A basic discriminative conditioning protocol, first experimentally studied by Pavlov (1927), involves pairing an initially neutral stimulus (A) with an unconditioned stimulus (US) on some conditioning trials, while on others presenting that same stimulus without the US and in compound with a second neutral stimulus (X). Of interest in studies employing this A-US/X-A, or feature-negative, protocol is the type of responding, if any, elicited by stimulus X. Under some circumstances, feature-negative training results in subjects' responding to X as if it signals the impending occurrence of the US, despite X's never having been paired directly with the US. In this case, secondorder conditioning is said to have occurred. However, with apparently minor procedural variations, the featurenegative paradigm can result in the subject's responding to X as if it expects that the US will not be immediately forthcoming. In this case, conditioned inhibition is said to have occurred. Surprisingly, despite the obvious similarities between the procedures yielding these behaviorally opposite outcomes, there has been little systematic research to determine when second-order conditioning or conditioned inhibition (or both) will result from this protocol.An investigation into the factors responsible for excitation or inhibition with this paradigm should ideally employ three dependent measures. To demonstrate that stimulus X is a second-order excitor requires that its responding be compared with that of another nonreinforced stimulus from a control condition providing little basis for excitation to develop (e.g., unpaired X and A, or unpaired A and the US; Holland & Rescorla, 1975;Rizley & Rescorla, 1972). In addition, most learning investigators accept that two separate tests, summation and retardation of acquisition, are necessary to establish whether X is a conditioned inhibitor (Rescorla, 1969). In the summation test, the conditioned response elicited by the simultaneous presentation of the putative inhibitor (X) and an independently trained excitatory conditioned stimulus (CS) is compared with the response elicited by the presentation of the independently trained excitatory CS alone. Reduced responding to the stimulus compound relative to the excitor alone provides evidence for conditioned inhibition, if care is taken to control for other response-decrementing processes, such as generalization decrement (Papini & Bitterman, 1993). In the retardation procedure, the potentially inhibitory stimulus (X) is reinforced, and excitatory reSupport for this research was provided by NIMH Grant 33881, and support for preparation of the manuscript was provided by NIMH Grant 064420. We thank Jeffrey Amundson, Francisco Arcediano, Daniel Burger, Oskar Pineño, Kouji Urushihara, and Daniel Wheeler for their insightful advice concerning all aspects of this research. Special thanks are due Raymond Chang, Jim Esposito, and Jennifer Kelschenbach for assistance in running the animals in what has become quasi-affectionately known as the "Monster Study." ...