Learning & Behaviour Flashcards
Pavlov
Classical or Pavlovian conditioning, subject has no control over response, eg dog salivation.
Skinner
Operant or instrumental conditioning, subject must consciously perform a task for reward, ie they learn to associate an action with a consequence, eg: rats push a lever for food.
Innate behaviours
Behaviours that are inborn and evolve across generations
Kinesis
Reflexes involving the whole body, eg animals placed in dry conditions will seek moisture, but those placed in a moist environment will stay.
Taxis
Similar to Kinesis except it is directional. Eg, moths attracted to light.
Testing for innate behaviours
Eibesfeldt’s squirrel studies
Deprivation experiments with rats
Yang kuo’s kittens and rats.
Displacement behaviours
Differs across species, inborn and pervasive behaviours triggered by conflict situation yet unrelated to conflict itself. Perhaps to delay the decision to either fight or flight. Eg, rats preening, fowl sham feeding, humans yawn or fidget.
Imprinting
Bond formed between young animals and particular class of objects during critical development period. Eg filial imprinting: response of young land birds to follow any moving object whether mother bird or human.
Harry Harlow
Monkey experiments, 1960. Found young monkeys attached to object resembling mother, monkeys bred without mother were savage, unfeeling and treated heir young badly. Suggests attachment approximate form of imprinting and influences capacity of animal to nurture.
Habituation
Form of simple conditioning, tendency for response to get progressively weaker to repeated stimulus. Eg living near airport, gradually not being disturbed by noise. Vital for freeing brain to process new information. Orientations response- new stimuli attracts your attention. Spencer 1960- habituation more rapid with weak stimulus, habituation will occur faster the second time stimulus is introduced.
Abherrant operant behaviours
Unusual phenomena that cannot be explained by traditional operant principles.
Instinctive drift
Naturally occurring behaviours sometimes intrude into operant training procedures.
Criticisms of behaviourist approach
Timberlake: experiments with captive animals fail to consider the biological significance of behaviours, the context they are produce in, and the form of naturally occurring stimulus to response. Proved natural behaviours could be activated, resulted in more naturalistic feeding routine of zoo animals.
Causal reasoning
Capability of animals to learn the relationship between stimulus and response, ie that there is a link between behaviours and outcomes.
Contingency
Behaviour or response of animal actually influence the rate or amount of reinforcement. Hammond 1980: contingency does clearly influence conditioning ( rats).
Non- contingency
Leads to the response being a redundant predictor of reward. Non response must be made more salient ( add stimulus) to break this effect.
Human contingency learning - David Hume
The cause and effect must be contiguous in space and time, the cause must occur prior to the effect, there must be consistent connection between cause and effect, the same cause must always produce the same effect, when several events cause the effect, it must be because the events share a component.
False contingency
Errors and false perceptions can arise under certain circumstances. Superstitious conditioning: whatever animal was engaged in prior to reinforcement had a greater probability of being reproduced.
Illusion of control
Subjective probability of success is higher than the objective probability brought about by the false perception that ones actions can influence outcomes when this hardly seems possible.
Can occur in both chance and skill based tasks.
We have an innate desire to feel in control and bolster our self esteem.
Our brains are designed to find connections and associations
Langer 1975- people find it hard to differentiate between tasks involving chance and skill ( situational confusion). This is because tasks involving chance often contain elements of skill tasks eg decision making and competition.
Primary control- use of direct skills and strategies to influence environment .
Secondary control- aligning oneself with higher power eg prayer
Interpretive/ predictive control - witchcraft, tarot, gamblers fallacy ( I’m due a win), law of averages ( three loses should follow up a win), availability heuristic.
