Search is a serial exploration of alternatives. Efficient search involves the ability to minimize costs (i.e., time/energy) and to keep track of alternatives already explored. The search abilities of 4 capuchin monkeys (Cebus apella) were evaluated by means of an apparatus featuring a set of suspended baited containers. The experiment featured conditions with different spatial configurations of the search space. Results show that the monkeys were able to search exhaustively 9 containers spatially distributed either as a 3 × 3 matrix or as 3 "patches" of 3 containers each. Search efficiency was higher in a search space suitable to organization in clusters or spatial chunks. In this condition, evidence for principled organization of search trajectories, as opposed to a random walk through the search space, emerges clearly and parallels search efficiency. This suggests that monkeys impose a structure over the search space and, by doing so, reduce the memory demands of the task.In cognitive science, search is defined as the problem of what to do next in situations that require an exploration of multiple alternatives (Stillings et al., 1987). Implicit in this definition is the serial nature of search. The search space can be formed either by exteroceptive stimuli or by representations of objects, events, or problem states. Whatever the material on which it operates, the search process becomes a challenge for the cognitive system when the space of alternatives is large. In this regard, the vast variety of tasks, so often used in animal studies, featuring binary choices may be considered to be a trivial search problem (De Lillo, in press). In a binary situation, a random choice followed by a default strategy warrants the exhaustive exploration of the search space. By contrast, the serial exploration of a large number of alternatives requires the ability to keep track of the moves that the system is performing to avoid spending time and energy in reconsidering alternatives already explored. An obvious implementation of a nontrivial search task in the realm of animal behavior are tasks in which an animal has to explore a large set of loci, one after the other, in order to find items of food.Following the pioneering work by Olton and Samuelson (1976), a vast amount of research has been conducted on rats running the radial maze. In a search space that affords strong spatial constraints, such as a radial maze, it is possible for the subject to deploy an algorithmic strategy consisting, for example, of visiting in succession adjacent arms following a particular direction of travel. This strategy would allow very efficient (no revisits) exhaustive searches

Consiglio Nazionale delle Ricerche By means of an apparatus featuring a set of suspended baited containers, search abilities of 4 capuchin monkeys (Cebus apeUa) were evaluated. The experiment featured different spatial configurations of the search space. Results showed that monkeys exhaustively searched 9 containers spatially distributed as a 3 × 3 matrix, a cross, a line, or a circle. Search efficiency was higher when the search space featured either a linear or circular arrangement of containers. When faced with a linear arrangement of containers, the subjects developed principled search trajectories from 1 end to the other of the linear array. This behavioral regulation was independent from search efficiency as measured by the amount of visits to containers already explored. The data suggest that monkeys use either the travel distance or the cognitive costs associated with unprincipled travel trajectories as currency for regulation.

Search abilities of mice (Mus musculus domesticus) were evaluated using an arena closed by a ceiling in which 9 food sources (which mice could reach standing on their hind legs) could be arranged according to 2 configurations: a 3 x 3 square matrix and 3 clusters each containing 3 food sources. Testing conditions prevented olfactory and visual cues from being left after visits to food sources, and mice were able to choose alternative routes between food sources. Results showed that mice were more efficient with the matrix than with the cluster configuration. Sex differences were observed: Females improved their performance with both configurations, whereas males improved only with the matrix one. Mice did not develop evident search strategies that would minimize task complexity. Comparison with data published on capuchin monkeys revealed differences, with monkeys performing better with the cluster configuration than with the matrix and applying searching strategies.

To analyze how search strategies are adapted according to the geometric distribution of food sources, the authors submitted rats to a search task in which they had to explore 9 food trays in an open field and avoid visiting already-depleted trays. Trays were spatially arranged in 4 independent configurations: a cross, a 3 x 3 matrix, 3 clusters of 3 trays each, and a random configuration. Rats exhibited differential search efficiency as a specific effect of the susceptibility of the configurations to being explored in a principled way: crosses were first, matrices or clusters were in the middle, and random configurations were last. Although no exhaustive searches or highly principled patterns were observed in any of the configurations, performances improved as the sessions went by. Thus, structural affordances of the environment influence the construction not only of search strategies but also of information linked to where the reward is.

No abstract