Fish were allowed to choose, in a T-maze, between various combinations of goalboxes containing either food or a mirror, or nothing. Only food and a mirror maintained swimming, but percent choice and speed were higher for food than for a mirror. Measures of time motionless in the runway, turning away from the goalbox, and air gulping supported the notion that a stimulus for aggressive display induces avoidance as well as approach tendencies and that the aversive component in the display situation is responsible for poorer performance for a mirror.Male Siamese fighting fish, Betta splendens, perform an elaborate aggressive display that is reliably evoked by the sight of a male Betta, a model of a male Betta, or the fish's mirror image. Simpson (1968) provides a detailed description of the behavior. It is well documented that the opportunity to view display-evoking stimuli is positively reinforcing (see Hogan & Roper, 1978, for a review). Nonetheless, studies comparing aggressive display and food reinforcement in B. splendens have shown differential effects on performance in a number of ways: When reinforced with the opportunity to display, fish show (a) lower asymptotic response levels, (b) more variable performance, (c) more rapid extinction, and (d) a failure to adjust response rates to ratio requirements (Hogan, 1967;Hogan, Kleist, & Hutchings, 1970). There is no clear explanation for these differences.One explanation that could account for some of these differencesis that aggressive display may possess negative as well as positive reinforcing properties, and performance for display would be determined by some combination of these two effects. Aggressive displays, in many species, have been shown to be motivated by both approach tendencies (aggression) and avoidance tendencies (escape or fear) (Hinde, 1970;Morris, 1956;Tinbergen, 1952). In Bettas, there is considerable evidence that a stimulus for aggressive display induces avoidance as well as This study was based on a dissertation submitted in partial fulfillment of the requirements for a PhD degree at the University of Toronto by R. J. Bois. It was supported by Grant A-Ol40 from the National Research Council of Canada. We thank Arturo Bouzas, Karen Hollis, Sara Shettleworth, and Gary Walters for their helpful criticisms of the manuscript. Don MacNaughton assisted with computer analysis. Requests for reprints should be sent to J.A. Hogan, Department of Psychology, University of Toronto, Toronto, Ontario M5S lAI Canada. approach tendencies. For example, experiments by Baenninger (1970) and Rhoad, Kalat, and Klopfer (1975) show that Bettas will work to avoid as well as to approach a mirror, model, or a live conspecific under certain conditions.Further support for the idea that display-evoking stimuli induce avoidance as well as approach comes from experiments by Bols (1977). She gave Bettas a choice between an empty goalbox and a goalbox with a display-evoking stimulus behind a clear partition. Three display-evoking stimuli were tested: a live displaying male B...