COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	Addenda to MENTAL
C00010 ENDMK
C⊗;
Addenda to MENTAL

creativity, counterfactuals, persuasion

When a philosopher tries to explicate a common sense notion like
"X believes Y", he is likely to come up with a single rather long
sentence whose words are to be taken in their ordinary English
meaning.  Such explicit definitions are useful when available,
but we know from science that there often isn't such a definition
corresponding to the usage of the term in a good scientific
theory.  Thus, "the internal energy of a certain mass of gas"
is meaningful only in the context of the theory.

	Philosophical logicians write as though they believed
the above but rarely seem to advance comprehensive theories.

John owes Bill a horse.
Terry thinks this involves thoughts.  In my opinion, humanity has
adopted an abstract notion of debt that doesn't necessarily
involve any human knowledge.  Thus an automatic action by
one company's machine may cause it to incur a contingent
liability that may never be known by any human because the
contingency may not occur or because the liability may be
lumped with other debts and never be considered separately.

This paper must be regarded as exploratory.  It would be better
if it could be much more formal and technical.

Technical issues:

	The concept of one theory approximating another needs to
be formalized.  In the case of physics, one can think of various
kinds of numerical or probabilistic approximation.  I think this
is misleading and won't help with explicating such concepts as
%2intentional action%1 as meaningful in approximate theories.
Instead it may go something like this:

	Consider a detailed theory ⊗T that has a state variable
⊗s.  We may imagine that ⊗s changes with time.
The approximating theory ⊗T' has a state variable ⊗s'.  There is a
predicate ⊗atp(s,T') whose truth means that ⊗T' is applicable when
the world is in state ⊗s.  There is a relation ⊗corr(s,s') which
asserts that ⊗s' corresponds to the state ⊗s.
We have

!!:	%2∀s.(atp(s,T') ⊃ ∃s'.corr(s,s'))%1.

Certain functions
%2f1(s), f2(s),%1 etc. have corresponding functions %2f1'(s'),
f2'(s')%1, etc.  We have relations like

!!:	%2∀s s'.(corr(s,s') ⊃ f1(s) = f1'(s'))%1.

However, the approximate theory ⊗T' may have additional functions
%2g1'(s')%1, etc. that do not correspond to any functions of ⊗s.  
Even when it is possible to construct %2g%1s corresponding to the %2g'%1s,
their definitions
will often seem arbitrary, because the common sense user of ⊗g1' will
only have used it within the context of ⊗T' 

2. Can a second order definition be converted back to a first order
definition?

3. General remark about separating technical from ideological issues.

	I thank the editor for the reference to (Dennett 1971) who
justifies ascribing mental qualities to machines on similar grounds.
I would require stronger evidence than Dennett for ascribing particular
beliefs.  In particular, I would not always ascribe a belief in
analytic truths, and I don't think that a desire to survive can
yet be ascribed to most machines.  (see p. 93 of Dennett).
I agree that the behavior of the mouse in seeking the cheese and
avoiding the cat can best be described by ascribing desires and
beliefs and some reasoning ability, but I believe that an economical
description of the mouse's mental structure will involve a fragment
of logic and few beliefs.  For example, we should be more ready to
ascribe instantiation of general beliefs than the formation of
general beliefs by generalization.  Thus deciding what mental structure
to ascribe to the mouse is a non-obvious technical problem.
I agree with the metaphor of %2taking out a loan of intelligence%1.

Moliere's doctor ascibes a %2virtus dormativa%1 to a sleeping powder.
This was regarded by Moliere and most everybody as pompous nonsense.
However, the hypothesis of a %2virtus dormativa%1 takes out a loan
of theory that could conceivably have been repaid.  Suppose, for
example, that the ability to put someone to sleep were conserved, i.e.
if you mixed the powder with something else, burned it and extracted
various substances from the residue, and it turned out that the total
ability to put people to sleep of the transformed substance turned out
constant.  Modern medicine refutes the hypothesis of %2virtus dormativa%1,

because the ability of a substance to induce sleep depends on a detailed
chemical structure and the ability to move this structure to
specific places in the body.  The similar hypothesis of electric charge
has been confirmed.