There have been many arguments about what makes people intelligent. It has been argued that the mind gains its power from its ability to reason symbolically. This belief is based on the notion that knowledge is stored in a form directly equivalent to a set of symbols, and that reasoning is manipulation of these symbols. One alternative position is that reasoning doesn't depend on anything so simple: rather, knowledge is stored in a distributed form, and is best represented as a kind of network. In this case, reasoning is usually represented as the adjustment of weights on the network's nodes. Such models of reasoning are sometimes described as non-symbolic.
Here, we adopt Newell's view expounded in his Unified Theories of Cognition - that intelligence involves both symbolic and non-symbolic processing, and that symbolic reasoning is related to tasks which require conscious thought with non-symbolic reasoning being associated with processing at an unconscious level. From this viewpoint a non-symbolic approach is a strong candidate for the way in which subsymbolic processing occurs.