Lecture 4 - Learning Flashcards
What is learning for?
Learning allows an organism to exploit and benefit from regularities in the environment.
What does a moon halo mean about upcoming rain according to first nations peoples?
The halo and the number of stars within the halo predict rain and the day on which the rain will come respectively.
Learning is about … and …?
Learning is about REGULARITY and INVARIANCE.
See bee example.
What is the example of the bees waggle dance used in the lecture?
Length of waggle gives an indication of distance, direction of path after the waggle give an indication of where the food is relevant to the sun. So, here only the direction and distance are important for the other bees - i.e. the number or colour of the flowers are irrelevant. This is an example of selective attention in learning in non-human animals.
What is counterfactual reasoning?
In terms of cues and outcomes, what is delta-P and what does it tell us?
Delta-P is the probability of an outcome given a cue.
Or, delta-P is the measure of the ACTUAL strength of the contingency between a cue and an outcome.
As delta-P approaches 1, the strength of the contingency increases, i.e. the presence of the cue is strongly correlated with the outcome.
As delta-P approaches -1 it tells us that the presence of the cue is likely to indicate the absence of the outcome.
Delta-P of zero indicates that there is no relationship between the cue and the outcome.
What experiment did Shanks (1991) do regarding delta-P?
In Shanks (1991) experiment participants were told of symptoms that occur or do not occur with certain made-up diseases.
They were then measured on how strongly they think a given symptom is a predictor of a given disease.
The more they were exposed to symptoms and symptom combinations and the made-up diseases the better they were at stating the actual statistical covariation.
What is the PROBABILISTIC CONTRAST MODEL and what is its relevance to calculating delta-P?
The probabilistic contrast model was proposed by Cheng and Holyoak.
This model says that we should only calculate delta-p when the cue and outcome and not being interfered with by other factors, i.e. when the background is constant.
What are some of the mechanisms involved in learning discussed in lecture?
Co-occurrence, delta rule, learned selective attention, and belief updating.
What is the blocking paradigm?
Learning isn’t just a reflection of co-occurance (see example of mice and red, blue lights and juice).
How does the experiment of the mouse, lights and juice illustrate the Blocking Paradigm?
What is the Delta Rule?
The Delta-Rule suggests that the learning is error-driven updating of associations.
What is the Delta Rule Explanation of Blocking?
What are some of the criticisms of the Delta-Rule?
The Delta-Rule cannot predict the Highlighting Effect.
The Delta-Rule also does not predict how quickly people learn associations.
The Delta-Rule cannot explain the Retrospective Revaluation.
What is the Highlighting Effect?
The Highlighting effect occurs in contingency learning in the following example:
A rat is trained to associate red light and bell with food. Then they are trained that red light and alarm lead to juice.
When they are presented with the bell and the alarm the rats go to the juice, even though they have had twice as much exposure to the bell being associated with food.
Because the red light is already associated with food, during the training phase it is the alarm that is learned to be the predictor of the juice regardless of the other cue and therefore when presented with the bell and alarm, the rat chooses the juice.
