Computational modeling of numerical cognition

Numerical cognition has been studied in both human and animal species for a long time. However, the computational basis of number representation and numerical skills has received very little attention, as compared with the computational basis of language processing, for example, reading (see Zorzi, 2004, for a review). In general, computational modeling is a powerful tool in cognitive science to evaluate or compare existing verbal theories (e.g., box-and-arrows models) and to make novel experimental predictions. In contrast to the loose formulation of traditional verbal theories, computational models need to be explicit in any implementational detail and can produce highly detailed simulations of human performance (e.g., they can be explicitly tested on any number of stimuli). Moreover, the performance after a “lesion” to the model can be readily compared to the behavior of neuropsychological patients.