PSYCHOLOGY 435
Spring, 1999: Test #1

Part I. Please answer all three questions in this section. Questions 1 and 2 are worth 30 points each, and Question 3 is worth 20 points. (Distribute your time accordingly!) As always, you should CITE EXPERIMENTAL EVIDENCE (including names!), where possible, to support your answers.

1. In this experiment, you will be conditioning two different groups using a shock with a maximum association value of 200, and a tone CS and a light CS. Each CS has a salience of .2. Here is the history of which CS is paired with the shock for each group on each trial:

                    Group 1                                    Group 2
        Trial 1     light                                         light & tone
        Trial 2     light                                         light & tone
        Trial 3     light & tone                             light
        Trial 4     light & tone                             light
        Trial 5     light & tone                             light

Given this, and calculating what the Rescorla-Wagner model predicts ought to happen,

(a) fill in the table on the next page (and SHOW YOUR CALCULATIONS on paper, so I know where and how you got your numbers. To keep things simple, round your numbers to the nearest tenth: i.e., use just one decimal place):
 
 
Group 1
VpresentCues 
(fill in if present): 
CSA: L         CSB: T 

  
TOTAL V
CSs 
CSA: 0         CSB: 0 
Trial 1 0 0 40 
40
Trial 2 40 40 32 
72
Trial 3 72                0 72 25.6                 25.6 
97.6                 25.6
Trial 4 97.6           25.6  123.2 15.4                 15.4 
113                  41
Trial 5 113            41 154    9.2                  9.2 
122.2               50.2
 
 
 
Group 2
VpresentCues 
(fill in if present): 
CSA: L            CSB: T 		  

TOTAL V
CSs 
CSA: 0             CSB: 0
Trial 1 0                         0 0 40                      40 
40                      40
Trial 2 40                     40 80 24                      24 
64                      64
Trial 3 64 64 27.2 
91.2
Trial 4 91.2 91.2 21.8 
113
Trial 5 113 113 17.4 
130.4
 
 

(b) Is there any overshadowing or blocking in this experiment, and why? (Justify your answer by talking about numbers!)

Light blocks tone in group 1; tone helps block light in group 2; CR to light (or tone) is less than expected on Trials 1 & 2 due to OS from compounding
(c) (i) By the end of the experiment, which group has the stronger tone CR, and why? (ii) Do the two groups have equal light CRs, since the light was presented on each trial for both groups? If not, which group has the stronger light CR, and why?

        (i) Group 2: 64 vs. 50.2
        (ii) No, because of B, OS: Group 2: 130.4 vs 122.2

(d) On the first trial, which group has the largest change? For which CS? What about Trial 2? Trial 3? Trial 4? Trial 5? (Justify your answer by talking about numbers!)

        T1: equal (40);    T2: G1(L=32);    T3: G2 (L=27.2);    T4: G2 (L=21.8);    T5: G2 (L=17.4)

(e) On each trial (as in the previous question), at the end of that trial, which group in theory has the weakest CR? For which CS? (Justify your answer by talking about numbers!)

        T1&2: G1 (T=0);    T3: G1 (T=25.6);    T4: G1 (T=41);    T5: G1 (T=50.2)

(g) If the associative value of the shock is reduced to 125 on a sixth trial, and both groups on that trial have the shock paired just with the light, will both still have an excitatory response to the light? Why or why not?

        Because Lambda < V[PC] in Group 2, a negative change (but CR will still be +)
 
 2. Four major theories of classical conditioning we've learned about were Pavlov; the Rescorla-Wagner Model; Miller's Comparator Theory; and Wagner's Memory Model.

(i) What would each have to say about inhibition in extinction, and what is learned in extinction?

Pavlov: extinction is the learning to inhibit a CR. Evidence for: spontaneous recovery; relearning; disinhibition

RW: Lambda becomes 0, driving the CR down to 0, so the CR is not inhibited. In support, Rescorla finds the CS doesn't pass the summation test

Miller: has no process or mechanism of inhibition in his model

Wagner: an expected UCS during extinction fails to occur. This is surprising, causing a round of new rehearsals, but of the CS alone.

(ii) Which of the four, if any, best handles the findings in learned taste aversions? Why? (Note that there are several findings here!)
The findings here you needed to worry about included long-delay learning, belongingness, and spontaneous recovery: Kalat & Rozin; Garcia & Koelling; Wilcoxin et al.

The only model we looked at that tries to handle all of these is Wagner's model, and I would have expected you to discuss studies like Krane & Wagner.

(iii) Which, if any, has a problem with the following, and why (you should be able to provide brief answers for this one):

a. Rizley & Rescorla's study of higher-order conditioning

goes against Pavlov who claimed that 1st-order inhibition (extinction) would cause 2nd-order inhibition: R&R found this didn't always happen
b. Potentiation
goes against RW who predict that the exact opposite effect (overshadowing) ought to occur
c. Antagonistic conditioning
goes against Pavlov who claimed that the CS should always resemble the UCS (studies by O'Brist et al., Siegel)
d. McAllister's study
goes against RW who have no parameter in their formula to account for contiguity effects modifying an association/CR
e. Dwyer et al. (Dwyer, Mackintosh, & Boakes)
Relevant to Wagner's SOP (though presents problems for most theorists): an association will form between two representations in the A2 state
3. Two of the very different models of classical conditioning we've briefly looked at were Mackintosh's Attentional Model (discussed mainly in the text) and Hull's model. I'm not asking you to discuss these models in detail here. But do take each of the people below, and tell me which of the two models each person would be more comfortable with, and why! (In other words, compare their philosophies/assumptions against those of the two models.)

                Aristotle                 Watson                 DesCartes                 Locke

The relevant contrast I expected you to focus on here was between contiguity approaches (behaviorist theories) and contingency approaches (cognitive theories), which would have allowed you to discuss each approach's roots in the Empiricist & Rationalist philosophies
Part II. Briefly identify five of the following (each is worth 4 points):

1. General process theory

            part of behaviorism: all species learn in the same way (via associations due to temporal contiguity)

2. Latent inhibition

            pre-exposing the CS results in slower CS-UCS association

3. Pre-exposure overshadowing effect

            Blaisdell et al.: LI & OS cancel out, as predicted by the comparator model

4. Krane & Wagner

            belongingness depends on interval between CS & UCS

5. SET

type of comparator model: CR to CS occurs if CS predicts UCS sooner than context by a certain time proportion: C/T > 2 (Gibbon & Balsam)
6. Positivism

            Comte: allow only reference to physically observable things