L17 - Contingency Learning and Causal Reasoning Flashcards
British Empiricists assumptions of the world
philosophical assumptions of the clean slate no innate knowledge, we compute causal relationships from the covariation of events via some internal mechanism
Michotte 1946
We have an innate ability to perceive a causal interaction in certain situations. But for most event relationships, causation must be inferred
Contingency Judgement
To what degree does performance of action (A) dictate the occurrence of outcome (O)?
Alloy and Abramson
Judgement of depressed and non depressed individuals
Action - press button
Outcome - green light
Participants judge degree their actions controlled outcome
Depressed Ss’ judgments of contingency were accurate
Nondepressed Ss overestimated the degree of contingency between their responses and outcomes
Illusory causation at low contingencies protects you from the helplessness of the world around you
What happens when Ps judgments approximate contingency function equals 0
Overestimated of causation when P(outcome) is high
Overestimation of causation when P(action) is high
Issue of using stat models of contingency judgements
mainly descriptive of function not mechanism
Associationism
Consequences are brought to mind by association
Dickinson, Shanks and Evenden
Show blocking in terms of Enemy tanks at the screen try to blow them up
P1 - Minefield
P2 - minefield + weapon
- There is a chance a mind field is blowing it up anyway
- Participants asked to rate how effective their shells were
0 contingency! Illusory causation
Explanations for blocking deficit
R/W learning deficit, salient cues
Arise from inferences in the test, ordering?
Shank 1985 - Backward blocking
Shows mines-only phase last yielded same effects
However other tasks such as allergist tend to yield stronger forward than backward blocking
Why talk about backward blocking?
S-R models do not predict learning in BB, but tasks do show some evidence of these effects, although typically weaker
Blocking prediction alternatives to associative models
Causal power - we embody representations and events in our world, hard to gage what it is, not measurable but certain types of events have the ability to cause outcomes and others don’t
An inferential account - Ps use reasoning and heuristics depending on the task: all associative learning is explicitly inferred
Lovibond: outcome magnitude
If A causes the outcome and B causes the outcomes, when you put them together the outcome should be stronger
Can enhance blocking to assist attributing causation
Mitchell et al (2006)
The residual blocking effect is related to a failure to encode associations between cue and outcome
Learned predictiveness in causal learning
Gang area groups
Demonstrates that selective attention and learning have a reciprocal relationship in causal learning
Selective attention biased by learning and biases future cause and effect learning