Ed Heil wrote:

> There's this intriguing book out by Gilles Fauconnier, called _Mental
> Spaces_ (translation of a French original).  It does away with
> traditional 'referential' semantic analyses of sentences in favor of a
> 'cognitive' analysis.  He sees sentences not so much as "carriers of
> meaning", where a sentence or a word "has" a meaning, but as minimal
> sets of instructions, which tell the hearer how to construct a meaning
> by manipulating, adding to, and subtracting from configurations of
> mental entities.
> Basic to the theory is the idea that meaning is distributed across
> sets of "mental spaces," so called because they are usually (in
> ordinary conversation) spoken of in metaphorical terms borrowed from
> physical spaces.  Any situation where meaning is compartmentalized is,
> in this theory, represented by different mental spaces: different
> times, different places, different contexts, different social spheres,
> reality and possibility, antecedent and consequent, and so on, all use
> exactly the same mechanisms.

This sounds similar to the localist case theory:

> Also basic to the theory is the idea that entities in one mental
> space may be linked to entities in another mental space (or in the
> same one!) via _connectors_ of different kinds.  Generally it is
> possible to use the name or description of an entity in one space to
> refer to an entity to which it is _connected_.  Identity is one
> connector among many.  Different kinds of analogy are also connectors,
> and so are well known functional relationships like those between
> authors and books -- "Plato is on the third shelf from the left."
> Connectors of another kind link roles and values.

There is also the idea of neural maps, pioneered by Michael Arbib
back in the 1960's ...

These ideas seem to be growing more popular, in linguistics,
psychology, and artificial intelligence.

> Definite and indefinite descriptions are easily explicated in mental
> space theory: a definite description identifies an entity already
> existing in a mental space; an indefinite description is an
> instruction to set up a new entity in a mental space.  (Broadly
> speaking.)
> Anyway, in mental space theory, negations are instructions to set up
> a mental space which contains the negated material, and to block the
> focussed entities from the negative space from being allowed into the
> reference space.  ("Reference space" is a relative term meaning
> something like the "parent" space with reference to which the
> negation-space is set up as a "daughter.").

A very cool analysis.