L16 - Learning and attention Flashcards
Lashley ‘jumping stand’
The jumping stand showed simultaneous discrimination; rats are placed on a pedestal and have to choose to jump onto two different platforms depending on colour for a reward.
Attention
Prioritising some stimuli coming into the environment over others
Against behaviourist thinking
Lashley thoughts on selective attention
Extreme
Proposed that when subjected in an experiment, when learning a discrimination they are only attending to one isolable feature at a time
Define intra-dimensional shift, positive or negative shift?
The same reinforcement property, but the way it has been used has changed.
- e.g. colour the tell but shapes have changed.
You get a positive transfer because you are able to learn about this shift a lot quicker.
Define extra-dimensional shift, positive or negative transfer?
Rule has changed but participant is still focused on previous rules, a lot of generalisation to the wrong features so interferes.
ED you get a negative transfer because they have to change what information is relevant.
Define easy-to-hard effects (transfer along a continuum)
Initial training on easy discrimination results in better performance on the hard discrimination, because subjects learn to attend to the important features.
Start simple and build complexity to generalise to hard versions
Explanation of easy-to-hard effect
Reason this is due to attention as providing easy version allows one to learn to attend to certain properties and ignore others like where it occurs (position habit). Giving easy version allows them to attend to correct feature and ignore distraction
Mackintosh and Little experiment:
Easy to hard effect with pigeons and colours
- Hard group finding it difficult
- Easy find it easy
- Easy finds it easier at first in second phase, easy reverse start bad but pick it up above hard
- Evidence that easy works in reverse too
Sutherland and Mackintosh 2 stage model explaining attention
Stage 1 - selective attention
- the organism analyses the stimulus on various features that compete for limited attention
Stage 2 - associative learning
- values of the features are determined byS-R consistent reinforcement when focusing on them individually
Define learned irrelevance
Presenting CS without US before the CS and US conditioning leads to retardation in later conditioning.
Kremer Experiment
3 conditions:
Control, CS only before (LI) or CS not leading to US when presented (association with context)
Results:
- Control learns association fastest, CS only delayed response
3 condition key: learned irrelevance -, no real conditioning after 8 trials, because CS is not informative does not attend to it even when it is relevant, takes longer to condition
Mackintosh (1975) thoughts on selective attention
Informative stimuli gain attention when they correctly predict meaningful outcomes. Conversely, redundant or irrelevant cues lose attention.
Flexible alpha after each trial on whether it was relavent
If alpha is the best predictor you will attend to alpha more in the future
Mackintosh application to blocking
- R/W says it is because you are predicting the outcome
- Mackintosh says B doesn’t predict anything, so will ignore it on the first trial
- So alpha of B decreases, attention is diminishing so learning is diminished
Mackintosh experiment of blocking and attention
- Blocking of unblocking
Phase one:
C: N - shock
E: N - Shock
P2:
E: L + N - shock
P3
C and E: L + N - Bigger shock
Results:
- Better conditioning in control, learning doesn’t occur to light in experimental phase as is seen as irrelevant
R/W model says this wouldn’t happen, issue of R/W model as this says some leaning would occur to light
Learned predictiveness
Cues that predict an outcome in a Phase 1 are learned faster in Phase 2 than cues that were unreliable in Phase 1.
It demonstrates selective attention in learning is based on the past utility of cues.