Using a cognitive architecture for addressing the question of cognitive universals in cross-cultural psychology: The example of awalé

Abstract

A central theme in cross-cultural psychology is the extent to which cognitive mechanisms are universal, or, alternatively, are specific to a given culture. We propose a new way to tackle this question: to use the same cognitive architecture, implemented as a computer program, for simulating phenomena in which individuals from different cultures perform a task familiar to their own culture. The CHREST architecture (Gobet et al., 2001; Gobet & Simon, 2000) has simulated a number of empirical phenomena related to the western board game of chess. Here we show that a model implemented in the same architecture accounts for several phenomena in awalé, a board game from the mancala family, which is commonly played in western Africa and in the Caribbean. CHREST first learns chunks by scanning expert-level games, and then is placed in memory experiments and problem-solving situations similar to those used with human youngsters. The model replicates empirical phenomena on memory for awalé positions reasonably well, although not perfectly, and also learns to play a fair, but far from perfect game using pattern recognition. The assumptions that learning is mediated by the acquisition of a large number of chunks and that the capacity of visual short-term memory is limited to three chunks are important in explaining the empirical data for the two games. The implications for theory development in cross-cultural psychology are discussed.