Chapter 3 Summary

Pavlov’s basic conditioning experiment provides a method for studying how organisms associate events that occur together in time. Today, it is generally believed that subjects learn to associate the CS and the US.
Second-order conditioning, sensory preconditioning, and generalization each provide ways in which stimuli that have never been directly associated with a US can elicit a conditioned response.
Most modern research on classical conditioning uses one of several basic methods. These include eyeblink conditioning in rabbits, fear conditioning in rats, autoshaping in pigeons, appetitive conditioning in rats, and taste aversion learning in rats. The conditioning that is observed in each of these systems has some unique characteristics, but it is interesting that the same laws of learning still generally apply.
The success of conditioning depends on several factors. The timing of the CS and US is important; for best conditioning, the CS should precede the US and should occur relatively close to it in time. Also, conditioning is more successful if conditioning trials are spaced apart rather than massed together. Conditioning is also best when the CS and the US are both novel and relatively intense or salient.
Researchers need to distinguish between responding to a CS that results from true conditioning and responding that results from pseudoconditioning or sensitization. Control groups that receive equivalent exposure to the CS and US are usually used for this purpose.
Conditioned excitors are CSs that predict a US; conditioned inhibitors are CSs that predict a decrease in the probability of a US. Inhibitors have effects on behavior that generally oppose the effects of excitors. To detect inhibition, it is often necessary to run summation and retardation-of-acquisition tests.
Conditioned inhibition results from several different procedures, including differential inhibition, conditioned inhibition, explicit unpairing (or negative correlation), and inhibition of delay. Latent inhibition and extinction do not produce CSs that satisfy the modern definition of inhibition; for example, they fail summation tests.
Conditioning is not an automatic result of pairing a CS and a US. Research on CS-US contingencies, blocking, overshadowing, and relative validity each indicate that CS-US pairings are not sufficient to cause learning. Instead, the CS must provide non-redundant information about the occurrence of the US for learning to occur. This idea has stimulated some important advances in our understanding of how conditioning works, some of which will be discussed in Chapter 4.