Illusion of control may be more or less likely due to individual and situational factors:
- confusion of luck and skill
- task framing/ instructional sets ( false contingency occurs more often when people are given naturalistic vs analytic instructions)
- need for outcome: more likely to perceive illusion of control if outcome is highly motivating ( Binet et al)
- more depressed people less likely to perceive control ( alloy & abramson)
- more likely to have illusion if anxious, “ hyper vigilance” ( fredland et al)
Thorndike ( classical theories)
Stimulus response theory
Cats escaping puzzle box via trial and error
“ instrumental learning” consequence binds or connects response to stimuli. Distinct from contingency learning as it occurs is association.
3 elementary behaviour laws
1. Law of effect- contiguity of stimuli and response, successful responses have a greater probability of being repeated
2. Law of readiness ( abandoned)
3. Law of exercise ( abandoned)
-notion of multiple response ( animals who vary their response are more likely to get it correct)
- associative shifting ( animals responses can be gradually shifted by making small changes to the stimulus) ( skinner fading or vanishing)
- response by analogy ( animals use a successful response from one situation to a new situation) ( generalisation)
- response availability ( modify response in reaction to variation in environment) ( discrimination learning)
Thorndike vs Guthrie
Thorndike- gradual trial and error learning vs Guthrie - correct response there all the time, matter of producing it in the right situation
Thorndike- rewards reinforce desired responses from animal vs Guthrie- rewards “ weld” behaviour to stimulus conditions but are not essential.
Tolman’s purposive learning
- animals are purposive, produce behaviours to achieve goals
- cognitive maps to navigate environment
- learning can occur in absence of goal
- place learning rather than response learning
- rats learned place even when released at the other end of the t maze
- vicarious trial and error
Influence of tolman’s work: - contiguity and contingency both important
- influenced development of expectancy value theory
- influenced development of Rotter 1966 locus of control construct
- distinguished learnt from performance
Hull
- not a radical behaviourist
- contingency + contiguity
- separation of learning and performance
- drive reduction
- learning= finding appropriate response to stimulus condition/ situation that reduces drive.
- included intervening variables in his theories
Reaction potential theory
Attempted to develop a unified system of behaviour
B.F. Skinner and operant conditioning
- definitive work: ‘ schedules of reinforcement’ fester and skinner 1957
- learning by consequence rather than association
- operant paradigm: increasing dominance of behaviours that lead to reinforcing consequences
- learning is inferred from performance
- positive reinforcement: presenting reinforcing stimulus after desired behaviour
- negative reinforcement: aversive stimulus removed after desired behaviour performed ( eg bob does the dishes to stop wife nagging)
- positive punishment: presenting negative consequence after undesired behaviour
-negative punishment: desirable stimulus removed after undesirable behaviour exhibited
Skinner’s 3 term contingency - discrimination learning: animals learn to recognise the conditions that signal whether making a response will have any effect eg. In the Skinner box, turning the lights off means pressing. The lever will have no effect.
- generalisation gradient: learning that is generalised to other stimulus, plot of response probability and stimulus similarity follows a normal curve.
- extinction and spontaneous recovery: animals deprived of reinforcement will gradually reduce their rate of responding to zero ( extinction). Responses return after an absence of exposure to extinction conditions ( spontaneous recovery).
Contiguity vs contingency
Contiguity : sequential occurrence or proximity of stimulus and response, causing their association in the mind ( Pavlov)
Contingency: conditioning more complicated than the number of CS-US pairings. If CS is paired with US then US is contingent ( conditional) on the CS. Positive, negative and zero contingency. Positive ( excitatory conditioning), negative ( inhibitory conditioning) ( Rescorla).
Behaviour modification
- shaping and successive approximations: selective reinforcement of responses closer and closer to desired response
- clinical applications teaching children with autism, ID and speech therapy
- response chaining & conditioned reinforcement: completion of one element of behaviour becomes the discriminant stimulus for the next element, eg. Teaching a rat to complete a complex series of behaviours eg barnabus maze.
