Three experiments delivered food at fixed or random intervals independently of the rat's behavior, always less than the amount eaten with food freely available. The results revealed a polydipsicresponse to this experimental suppression of eating, and total drinking decreased as total eating increased. When we added a lever that signaled each food delivery, leverpressing and drinking rose far above their baseline levels; both responses decreased as total eating increased. When a similar schedule presented lever and food independently, rats still became polydipsic, but showed no sign of autoshaped leverpressing. A fourth experiment revealed a hypophagic response to schedules that suppressed drinking; total eating increased with total drinking. As mutual substitutes in the economic sense, one behavior falls as the other rises; as mutual complements in the economic sense, the two behaviors rise or fall together. We discuss polydipsia and autoshaping in terms of drinking as an intrinsic substitute for eating, and leverpressing as a learned substitute for eating. The results suggest a revision of conservation theory, whichviewsdrinking and eating as substitutes when the schedule suppresses eating but as complementswhen the schedule suppresses drinking.Given small portions of dry food spaced several seconds apart, the rat may drink from a nearby water tube two or three times as much as it does with food available continuously (Falk, 1961). This excessive drinking, called "schedule-induced polydipsia," has inspired several theoretical accounts, none of which has yet gained widespread acceptance. As a wellspring of theoretical speculation, polydipsia rivals another example of excessive responding, called "autoshaping" (Brown & Jenkins, 1968; Schwartz & Gamzu, 1977)or "sign tracking" (Hearst & Jenkins, 1974), studied mainly in pigeons, but also seen in rats. If we replace the water tube with a retractable lever and present the lever a few seconds before each delivery of food, the rat may press the lever much more than it would ordinarily, even though the delivery of food does not depend on pressing the lever Peterson, Ackil, Frommer, & Hearst, 1972).Conservation theory has seen some success in accounting for instrumental performance under a wide variety of response-contingent schedules (Allison, 1976(Allison, , 1980(Allison, , 1981bAllison, Miller, & Wozny, 1979;Shapiro & Allison, 1978).This paper shows how conservation theory may also account for two kinds of excessive responding under noncontingent arrangements, schedule-induced polydipsia and autoshaped leverpressing. Experiment 1 tests a conservationThe research was supported by Grant MH34148 from the National Institute of Mentel Health. We thank Gary Lucas for his comments on the manuscript. Requests for reprints should be sent to James Allison, Department of Psychology, Indiana University, Bloomington, Indiana 4740'. model for the polydipsic response to a fixed-tim schedule that delivers free food at flxed periods 0 time. Experiment 2 extends the model to a setti...
