## Symbols and Numerosities

Other studies have directly addressed the ability of nonhuman animals to learn the relationship between symbols and numerosities. For example, Matsuzawa (Mat-suzawa, 1985; Matsuzawa et al., 1991) and Boysen (Boysen and Berntson, 1989) each trained a chimpanzee (Ai and Sheba, respectively) to choose the appropriate arabic numeral when presented with visual arrays of one to six objects. By the end of training, Ai was able to label the color, shape, and numerosity of sets of household objects. Sheba could both choose the arabic numeral that corresponded to an array of objects and choose the array of objects that corresponded to the numeral (see Figure 6.3). Pepperberg (1987) similarly trained Alex, an African gray parrot, to verbally label the numerosities 1 to 6 with the appropriate English words. Alex was even able to selectively enumerate one type of object in a visual display and ignore

FIGURE 6.3 A drawing of a chimpanzee named Sheba choosing the array of objects that depicts the numerosity of the numeral displayed on a computer monitor.

others (Pepperberg, 1994). For example, when presented with red and blue corks, Alex can answer, "How many blue cork?"

Although it is extremely impressive that chimpanzees and parrots can learn the relationship between numerosities and symbols, a striking aspect of these data is how very laborious it is to train this behavior. In fact, there is no indication of any positive transfer from one numerosity to the next (Murofushi, 1997). For example, after having been trained with the arabic numerals 1 to 3 and the corresponding numerosities 1 to 3, one might expect that when presented with a novel symbol and 3 + 1 dots that the ape would infer that the new numerosity should be paired with the new symbol. However, this is not the case. The chimpanzees required the same number of additional training trials for each new numerosity-symbol pair added to the mix. In addition, when chimpanzees were required to report the color, shape, and number of a set of objects in any order, they chose to report number last (Matsuzawa, 1985). These findings collectively suggest that the pairing of arbitrary symbols and numerosities heavily taxes the cognitive resources of chimpanzees.

0 0