, and rejected all concepts such as drives that were not directly measurable.
Mowrer two factor theory of avoidance learning
Avoidance learning, two factor theory - classical conditioning (fear of warning signal paired with stimulus) + instrumental conditioning (escape response from fear).
- avoidance becomes the reward.
Animal placed from one compartment, warning signal, if the animal stayed it would be shocked, if it moved it would be saved.
The reward is the decrease in fear - reinforcer (instrumental conditioning)
How did Noam Chomsky’s critique of Skinner’s Verbal Behaviour affect the field of psychology?
It was part of a broader trend to reject radical behaviourism and include mental phenomena in learning models, called the “cognitive revolution”
Describe the blocking experiment and Why was Kamin’s discovery of blocking an important factor in the cognitive revolution?
Used conditioned suppression to measure fear
Stage1: noise - shock
stage 2: light + noise - shock
stage 3: test light alone does it shock?
Result: no fear of light - still neutral, wasn’t paired
Learning occurs only when something unexpected happens
Learning about the noise blocked learning about the light.
Blocking cannot be explained with simple S-R models. If stimuli have difference SALIENCE to animals, that requires including ATTENTION in learning models.
List six features of associative learning theory that developed post-cognitive revolution.
- distinction between learning and performance
- theory is not monolithic. general associative principles apply, but vary within specific species and paradigms
- initially focused on classical conditioning in animals
- Stimuli compete for associative strength
- Limited interest in applications (contrast with Skinner)
- Rescorla-Wagner and later theorists applied associative learning theory to causal learning in humans
Brewer’s review (1974)
There is no convincing evidence for operant and classical conditioning in human beings - poor evidence when reviewing, humans can’t be conditioned to do things without being aware of contingencies
Konorski (1967)
conditioned inhibition - different to pavlov, preparatory vs consummatory responses. Inspired lots of new research such as Kamin’s