Second-order conditioning has been frequently observed with the fear response but not with the eyelid response. The present experiments manipulated the temporal relationship between the second-order and first-order stimulus on second-order conditioning trials. Our results indicated that a trace second-order procedure is not effective with either response system. Second-order fear conditioning was most prominent when the second-order CS terminated at the onset of the first-order CS. This arrangement, however, did not produce second-order eyelid CRs. In eyelid conditioning, the second-order CS appears to inhibit responding to the first-order CS which immediately followsit.Pavlov (1927, p. 33) reported that a response originally conditioned to one stimulus could be conditioned to another when the two were paired in the absence of the US. Investigations employing the conditioned fear response have demonstrated reliable second-order conditioning. McAllister and McAllister (1964) trained rats to hurdle-jump to avoid shock. A group which received forward pairings of the second-order and first-order stimulus (CS z and CS.. respectively) showed consistent secondorder responding, whereas a backward conditioning control group did not. Other control procedures, such as noncontingent presentations of either CSt and the US (Rizley & Rescorla, 1972, Experiment 1) or CS z and CSt (Kamil, 1968), have also been examined. These studies demonstrate that both CSt and the US, as well as CS z and CS.. must be paired in order for responding to CS z to develop. These outcomes clearly establish second-order fear conditioning as a bona fide associative phenomenon.In contrast, investigations of second-order con- second-order fear conditioning with an analogous set of procedures.The difficulty in obtaining second-order .eyelid conditioning may be due to the use of an inappropriate time relationship between CS z and CSt. This hypothesis is based on Pavlov's (1927, p. 33) specification that for second-order conditioning to be effective, a temporal gap must separate CS z and CSt; i.e., "the essential condition is that the new stimulus should be withdrawn some seconds before the primary stimulus is applied." Fear conditioning data have not supported this notion. For example, Maisiak and Frey (1977, Experiment 1) found that second-order fear conditioning was less probable when a gap separated CS z and CSt than when the onset of CSt coincided with the termination of CS z. Similarly, Rescorla (1973) found that a delay secondorder fear conditioning procedure was superior to a trace procedure. The present experiments provide a more detailed analysis of Pavlov's hypothesis by systematically manipulating the temporal relationship of CS z and CSt in second-order conditioning.
EXPERIMENT 1 MethodSUbjects. The subjects were 48 male albino rabbits ranging from IS to 20 weeks in age and weighing approximately 1.8 kg. The animals were housed individually with free access to water and Purina Rabbit Chow. Four animals died of respiratory infections, and the...
