Classical Conditioning Demonstration of the
Rescorla Wagner Model, Learning 301
Introduction:
This cute little applet demonstrates the Rescorla-Wagner mathematical moddel of classical conditioning introduced in "Learning and Memory" by Schwartz and Reisberg (page 101) and "Learning and Behavior" by Schwartz and Robbins (page 114), both published by Norton. The model is particularly powerful in predicting blocking, conditioned inhibition, and context effects (with compound conditioned stimuli). It is wise to read these or other appropriate texts to fully understand the predictions of this model. However, some insight can be gained by running the applet according to the instructions below.
The Rescorla-Wagner model:
The model is expressed by the formula: DVn = K (l - Vn-1), where DVn = change on trial n of the associative strength between a conditioned stimulus (CS) and an unconditioned stimulus (US), Vn-1 = total associative strength between all CS's and the US after trial n-1, K = the effectiveness or salience of the CS, and l = the maximum strength of the US, hence the maximum value of the CR . The formula, Vn = Vn,tone + Vn,light + Vn,context + ........ is used to reflect the total associative strength between two or more CS's and a US when the CS's are presented together as a compound in a tria l.In the demonstration you can administer tone and light CS's, followed by shock and see changes in associative strengths between the tone, the light, and the context (which is always present). It is recommended that you compute the values yourself for each trial.
Running the applet:
1. Set the effectiveness (salience) of each conditioned stimulus; The parameter K represents the salience of a conditioned stimulus. It's value reflects attention given to a conditioned stimulus or how effectively the stimlus is precieved. It is affected by the strength, modality (e.g. vision), and past experience. In this demonstration, the salience of the light and tone should be set much higher that that of the context, to which little attention is paid. I suggest values of .30, .30, and .05 (much higher than that of a "real" conditioning situation, but good for this demonstration).
2. Set the stength of the unconditioned stimulus. l represents the strength of the US and determines the maximum psssible conditioned response (CR). A setting of 150 - 200 will knock your socks off, and a value of 0 - 50 is weak.
3. Determine what conditioning effect you wish to simulate:
Simple Conditioning: Trials consists of a tone (of light) followed by a shock. Trials continue until a substantial CR is produced to the tone.
Blocking: Consists of two conditioning phases followed by a test phase.
1: Trials consist of tone followed by shock until a substantial CR is produced.
2: Trials consist of a tone and light, then shock.
3: Run a trial with the tone only, and another with light only. Note the difference in CRs to the tone and light.
Conditioned Inhibition: Same as blocking, except that in Phase 2, the shock is not administered at all. Again, note the difference in CR's to the tone and light.
Context Effects: Run a number of shock trials (10-20) without any CR, then introduce the tone into the mix (tone -> shock trials). Notice how slow the tone acquires a CR compared to a situation in which the shock was never presented alone.
4. Running a trial: Press the "start" bar to start a trial. You have five seconds to administer a combination of tone, light, and shock. Then the a new trial will begin. You don't need to click the context. it's always there. Use the reset bar to set all values back to their initial defaults.
Created by Donald Walter, Ph.D., University of Wisconsin-Parkside© 1999, University of Wisconsin-Parkside.