180 likes | 473 Views
RESCORLA-WAGNER MODEL. What are some characteristics of a good model? Variables well-described and manipulatable. Accounts for known results and able to predict non-trivial results of new experiments. Dependent variable(s) predicted in at least relative magnitude and direction.
E N D
What are some characteristics of a good model? Variables well-described and manipulatable. Accounts for known results and able to predict non-trivial results of new experiments. Dependent variable(s) predicted in at least relative magnitude and direction. Parsimonious (i.e., minimum assumptions for maximum effectiveness).
STEPS IN MODEL BUILDING • IDENTIFICATION: WHAT’S THE QUESTION? • ASSUMPTIONS: WHAT’S IMPORTANT; WHAT’S NOT? • CONSTRUCTION: MATHEMATICAL FORMULATION • ANALYSIS: SOLUTIONS • INTERPRETATION: WHAT DOES IT MEAN? • VALIDATION: DOES IT ACCORD WITH KNOWN DATA? • IMPLEMENTATION: CAN IT PREDICT NEW DATA?
PRINCIPAL THEORETICAL VARIABLE: ASSOCIATIVE STRENGTH, V
ASSUMPTIONS 1. When a CS is presented its associative strength, Vcs, may increase (CS+), decrease (CS-), or remain unchanged. 2. The asymptotic strength () of association depends on the magnitude (I) of the UCS: = f (UCSI). 3. A given UCS can support only a certain level of associative strength, . 4. In a stimulus compound, the total associative strength is the algebraic sum of the associative strength of the components. [ex. T: tone, L: light. VT+L =VT + VL] 5. The change in associative strength, V, on any trial is proportional to the difference between the present associative strength, Vcs, and the asymptotic associative strength, .
Overshadowing k= .1, .6
Accounting for Known Data • Un-Blocking • What value can we change to allow associative strength to accrue to CS2?
Accounting for Known Data V1= 90 V2= 0 k= .2
Novel Predictions • Overexpectation effect • CS1+UCS… • CS2+UCS… • What happens when you present CS1+CS2?
Weaknesses of Models • Errors of commission • Errors of omission
Weaknesses of the R-W Model • Extinction curve not mirror image of Acquisition • Facilitated re-acquistion • What about time? • Spontaneous recovery • CS pre-exposure effect (latent inhibition) (see review article Miller et al. 1995)
Importance of Context • UCS pre-exposure effect • Present UCS by itself • Then pair UCS with CS • Explained by R-W, how?
Probability Space Random Control: 0 < P(UCS|CS) = P(UCS|~CS) Inhibitory CS: 0 < P(UCS|CS) < P(UCS|~CS) Conditioning: 0 < P(UCS|~CS) < P(UCS|CS)
Combining Conditional Probability and Rescorla-Wagner • How can R-W account for no conditioning to the CS in the following: • 0 < P(UCS|CS) = P(UCS|~CS) • Hint: CS is a compound (CS + CTX) • ΔVCS= ΔVCS +ΔVCTX
Contiguity in R-W Model If P(UCS|CS) = P(UCS|~CS), we really have CS = CS + CTX and ~CS = CTX. Then: P(UCS|CS + CTX) = P(UCS|CTX) V (CS + CTX) = VCS + VCTX (R-W axiom) V CS = V (CS +CTX) = VCS + VCTX but: V CTX = V ~CS so: V (CS + CTX) = V CS + V~CS = 0 No significant conditioning occurs to the CS Pavlovian Conditioning