Classical Conditioning II Flashcards
Rescorla-Wagner Model
Learning will occur only when the subject is surprised
Big surprise= lots of learning
Small surprise= less learning
Blocking Experiment
Developed by Leon Kamin
Blocking- phase 1- Light paired with US phase 2- Light and tone paired with US Test- Tone presented Result- No CR (US in phase 2 is not surprising because US is predicted by light)
Control-
phase 2 (no phase 1)- light and tone pairs with CS
Test- tone
Result- CR
Conditioning is not an automatic result of CS-US pairings
For conditioning to occur, the CS must be informative and the US surprising
* Once you have the perfect predictor you don't need anything else * CS blocks learning to other CS (only care about light, don't care about tone)
Associative Strength (V)
Associative strength between CS and US
Learning strengthens associative strength
Each trial changes associative strength, but amount of change gets smaller and smear because US becomes less and less surprising
Conditioned inhibitors (CS-) have negative associative strength
Quantification of How Surprising US IS
Delta V= V max - V n
V n - strength of association at beginning of trial n
Delta V - change is the strength of association produced by trial n
Learning Curves
Differ in terms of V max, rate of acquisition
V max
Determined by magnitude of the US
Stronger US= bigger V max
Alpha
Salience of CS
Value between 0 and 1
Beta
Salience of US
Value between 0 and 1
Rescorla Wagner Model
Learning will occur only when the subject is surprised
Big surprise= lots of learning
Small surprise= less learning
Delta V n= alpha * beta * (Vmax - - V n)
or
Delta Vn= alpha beta(Vmax- sum of V n)
Extinction of a Conditioned Response
Weakening of conditioned response when CS is presented by itself (no US)
Vmax= 0
Explanation of Blocking
Beginning of trail- when 2 CS are used, associative strength of both stimuli is the sum of the individual stimuli
Blocking group- V max (light) is 1.0 due to extensive conditioning, but the tone is new (introduced in phase 2) so V(tone)=0
V (light+tone)=1.0+0= 1
Delta V(tone)= 0.3 (1.0-1.0)=0
Over Expectation
Extensive conditioning -Pair tone with shock, then pair light with shock, when tone and light presented, expect larger shock because both stimuli are presented
When the same intensity of shock is given, subject becomes less fearful of both stimuli
Rescorla: Over Expectation Experiment
Phase 1:
Experimental Group- Tone + Light -> shock
Control group- nothing
Phase 2:
Both groups-
Tone-> shock
Light -> shock
Results showed that control group (no extensive conditioning) was more fearful
Contextual Stimuli
When US not contingent to CS conditioning will be strong to contextual stimuli but not to CS
CS is not always presented, but context (background) is always there, so a response to CS decreases, and a response to context increases
Problems with Rescorla-Wagner Model
Exclusive focus on “surprisingness of US”
Assumes alpha does not change, but it is not always constant
Conclusion that extinction removes original learning (ignores spontaneous recovery, renewal, reinstatement, rapid reacquisition)
Spontaneous Recovery
Reappearance of a CR to a CS after a period of time following the last extinction trial
Extinction adds new learning, just a different kind (competes with original learning)
Renewal
Reappearance of CR to CS due to return to training environment, instead of environment used during extinction
Reinstatement
Reappearance of CR to CS due to a brief presentation of US
i.e Tone/Shock pairing, extinction, then shocked night before…. next day, CR to CS (tone) appears again
Rapid Reacquisition
Rapid return of a CR to a preciously extinguished CS (shorter amount of time to recondition)
Prediction Error
Delta V= prediction error
Large delta V, US not predicted, very surprising
Small delta V, US predicted, much less surprising
Dopamine
Neuromodulator involved in learning, motivation, and variety of psychobiological functions
Wolfram Schultz
Juice comes out of spout, measures response from monkeys
Presents visual stimuli, monkeys associate image with juice
Discovered that after conditioning, dopamine spike after CS, not after juice
Therefore, dopamine is involved in prediction, not pleasure. Overtime something meaningful that is not predicted happens, dopaminergic neurons respond
Autoshaping/ Sign Tracking
Doesn’t require action from subject to get reward
Pair light with grain, pigeon starts pecking at light (even though it doesn’t need to)
Goal Tracking Behaviour
Tracks goal (food)
Primarily uses US
(tracks food, not light)
Dopamine response to CS and US, response to both stay the same
