ed, is better described by non symbolic models. The symbolic rules and the underlying logic of their implementation tend to produce rigid and brittle artificial intelligence systems.
As opposed to computational models, connectionists propose a brain model of the mind (Lloyd, 1990, pp. 116 118). Connectionist models are large networks of very simple computational units that are interconnected and running in parallel. In these models, knowledge is encoded within the pattern of the numerical strengths of the connections between the processors in the system, as opposed to being encoded in symbolic structures.
Connectionists contend that cognition is implemented in networks of neuronlike basic processors (Lloyd, 1990, pp. 118 122). Such processors, it is contended, share the following characteristics with neurons.
1. The processors are stupid. These processors are limited in capability to a simple summation of their inputs, and the outputs of these processors are unarticulated. These processors do not emit strings of interpretable code. Rath
...