- conditioned stimulus: one which is not naturally reinforcing but which comes to be a reinforcer because its presence signals the presence of primary reinforcement or that it is about to follow
- forward chaining: initial sequence in the chain of behaviour is trained first followed by all subsequent ones eg teaching skills to people with ID
- fading: behaviour modified by eliciting responses to successive variations of an original stimulus object ( used in the past to change sexuality)
- assisted or forced learning: physical intervention in the subjects attempt to learn new behaviours eg lifting an autistic child’s head for eye contact
- intermittent reinforcement: animal is not rewarded for every response; ratio or interval
- ratio: reward every (x) responses
- interval: reward response after ( x) seconds or minutes
- fixed ratio schedule: animal is only reinforced after x number of responses. Eg: FR5 = reward only on 5th, 10th, 15th etc
- fixed interval schedule: animal must wait a certain length of time for reinforcement. Eg: FI3 = animal reinforced for first response after 3 minutes then after 3 minutes has elapsed again etc
- variable ratio: reinforcement of varying number of responses, produces rapid rate of responding
- variable interval schedules: reinforcing the animal after a variable number of minutes eg VR4 on average after every four minutes.
Rescorla-Wagner model,Hull-Spence model, Mackintosh and Pearce - Hall models
-latent inhibition effect: if a CS ( light) is repeatedly presented without consequence during the exposure phase, then animals subsequently learn to associate it with a US more slowly ( compared to a novel CS (tone)). The light does not become inhibitory during the exposure phase ( better name is CS preexposure effect) Rescorla- Wagner and Hull- Spence models cannot explain this.
Rescorla- Wagner model: model of classical conditioning in which the animal is said to learn from the discrepancy between what is expected to happen and what actually happens. Predicts blocking condition ( In Kamin’s blocking effect the conditioning of an association between two stimuli, a conditioned stimulus (CS) and an unconditioned stimulus (US) is impaired if, during the conditioning process, the CS is presented together with a second CS that has already been associated with the unconditioned stimulus.)
-Hull-Spence model, same as above but cannot predict blocking condition.
-Pearce-Hall model - makes the following assumptions: learning is proportional to attention, attention is proportional to the amount of surprise, unpredictable events enhance attention( correct, proved by Kay & Pearce) once a US is fully predicted by a CS, attention to the CS diminishes. Predicts blocking and latent inhibition effect ( but NOT partial reinforcement extinction effect:The partial reinforcement effect (PRE) is the empirical finding that resistance to extinction is greater following acquisition where some, but not all, responses are reinforced (PRF); compared to acquisition all responses are reinforced (CRF).)
-mackintosh: selective attention theory, different explanation for blocking effect, explains latent inhibition, does not always make correct predictions, does not assume that stimuli are automatically attended ( unlike RW and HS) does not explain little to large US switch)
Prediction error
Actual outcome- expected outcome
Assumed to drive new learning
Dopamine neurons
- involved in learning, evidence that brains compute prediction error
- activity in dopamine neurons matches associative strength of CS
- show “ blocking effect” that mirrors behavioural effect
- activity is consistent with Rescorla-Wagner 1972 model
Catastrophic interference
- new learning erases old learning ( most models of learning suffer from this including RW)
- the brain does not suffer catastrophic interference( interference does occur but it isn’t catastrophic)
Memory
- memories are stored in more than one structure, this might explain why we don’t suffer from catastrophic interference.
- brain might store similar versions of the same memory in the hippocampus and in the neocortex, but the rate that the memories are strengthened is different.
- memories are stored in the hippocampus, which learns new associations fast
- at the same time, memories are gradually stored in the Neo-cortex, which learns slowly.
- implies that recent memories rely on the hippocampus and old memories rely on the Neo cortex.
Damaged hippocampus
- unable to form new episodic ( what, when, where) memories but can form procedural memories ( they require slow changes in the neocortex)
- retrograde amnesia ( forget memories that happened before damage)
- amnesia is temporally graded, recent memories are far more likely to be lost than old ones.
- old memories stored in neocortex independent of hippocampus.
