lecture 3 Flashcards
relative stimulus validity Wagner et al., 1968
Two groups receive the same number of exposures to the same things paired 50% with US.
The cues that accompany CS1 are manipulated:
True discrimination (TD) condition (AX+ BX-): CS2 perfectly anticipates when the US will occur
Pseudo-Discrimination (PD) condition (AX+/- BX+/-): CS2 is uncorrelated with when the US will occur
Results show that response to CS1 is greater in TD condition than PD condition which must be due to relative value
implications of realtive value for RW 1972
More valid cues interfere with learning about less valid cues
There is some competition for association
Overexpectation A+ B+ ABC+
decline in responding to a pair of well established conditioned stimuli (CSs) that have been given further reinforced training in compound with each other.
how is learning determined by the predicitve validity of cues
Training a cue as an inhibitor means that the expectation for an outcome is negative
Superconditioning occurs when the expectation difference is very large, for example when a stimulus is paired with something negative.
Inhibitors can be used to protect from extinction
The response is determined by what the animal already knows
what factors of schizophrenia are linked to associative learning
Paranoia, interruption in the perceptual world.
Cognitive differences in learning relationships in their environment
Difficulty attending to stimuli on the basis of the meaning - learn about things that are not relevant
Amphetamine- Dopamine agonist produces symptoms
Dopamine (D2) Receptors common treatment target
They learn as well or worse in many standard situations, but with relative value they show ‘enhanced’ learning of cues that controls learn less about (ie Blocking)
Gray (1998) suggested that a failure to learn selectively may underlie disorders (e.g., psychosis, schizophrenia, anxiety).
serra et al 2001
Diagnosed schizophrenics, genetic relatives or controls exposed to computer based learning tasks.
Given in a blocking group and non-blocking group stimuli
Asked to learn about different aspects of the setup
Predictiveness: regularity in blocking stream over trials (blue always followed by yellow)
When flankers are presented on blue trials they are blocked in normal individuals but not schizophrenics or relatives of schizophrenics
Demonstrating the symptoms of schizophrenia are accompanied by different learning
how can you increase learning in the case of latent inhibition according to RW
1)decrease the ΣV by removing cues that predict the CS or 2) increase the value of λ.
latent inhibition
Once a stimulus has been learnt about as having no consequences, subsequent presentation of the stimulus having consequences results in interferences
Retardation in the learning in phase 2 if you’ve been preexposed with a different contingency
However people with schizophrenia do not show this effect
Dependent on position on schizoptipi
Low latent inhibition individuals may see relationships other people cannot see aiding creativity when it is not extreme
Increasing the strength of the US (λ) Dickinson, Hall & Mackintosh (1976)
Design compares the traditional Kamin’s blocking procedure with an Unblocking procedure
Predicted that a change to the outcome should renew attention.
Whereas Rescorla-Wagner model only predicts unblocking if there is an increase in λ.
Found that blocking can be diminished by changing the salience of the outcome (by doubling or halving the outcome)
A model of selective attention (Mackintosh 1977)
surprising events drive attention to CS
Suggested that alpha (α) increases if a CS predicts an otherwise unexpected event, while alpha should decrease if the CS predicts no change
neural evidence for a reward centre
Milner and Olds 1954
intracranial brain stimulation implicates dopamine
Dopmaine (Shultz 1988)
Shultz (1988) recorded extracellular dopamine levels and found that experience of reward or better than predicted outcome resulted in midbrain dopaminergic neurons exhibiting a phasic burst of firing
Shultz & Romo (1990) suggest two functions of the dopamine system
Phasic transmission of reward information
A tonic enabling of postsynaptic neurons (in striatum, frontal cortex and amygdala)
where are the cell bodies of dopamine neurons located
mostly in the midbrain wiht axonal action in striatum and frontal cortex (Schultz, 1998)
indiscriminative type of appetitive US
no prediction
reward occurs
dependent on predictability: excitation A+
reward predicted
reward occurs
depresssion of activity when no reward occurs: extinction A-
reward predicted
no reward occurs
