Learning & Memory
Chapter 3
Theoretical Issues in Classical
Conditioning
I.
Two major questions:
(1) What are the necessary and sufficient conditions for classical conditioning
Necessary conditions--minimum required for an event to occur. If the necessary conditions are absent, the event __________________.
Sufficient conditions--those conditions that should always produce a particular outcome.
(2) What is learned during classical conditioning?
a relation between the CS and the US (panel b)? OR
a relation between the CS and the UR/CR (panel a)?
II. Answering Question 1
Explaining
necessary & sufficient conditions
A. Contiguity Theory
Temporal arrangements of a CS and US constitute the necessary and sufficient conditions for CR development.
Assumptions:
(1) Excitatory conditioning depends on the occurrence of an effective CS and US
(2) Neural activity of CS must overlap with neural activity of US.
For the neural activity to overlap, CS and US must occur close together in time with the CS preceding the US
Basic findings are consistent with the contiguity theory
(1) Does excitatory conditioning ever fail to occur when these assumptions are met? YES
Several studies have shown that classical conditioning may not occur, even when "effective" stimuli are used.
E.g., Colavita’s (1965) study of salivary conditioning in dogs
Amount of conditioning cannot be determined solely on the basis of CS-US arrangement and temporal contiguity.
E.g., Rescorla (1968)-- Two groups receive same number of CS-US pairings, but one also receives several US-alone presentations. Conditioning differs b/tw the two groups.
Blocking (Kamin, 1968, 1969)--the amount of conditioning to a given CS cannot be predicted solely by looking at the number of times the CS has occurred in a temporally contiguous manner with the US
(2) Does conditioning ever occur when the conditions for classical conditioning are NOT met? YES
Taste aversion conditioning
E.g., Garcia & Koelling (1966)--Assumptions of contiguity did not seem to be met but taste aversion learning occurred.
However it has been shown that shorter intervals in taste aversion studies generally lead to stronger conditioning than longer intervals.
Still, to obtain conditioning when the CS and US are separated by HOURS called into question the assumptions of contiguity theory.
Backward conditioning-- certain backward-conditioning situations do promote excitatory CRs.
E.g., Mahoney & Ayres (1976)--found evidence of excitatory conditioning when only a few backward trials were presented
B. Contingency theory
Conditioning
depends on how adequately a CS serves as a predictor for US occurrence…time
itself is not important. Overall pattern of events in the environment is
important.
Perhaps the strongest and most direct support for contingency theory--Rescorla (1966, 1968).
However, there are cases in which the predictions of contingency theory were not supported. The following discussion will look at situations in which conditioning fails despite a strong CS-US contingency and examine some findings that indicate that conditioning can occur when there is no CS-US contingency.
(1) Conditioning Deficits Not Predicted by Contingency Theory
In some cases, stimuli that reliably elicit a UR fail to serve as effective USs in a conditioning situation, even when there is a strong CS-US contingency (remember Garcia & Koelling).
The blocking phenomenon (Kamin, 1968, 1969) also poses difficulty for contingency theory.
Kamin showed that if one of the elements of a compound CS has been paired with shock before the compound was established, conditioning to the novel element is blocked, despite the positive contingency between the novel element and the US.
Therefore, the amount of conditioning to a given CS cannot depend only on the contingency between the CS and a US.
(2) Conditioning in the Absence of CS-US Contingency
Tested by Durlach (found evidence for conditioning in an autoshaping procedure even when the CS-US contingency was zero).
C. Rescorla-Wagner theory—a more complex version of contiguity theory.
Conditioning depends on the surprisingness of the US . The model describes this idea conceptually and mathematically.
The model provides a trial-by-trial description of the increments in conditioning that occurs to each CS that is present.
Symbols & definitions:
Can use this formula to specify how much a CS will become conditioned on any given trial, based on how much conditioning has occurred in the past.
The Evidence for Rescorla-Wager Model
Can handle most of the basic conditioning phenomena (e.g., better conditioning with salient CSs and stronger USs; better conditioning to forward arrangements, etc.).
Can also account for phenomena that cannot be explained by contiguity theory or contingency theory, such as:
(1) Effects of Stimuli other than the CS--(e.g., blocking of conditioning to an element in a compound CS)
A novel CS paired with a US will not develop the capacity to produce a CR if, at the time of the pairing, the US is already predicted by other stimuli in the environment.
(2) The Occurrence of Inhibitory Conditioning
Problems with R-W model—Fails to account for certain details of the blocking phenomenon. For this reason, some of the assumptions of the model have been modified in more contemporary versions of the R-W model (e.g., Pearce & Hall, 1980). Also, fails to account for certain anomalies that occur in certain conditioning situations.
R-W model remains a very influential theory of classical conditioning
III. Answering Questions 2
What is Associated in Classical Conditioning?
Two main points of view: (1) Organisms learn a relation between the CS and the UR. (2) Organisms learn a relation between the CS and the US.
Stimulus-Response or Stimulus-Stimulus Learning?
A. Stimulus-Response Learning
Assumptions:
CS comes to elicit the same reflexive response produced the US (e.g., Hull, 1943).
The CS becomes a second stimulus (in addition to the US), which is linked to and is capable of producing the reflexive response (UR).
However, Light & Gantt (1936) prevented dogs from making the UR on training trials...
B. Stimulus-Stimulus Learning
During conditioning, organism learns a relation between the CS and the US.
Once conditioning has occurred, presenting CS stimulates an experience (mental representation) of the US.
This US representation then provokes the appropriate conditioned response.
Instead of the CS coming to produce a response directly, the CS triggers an experience of the US, and the response occurs as a reaction to this US experience.
Evidence—difficult to resolve the question of what is learned…Much of the available evidence does appear to favor the view that conditioned responding depends on stimulus-stimulus learning.