## Classical Conditioning

### Generalization and Discrimination

Generalization exists when a response conditioned to a CS occurs also to other stimuli, based on the similarity of those other stimuli to the CS. Usually, however, the response to these other stimuli will not be as strong as the response to the CS. Discrimination exists to the extent that subjects respond differently to these other, similar stimuli than they do to the CS. Thus generalization and discrimination can be viewed as two sides of the same coin, or as two ends of a continuum ranging from complete generalization (responding as strongly to the similar stimuli as to the CS) to complete discrimination (responding only to the CS).

A young child who has been bitten by a particular dog will become classically conditioned to the sight of that dog; that sight will evoke strong fear. If the child now also shows signs of fear to other dogs, merely because of their similarity to the dog that inflicted the bite, these responses would be examples of fear generalizing to the new (dog) stimuli. Later, the child may learn that these other dogs are friendly, so that they no longer evoke fear, although the original dog still does. This difference in response to similar stimuli would be an example of discrimination.

A generalization gradient is a graph showing the strength of the conditioned response as a function of some CS variable. Subjects are first conditioned to respond strongly to the CS (e.g., a 1000 Hz tone) by pairing it with a US. They are then tested for their response to other stimuli identical to the CS except for the values of one variable (e.g., tone frequency). Typically, subjects will respond most strongly to the original CS and less to the other test stimuli as they differ more and more from the CS. The result is a generalization gradient whose peak is located at the value of the original CS, with response strength falling off in either direction (e.g., higher or lower frequency tones).

### Procedures Defined by their Temporal Parameters

Pavlov distinguished several classical conditioning procedures according to differences in the ordering and timing of events:
• Delay Conditioning
• The US follows the CS at some specified delay (the CS-US interval).
• The CS continues to be present during the entire CS-US interval.
• This procedure is the best for classical conditioning when used with an optimal CS-US interval. It works less well if the CS-US interval is either too long or too short.
• What is an optimal CS-US interval depends on the nature of the response. For salivary conditioning the optimal interval is on the order of five to ten seconds. For eye-blink conditioning, in contrast, the optimal interval is on the order of only 0.25 to 0.50 seconds.

• Trace Conditioning
Same as delay conditioning, except that the CS remains present only briefly, for only a portion of the CS-US interval. The time between the end of the CS and the beginning of the US is called the trace interval. With equal CS-US intervals, trace conditioning is somethat less effective than delay conditioning.

Delay and trace conditioning both arrange for the CS to precede the US, so both are forms of a more general category called forward conditioning.

• Simultaneous Conditioning
The CS and US are presented simultaneously. This provides the best possible temporal contiguity and so might be expected to be a highly effective classical conditioning procedure. However, it is almost completely ineffective.

• Backward Conditioning
As you might have guessed from the name, in backward conditioning the usual CS-US ordering is reversed: the US comes first, followed by the CS. Pavlov believed that backward conditioning was ineffective, but later research showed that in fact it is effective, but not in the way you might have expected. The backward-paired CS becomes inhibitory; that is, it gains the ability to inhibit, or suppress, the salivary response. One way to demonstrate this is by using a summation test, presenting an ordinary CS and the backward-paired CS at the same time. The salivation normally evolked by the ordinary (excitatory) CS is reduced or even prevented by the simultaneous presentation of the backward-paired CS. (In this test the excitatory and inhibitory effects summate, or add together, but since the effects are in opposite directions they tend to cancel each other out.)

• Temporal Conditioning
The temporal conditioning procedure differs from the others in that no explicit CS is presented. Instead, the US is presented at regular time intervals (e.g., every 30 seconds). After sufficient experience with this procedure, the dog begins to salivate just prior to US delivery, at the end of each interval. Temporal conditioning involves internal, time-correlated stimuli generated in the dog's brain as a result of the regular interval of US delivery. Those time-based stimuli that occur close to US delivery become classically conditioned CSs that evoke salivation just prior to each US delivery.