Repository 1-1995 Responses to Quantity : Perceptual Versus Cognitive Mechanisms in Chimpanzees ( Pan Troglodytes )

Two chimpanzees were trained to select among 2 different amounts of candy (1-6 items). The task was designed so that selection of either array by the active (selector) chimpanzee resulted in that array being given to the passive (observer) animal, with the remaining (nonselected) array going to the selector. Neither animal was able to select consistently the smaller array, which would reap the larger reward. Rather, both animals preferentially selected the larger array, thereby receiving the smaller number of reinforcers. When Arabic numerals were substituted for the food arrays, however, the selector animal evidenced more optimal performance, immediately selecting the smaller numeral and thus receiving the larger reward. These findings suggest that a basic predisposition to respond to the perceptualmotivational features of incentive stimuli can interfere with task performance and that this interference can be overridden when abstract symbols serve as choice stimuli. Evaluation of numerical capabilities in animals in recent years has expanded the range of skills previously demonstrated in nonhuman species, provoked new questions regarding possible processing mechanisms that might subserve these capacities, and encouraged new directions for research (Boysen & Capaldi, 1993; Thomas & Lorden, 1993). Findings from our laboratory have indicated that the chimpanzee is an apt subject for the study of numerical competence and, with appropriate training, can demonstrate facility with counting, basic summation abilities, symbolically mediated summation, and an understanding of ordinality (Boysen, 1992, 1993; Boysen & Berntson, 1989; Boysen, Berntson, Shreyer, & Quigley, 1993). The establishment of such skills in chimpanzees is significantly labor intensive (Boysen, 1992) and requires years of gradually expanded training. Although this may appear at variance with the seemingly effortless acquisition of basic counting skills and number concepts in human children, an appraisal of the children's counting literature suggests quite a different picture. Counting in children entails a highly complex set of skills acquired gradually over years, facilitated by the emergence of verbal abilities and concomitant cognitive development (Fuson, 1988; Gallistel & Gelman, 1992; Gelman & Gallistel, 1978). Although the range of number skills acquired by chimpanzees may be more limited than possible in children, our animals have nevertheless been successful in each number-related task that has been undertaken, including those using numerical symbols (see Boysen, 1993, for an overview). An obvious question arises as to the nature of the symbolic representation evoked by Arabic numerals in animals with an extensive repertoire of numerical competence. The present study offers additional insights into this issue. Our initial goal was to explore the possible use of deception by chimpanzees who had previous training in counting and number comprehension (Boysen & Berntson, 1989). Preliminary training, however, revealed that the animals were unable to perform optimally in a simple discrimination task that required the selection of the smaller of two arrays of food. This inability was striking in view of the complex numerical tasks that the animals had performed successfully. The current study explored this performance deficit and its alleviation by the use of numerical symbols. Results suggest that symbols may come to represent only selective attributes of their referents and that this selective representation may be fundamental to the ultimate adaptive utility of symbols.