Lectures 16 - 20 Flashcards
What are some factors affecting the rate of discrimination learning?
- similarity between discriminative stimuli
- salience of consequences
- contingencies that differentiate the stimuli
- prior exposure to the discriminate stimuli
describe the phenomenon of perceptual learning
perceptual learning is learning to distinguish between two stimuli = discrimination learning
How can we reconcile perceptual learning and LI
LI describes a phenomenon where pre-exposure of a cue prevents learning about that cue and a US. In perceptual learning, the animal is exposed to the common elements twice as much as the distinct elements, therefore LI of the common elements occurs such that it increases the relative silence of the distinct elements. Therefore mere exposure of stimuli increases an animal’s ability to distinguish between them
Describe the peak shift effect and give 2 examples in everyday life.
subjects respond most frequently to a stimulus they have not encountered before –> stimuli that are further away from S+ and in the opposite direction to S-.
- caracatures
- zebra finches and females prefer very red beaks
what are the conditions for peak shift
- occurs when S+ and S- are very similar
- requires expectation of reward when S- is presented - errors
what are the phenomena that suggest that animals learn which properties to attend to, and cast doubt on non-attention theories like RW.
- interdimensional vs extradimensional shifts
- easy to hard effect
- overtraining reversal effect
Explain Mackintosh’s theory
informative stimuli gain attention when they correctly predict meaningful outcomes. Conversely redundant cues loose attention. the associability (alpha) of a cue A increased if the prediction error of a < prediction error of another cue e.g. B. this means that A is a good predictor of the outcome, therefore attention increases. and Vice Versa
How does Mackintosh predict LI
when a CS is presented without the US during pre-exposure, the value of alpha to the CS predicted goes down. When the CS is paired with the US in a second phase, the value of alpha starts quite low, making increments of associative strength that happen on that trial quite small.
How does Mackintosh predict Blocking
When cue A-US in trial one, the prediction error of A < prediction error of B, meaning Alpha A increases and alpha B decreases. So A is a good predictor of the outcome, so the animal attends to A, therefore learning about B is attenuated.
Describe Learned Predictiveness
Learned predictiveness is the phenomenon whereby an animal attends to cues that were previously useful. So learning is biased towards cues that were previously useful and learning is impaired towards cues that were previously redundant.
Describe Pearce and Hall’s theory of Negative Transfer
When the consequences of the CS are already well known there is little need to attend to it. Alternatively if the US is surprising, the CS isn’t well understood, and this increases attention on the next trial.
How can Mackintosh and Pearce and Hall’s theory be reconciled.
Attention reduces to all stimuli as the outcome is predicted but, the best predictors still win over other stimuli presented in the trial. Functional distinction - exploitation of information (Makintosh) or exploration (Pearce and Hall)
Can associative models account for backward blocking ?
No, many associative models can’t predict backward blocking because they emphasize order effects. they suggest that nothing is learned from presenting A+ after AB+ because the CS-US association has already been reinforced in phase 1 (attention doesn’t change).
How can within-compound associations account for backward blocking?
AB+ A+
A is becoming associated with B such that when you see A it retrieves a memory of B. So your generating a prediction that B should be there, because AB previously made, but it isn’t, therefore AB association weakens, therefore inhibitory learning such that B inhibits the outcome. This is why the presentation of B decreases CR.
How do the findings of learning without awareness differ for trace and delayed learning?
Trace conditioning acquired by aware participants only, delayed conditioning acquired by aware and unaware.
