The neuroanatomical similarity of all cortical areas naturally leads
to a conjecture that there may be some commonality in the computation
performed across functions from vision to abstract thought. From the
perspective of traditional AI this seems a naïve, utopian
suggestion. What commonality is there between estimating fundamental
matrices for reconstruction of 3D scenes and chaining through sets of
domain knowledge rule sets? The mechanism described here evolved from
an effort to reverse engineer the visual cortices to create a viable
machine vision mechanism, but has proven to be capable of solving
problems of thought in domains seemingly remote from vision. The
common element in the solution of each of these problems is the
discovery of a composition of mappings which reflect the essential
structure of the problem. The general mechanism has come to be called
a map-seeking circuit because its mathematical expression has an
isomorphic implementation in realistic neuronal circuitry.