Three rats were trained to lever press on concurrent random interval 2-min random interval 2-min schedules of milk reinforcement. With a 5-sec changeover delay, relative response rate matched the relative reinforcement duration associated with each lever. A stimulus, during which unavoidable shocks occurred at random intervals, was superimposed on this concurrent baseline, and shifts in preference were found. However, data from this procedure were ambiguous, apparently confounded by shock-elicited response bursts. Termination of the shocks during the stimulus resulted in a rapid recovery of matching, which was preceded by a brief facilitation of responding on the less-preferred lever. The procedure was then changed to a conventional conditioned anxiety paradigm with a variable duration pre-shock stimulus. A marked shift in relative response rate towards the preferred lever was found in all three rats; that is, responding on the preferred lever wvas far less suppressed during the pre-shock stimulus than responding on the less-preferred lever.Varying the duration of the grain reinforcer for responses under a single-key variableinterval (VI) schedule, Catania (1963) found that response rates of pigeons failed to change systematically with reinforcer duration. However, when subjects were allowed to choose between two keys associated witlh different durations of the reinforcer on concurrent VI 2-min VI 2-min schedules with a 2-sec changeover delay (COD), relative response rate on each key came to match the relative reinforcer duration associated with that key.These results were confirmed and extended in an experiment by Neuringer (1967), who showed that whereas key preference in concurrent chain schedules varied directly with reinforcer duration, absolute rates of responding on each key in the terminal links were comparatively insensitive to clhanges in reinforcer duration. Neuringer noted that amount and frequency of reinforcement appear to influence the relative strength of a response among a set of response alternatives far more than the absolute strength of a response in isolation.Other eters of punishment have a greater effect on responding in a concurrent than in a singleresponse situation. Azrin and Holz (1966) demonstrated that an intensity of shock that only slightly suppressed responding in a single-key punishment situation, when used as punishment of one response in a concurrent situation, produced complete suppression of that response, accompanied by an increase in the rate of the concurrent response. Similar results have been found using timeout, or a loud noise as the punishing stimulus (Azrin and Holz, 1966). However, when both component responses are punished in concurrent VI VI schedules in which reinforcement rates are asymmetrical, the absolute rate of each response is reduced but their relative rates are not affected. The organism's preference still matches relative reinforcement rate, even when overall response rate is considerably suppressed (Holz, 1968). Catania (1966) emphasized that the is...