ABSTRACT
Communication, lending context sensitivity and then knowledge enhancement, becomes an ecological essentia] as cognition advances. Agents that can communicate have an extended range of knowing what is going on and of influencing it. From the viewpoint of interactive realism, real-world process rather than inventory of surface forms (words, syntax) becomes the focus. Communication is seen to be based on cognitive action to “simplify” in the face of obvious complexity, to reduce local apparent entropy in the face of ever-increasing system entropy. Communicating involves direct action supported by chemo-physical sensor-effectors that provide values along sensory parameters. The potential for communication goes way back in evolution, to molecular reactions like crystallizing and dissolving that are “sensitive” in “seeking” or “fleeing” certain things. Sensors, and cognition with them, develop through selective recombination. Then, in the face of ever-increasing complexity (“disorder,” the ever-increasing entropy of the Second Law of Thermodynamics), cognition advances by selecting sensory parameters as relevant, and building up a “simpler” (or at least more helpful) cognitive meta-universe.
The key link between cohorts is for communicative gestures to be held in common gestures that cue cognitive parameter values and processing. Recent experimental evidence shows that in human infants, mimicry and discriminations matching sound to mouth position are instinctive (automatic behavior; Meltzoff & Moore, 1983). Automatic mimicry (along with other cognitive symmetries) appears to push individuals toward sharing code systems under dynamic conditions. Search for the cause of linguistic drift whenever languages are in use led Candelaria de Ram (1988) to posit the existence and critical function of automatic mimicry even in adults as providing ineluctable pull toward new variants.
Simple neural net models simulating phonemic alternation (Loritz) and multistage, multilayer nets for parsing (Gigley) point the way to adequate modeling. Pragmasemantics™ process descriptions can be cast within a special grounded, sortal logic (Candelaria de Ram 1992a) as high-level computer programs. They register real-parameter values and send out signals, so as to act as communicators. These artifices are, by virtue of their underlying 426operational logic, networks. The idea is to make them like natural communicators, exposing individual-agent communication reasoning and behavior-control detail. As shown in the companion work (Candelaria de Ram, 1994), the operators (functors) of Pragmasemantic logic apply for molecular and cellular level process components just as they do in processes at the speech act level. Rudimentary operations in a dynamic combinatory system eventually enable full-fledged communication.
