Sunday, December 18, 2005
Lattice of ontologies
Communication
from John Sowa
To separate the entangled points in a more extended summary, let me start with a description of the processes of perception along the lines that cognitive psychologists have been discussing for the past 40 years or so:
1. Incoming signals that impinge on our nerve endings are signs.
2. Perception involves the interpretation of those signs by
a process of retrieving one or more chunks of previously
experienced signs (traditionally called _percepts_) that
are assembled into a pattern that matches the new signs.
3. The process of perception is rarely an exact match of
old percepts to new signs. Instead, the assembly of
percepts has the nature of a hypothetical construction,
which may involve a considerable amount of deformation
and adjustment. The process is definitely fallible and
possibly ambiguous, in the sense that several different
selections of percepts could be adapted to match the
any given incoming pattern of signs.
4. Peirce called the process of perception a kind of abduction,
and that term has been applied by computational linguists,
such as Jerry Hobbs, to the process of parsing natural language
sentences (which are usually processed by methods similar to
the steps outlined in #2 and #3 above).
5. Not all percepts are associated with words in a natural
language, but many of them are. A natural language description
of the perceived experience could be generated by assembling
the words that correspond to the percepts in an natural language sentence
according to the syntax of the language and with the help of
auxiliary morphemes, such as function words and inflections.
Assuming a process of perception of this sort (which is a common hypothesis in cognitive psychology), I would expand my earlier sentence thus: the process of perception involves the matching of incoming signs to a hypothetical construction from stored percepts. The pattern matching process is fallible, and any particular construction is a hypothesis that could be, and often is, falsified or at least corrected by future experience.
What we call "objects" are the external projections of our internal constructions that have proved to be useful over long periods of time. Peirce said that a great many of our beliefs that have survived extensive testing are probably true within the limits of our ability to perceive and verify. We can probably be sure that our cherished beliefs will survive tests that are similar to our past experiences, but we can never be certain how far they can be trusted.
Note the ubiquitous signs: the original signals from the nerve endings, the store of percepts in the brain, the assemblies of percepts used to interpret the new signs, the stable constructions, which we call "objects", and the words and sentences we use to describe them. They're all signs.
Different psychologists may have different theories and different terminology for the processes of perception, but what they're discussing is still signs and signs of signs.
John