Comparing the symbol string the rules

278 Chapter Six—Thinking Is Social

Note that some theorists code binary node activations in {0, 1} and some in {−1, +1}. The difference between these two implementations is seen in the effect of a node when it is turned off; a node with zero value has no effect on the units it is connected to, while a node value of −1 ac-tively tends to inhibit them. This results in some paradoxical cases, where in order to achieve the desired result positive relationships must actually be coded using negative connections. The present model emu-lates connectionist networks with {0, 1} activations and as such avoids that anomalous instance.

When the program is executed, a population of random symbol strings is generated. A string is evaluated by comparing it to each rule in the rule file. Weighted matches between the test string and the rules are summed through the rule set. A larger total indicates that more con-straints were satisfied. The ACM algorithm is applied by selecting an in-dividual and a neighbor, comparing their evaluation totals, and interact-ing when the neighbor’s total is greater than the individual’s.

Experiment Six: Symbol Processing 279

population dynamics allow important insights into the cognitive opera-tions involved. Part of the definition, indeed, of a stereotype is that it is a belief shared by a group about members of another group. ACM shows the development of stereotyped thinking as it spreads through a popula-tion. A set of commonly held beliefs is arranged in various ways until the best explanations are found. The search is shared by the population, and the successful results spread to all members.

Because trips through the graph vary in the number of nodes they pass through to reach the goal, an adjustment in the performance mea-sure was necessary. If simple tour length were used as a performance