Spatial relational learning and foraging in cotton-top tamarins

Abstract

Spatial relationalleaming can be defined as the use of the spatial (geometric) relationship between two or more cues (landmarks) in order to locate additional points in space (O'Keefe and Nadel, 1979). An internal spatial representation enables an animal to compute novel locations and travel routes from familiar landmarks and routes (Dyer, 1993). A spatial representation is an internal construct mediating between perceived stimuli in the environment and the behaviour of the animal (Tolman, 1948). In this type of spatial representation the information encoded must be isomorphic with the physical environment such that the geometric relations of distance, angle and direction are maintained or can be computed from the stored information (Gallistel, 1990). A series of spatial and foraging task experiments were conducted to investigate the utilisation of spatial relational learning as a spatial strategy available to cotton-top tamarins (Sag uinus oedipus oedipus). The apparatus used was an 8x8 matrix of holes set in an upright wooden board to allow for the manipulation of visual cues and hidden food items such that the spatial configuration of cues and food could be transformed (translated or rotated) with respect to the perimeter of the board. The definitive test of spatial relational learning was whether the monkeys relied upon the spatial relationship between the visual cues to locate the position of the hidden food items. In a control experiment testing for differential use of perceptual information the results showed that if given the choice, tamarins relied on visual over olfactory cues in a foraging task. Callitrichids typically depend on olfactory communication in socio-sexual contexts so it was unusual that olfaction did not also play a significant role in foraging. In the first spatial learning experiment, the tamarins were found to rely on the three visually presented cues to locate the eleven hidden food items. However, their performance was not very accurate. In the next experiment the task was simplified so that the types of spatial strategies the monkeys were using to solve the foraging task could be clearly identified. In this experiment, only two visual cues were presented on either end of a line of four hidden food items. Once the monkeys were trained to these cues, the cues and food were translated and/or rotated on the board. Data from the beginning and middle of each testing session were used in the final analysis: in a previous analysis it was found that the monkeys initially searched the baited holes in the beginning of a testing session and thereafter predominantly searched unbaited holes. This suggests that they followed a win-stay/lose-shift foraging strategy, a finding that is supported by other studies of tamarins in captivity (Menzel and Juno, 1982) and the wild (Garber, 1989). The results also showed that the monkeys were searching predominately between the cues and not outside or around of them, indicating that they were locating the hidden food by using the spatial relationship between the visual cues. This provides evidence for the utilisation of spatial relational learning as a foraging strategy by cotton-top tamarins and the existence of complex internal spatial representations. Further studies are suggested to test captive monkeys' spatial relational capabilities and their foraging strategies. In addition, comparative and field studies are outlined that would provide information regarding New World monkeys' spatial learning abilities, neurophysiological organisation and the evolution of complex computational processes.