Abstract

We discuss recent computational network models of elementary number processing. One key issue to emerge from this work is a crucial distinction between symbolic and non-symbolic number representation, and the related distinction between number-selective and number-sensitive coding. Empirical predictions from the models were tested, and are here summarized. Another issue is the relation with task-based decision making mechanisms. In both lab and real-life settings, representations are seldomly accessed in a task-neutral manner, rather subjects are usually presented with a task. A related theme is the functional association between number representations and working memory. In these issues also, both modeling and neuroimaging work is summarized. To conclude, we propose that the combined modeling-neuroimaging approach should be extended to tackle more complex questions about number processing.