PSYCHOLOGICAL SCIENCE Research Report TRUE IMITATIVE LEARNING IN PIGEONS Thomas R. Zentall, Jennifer E. Sutton, and Lou M. Sherburne University of Kentucky Abstract —Providing evidence for imitative learning iti animals has been made difficult by the need to control for a ntimber of possible nonimitatiort accounts (e.g., mere presence of another animal, attention drawn to a location, attention drawn lo an object being manipulated) that often have not been recofittized in previous research. In the present experiment, we used a ver- sion of the two-action method in which a treadle could he op- erated by a pigeon in one of two distinctive ways: with its beak by pecking or with its fool by stepping. What is unique in this experiment is not only the distinct response topographies, hut also that both responses have the same effect on the environ- ment (depression of the treadle followed by food reward). When pigeons that had observed one of the two response topogra- phies were given access to the treadle, a significant correspon- dence was found between the topography of the observers' re- sponses and that of their respective demonstrators' responses. Imitative learning in animals has attracted considerable recent research interest (see, e.g., Heyes & Galef, 1996; Zentall & Galef, 1988), in part because of its cognitive implica- tions. For example, it has been suggested that evidence that an animal can acquire a response having observed the response made by a demonstrator may indicate that the observer '"un- derstands" how it looks itself when making the same response (see Zental!, 1996). A ntitnber of examples of imitation by aniinals have been reported in the literature. For example, Gardner and Gardner (1969) reported that Washoe, a chimpanzee, would bathe a doll in much the same way that she had been bathed by humans. Similarly, Breuggeman (1973) reported observing a young female rhesus monkey clasp a piece of coconut shell to her stomach in the same manner as her mother clasped the young monkey's infant brother. Finally, Russon and Galdikas (1993) reported that orangutans living free in a rehabilitatloti center manipulated human artifacts in humanlike ways, such as brush- ing teeth, using a knife, sharpening an axe, and applying insect repellant. To recognize the relation between one's own behavior and that of another appears to require what developmental psychol- ogists refer to as point of view or perspective taking (Piaget & Inhelder, 1948/1967). Although perspective taking does not ap- pear to be within the cognitive capacity of laboratory animals such as rats and pigeons, considerable research on imitation learning has been conducted with these species (see, e.g., Galef, 1988; Zentall, 1996). Before one can claim that rats or pigeons are capable of imitative learning, however, one mtist rule out nonimitative factors (see Galef, 1988; Thorpe, 1963; TomaseUo, 1996; Whiten & Ham, 1992; Zentall, 1988). Address correspondence to Thomas R. Zentall, Psychology Depart- tnent, University of Kentucky, Lexington, KY 40.506-0044. One paradigm that has been used to study socially transmit- ted learning involves the association of a nominally neutral ob- ject with fear. According to Mineka and Cook (1988), labora- tory-reared monkeys that are exposed to a toy rubber snake show little evidence of fear, whereas wild-reared monkeys often show a clear aversion to the toy. If a wild-reared monkey is exposed to the toy snake in the presence of a laboratory-reared monkey, fear of the snake is readily transmitted to the previ- ously unaffected laboratory monkey. However, most research- ers would not classify this example of social transmission as imitation because it could be explained in simpler terms. First, fear can readily be transmitted from one animal to another through a process that Thorpe (1963) called contagion. Second, the association of the putative neutral snake whh fear could occur through the Pavlovian pairing of the two (see Whiten & Ham, 1992). That fear of snakes is more readily transmitted from one animal to another than fear of other stimuli (e.g., flowers; Mineka & Cook, 1988) suggests that animals may be predisposed or prepared to make some associations rather than others (humans readily develop a fear of the dark, whereas they rarely develop a fear of electrical outlets; Seligman, 1972). Thus, most imitation experiments with animals have been con- dticted in appetitive, rather than aversive, contexts. In the sitnplest form of such an experiment, one a.sks, for example, if giving a rat the opportunity to observe another rat bar pressing for food reward results in faster acquisition of bar pressing than if the rat had to acquire the response in isolation (i.e., a trial-and-error control). However, motivational changes produced by the mere presence of another rat (i.e., social fa- cilitation; Zajonc, 196.'i) could account for such facilitated ac- quisition. Futhermore, in exposing an observer to a bar-pressing dem- onstrator, one may merely draw the observer's attention to the demonstrator, and consequently draw the observer's attention to the location of the bar (a phenomenon known as local en- hancement; Thorpe, 1963). Attempts have been made to avoid local enhancement by separating the demonstrator's bar from that of the observer using the duplicate-cage procedure (War- den & Jackson, 1935; Zentall & Levine, 1972), Then, if the observer's attention is drawn to the demonstrator and its bar, the observer's attention should be drawn away from the ob- .server's own bar—an outcome that should lead to retarded ac- quisition. But, typically, the observer's bar is quite similar to that of the demonstrator. Thus, it is possible that the observer's attention will be drawn not only to the demonstrator and its bar, but also to the observer's own bar (a phenomenon that has been called stimulus enhancement; Gaief, 1988; Spence, 1937). Control procedures for stimulus enhancement effects are dif- ficult to design because the appropriate control group is not obvious. One approach, suggested by research by Dawson and Foss (196.')), is to identify a task that can be accomplished in more than one way (e.g., the two-action method; Whiten & VOL, 7, NO. 6, NOVEMBER 1996 Copyright € 1996 American Psychological Society 343