Stances: Production systems

Cognitive Universals " Fiiirre! " If someone were to shout that while you were in the midst of reading this essay, you would, like most people, stop reading and look around the room for the source of the shout, or the fire itself. You would also consider whether or not the likelihood of a fire was sufficiently high to cause you to take appropriate action e.g., locate a fire extinguisher, call the fire department, or leave the room. Of course you have not been sitting around waiting for someone to yell " fire! ". You were engaged in some task (unrelated to fires) and yet were able to react to that stimulus. In contrast, if you were a police officer who was being trained at a firing range, you would be doing exactly that (i.e., waiting for someone to yell " fire! "), and when you heard " fire " you would pull the trigger on your weapon, rather than look around for the source of the cry. In both situations, two kinds of knowledge would be brought to bear in responding. On the one hand you would have facts (e.g., that someone yelled " fire " , that you are in the library or on a firing range, etc.), and on the other, you would have skills for how to respond (e.g., how to find an exit, how to fire a gun, etc.). Furthermore, responding could involve both pre-existing knowledge of both types (e.g., that you are on a firing range, or how to fire a gun), and newly acquired knowledge of both types (e.g., that someone yelled fire, or where to look for the exit in this particular room). This example illustrates the cognitive universals that builders of production system models believe to be fundamentally important aspects of intelligent behavior (both human and artificial). First, at any point in time there are always many possible actions from which one must be