L8 - Contingency and Conditioning Flashcards
Temporal Correlation
Not just contiguity that is important but correlation.
This is shown by the truly random sample
Thus conditioning is about understanding causal relationships among events in external world
True or false causality always means contingency but contingency only implies causality
TRUE
How did Rescorla 1968 show contingency affects conditionings
He trained rats on a partial reinforcement schedule: noise (CS) followed by shock US on 40% of CS presentations
In instances when you didn’t have the CS there were 4 groups
- When the US never occurred
- Occurred 0.1
- 0.2
- 0.4
What gives us the best learning is the largest difference between p(US-CS) - US never occurs without the CS
- As the US is played more without the CS learning decreases incrementally
Inhibitory conditioning
contingency is less than 0, learning can occur with the negative contingency (learning that US occurs without CS)
In this case when pairing CS to US it will then take longer as it start below 0 after phase 1
Contiguity
For learning to occur the response must occur in the presence of or very soon after a stimulus is presented, or an association will not occur (time).
Behaviourist approach
Contingency
(Actual reinforce) refers to the predictive relationship between events
Positive Contingency
Probability of US occurring when CS present is greater than probability of US occurring without CS.
Zero Contingency
Probability of US occurring when CS is present equals probability of US occurring without CS.
Gallistel Conditioning as Information
Learning as a basic computational process in which animal estimates when (or if) important events occur
CS must provide worthwhile info about when the US will occur
The CS that carry the most info we will learn the most about
Shannon’s information theory
The CS must reduce the animals level of uncertainty about when the US will happen
Animal learns to respond to CS if it is a useful predictor of US - if CS carries no useful info about US, not worth responding
Comparative Rate Estimates
Responding determined by the rate of US occurring when CS is present, relative to base-rate of US occurrence
Despite big variations in rate of US occurring in presence of CS, conditioning is equivalent as long as temporal conitguity ratio is the same
e.g. Interval between trials 4 and time between trials 24 or 8 and 48 respectively
24/4 = 6
48 / 8 = 6
It is not overall amount of training it is the specific types in which we discipline
When does conditioning occur in terms of contingencies?
Conditioning occurs when there is a +ve contingency
We care about what happens when the CS is present and when it is absent
What does a suppression ratio of 0 imply?
You’ve learnt a lot
Durlach experiment: Will conditioning occur in this situation?
CS1 - light
CS2 - tone
US - food
Food is always delivered and CS1 and 2 alternate
Even though in theory 0 contingency they learning does occur due to the validity of predictiveness.
US never occurs without the CS so prediction is valid
Contiguity vs contingency
Contiguity: CS co-occurs with the US: they are contiguous, or close together, in space and time.
Contingency: the CS predicts the US: the occurrence of the US is contingent on the prior occurrence of the CS.
