Anim Behav., 1995,50, 104-1045 Honey bees are predisposed to win-shift but can learn to win-stay GREGORY E. DEMAS & MICHAEL F. BROWN Department of Psychology, Villanova University (Received 7 October 1994; initial acceptance 12 December 1994; final acceptance 6 February 1995; MS. number: A7135R) Abstract. The ability of honey bees, Apis mellifra, to avoid returns to locations recently depleted of sugar solution (win-shift) or to return to locations recently depleted of sugar solution (win-stay) was tested. Bees collected sugar solution from a small matrix of six cells. During each of a series of trials, they first visited a randomly ‘determined set of three cells. They were then allowed to freely choose between the six cells, with the contingencies encouraging either win-shift or win-stay behaviour. Previous research indicates that honey bees use spatial workin’g memory to discriminate previously visited cells from unvisited cells in this experimental preparation (Brown & Demas 1994, J. camp. Psychol., 108, 344352). In the present experiment, bees in the win-shift condition tended to choose previously unvisited cells throughout the experiment. Bees in the win-stay condition learned to choose previously visited cells over the course of the experiment. These results indicate that bees choose locations based on previous visits, either being attracted to, or repelled from, locations recently depleted of forage. 0 1995 The Association for the Study of Animal Behaviour Although a food reward typically increases the probability of a behaviour, in the context of spatial food-gathering tasks many animals tend to search elsewhere for food after finding it and gathering it from one location. This avoidance of revisits to locations where food has recently been found (and typically depleted) is referred to as a win-shift tendency. A win-shift tendency has been found in a wide range of vertebrate animals, but has been studied in detail using the radial-arm maze, a laboratory task in which rats gather food from a set of locations (usually eight or 12), arranged in a circle (Olton et al. 1981). Olton & Schlosberg (1978) showed that young laboratory rats (with no natural foraging experience) readily learned to choose locations following a win-shift rule, but slowly or never learned to choose locations fol- lowing a win-stay rule. They concluded that rats have a natural win-shift predisposition, but that choices are also sensitive to the pattern of food distribution in the environment. In the present study we examine the win-shift or win-stay tendencies of honey bees foraging in a Correspondence: M. F. Brown, Department of Psychol- ogy, VillanovaUniversity, Villanova,PA 19085, U.S.A. (email: brownmf@vuvaxcom.bitnet). G. E. Demas is now at the Department of Psychology, 223 AmesHall, JohnHopkins University, Baltimore, MD 21218, U.S.A. semi-naturalistic laboratory situation for sugar solution. In nature, honey bees forage for both nectar and pollen from a wide variety of flowers (e.g. Seeley 1985; Winston 1987), with nectar foraging much more common during the late summer, which is when we conducted the present experiment. Most flower species exploited by bees produce nectar at a sufficiently low rate that the flower is depleted of nectar during a single visit (Pleasants 1981). Thus, the contingencies operat- ing in the natural environment of bees appear to encourage a win-shift tendency. On the other hand, flowers vary widely in nectar production rate, acrossboth flower species and environmental conditions (e.g. soil and rainfall). Thus, it may be advantageous for honey bees to be somewhat flexible in terms of win-stay versus win-shift tendencies. Brown & Demas (1994) recently developed a task for the study of spatial choice in honey bees that is analogous to the radial-arm maze task used with rats and other vertebrates. We showed that bees avoided revisits to recently visited locations in accordance with a win-shift tendency, and did so using memory for the identity of previously visited locations. As indicated above, however, honey bees might also behave according to a win-stay tendency if the contingencies encouraged doing so. The present experiment used the basic 0003-3472/95/101041+05 $12.00/O Q 1995 The Association for the Study of Animal Behaviour lnA1