In Support of Pragmatic Computation

This paper argues that human cognition is structured to sup-port solving in unmanageably large problem spaces. Ma-chines face similar challenges, and can therefore benefit from the same kinds of approaches. Pragmatic human cog-nitive mechanisms already appear in AI artifacts, and more should be considered. Examples are drawn from human and programmed approaches to game playing, constraint satis-faction, and protein sequence alignment. People and artificially intelligent machines both solve problems. From a machine’s perspective, however, the ability of people to function not only adequately but ex-pertly is nothing short of miraculous. Human memories are clearly limited and often inaccurate, human algorithms are mysterious, and human heuristics are naïve. Although en