> Na, I think the above is wrongly states, we needssetsdefined afterquantifierfreeformulas and in aquantifierfreecontext, a simple
> trial would be the following.
>
> To the language of FOL- (=FOL - quantifiers + {|}) add the following
> rule.
>
> If phi(x,y1...yn)is a formula in which x,y1,...,yn are the sole
> variables in it, then
>
> {x|phi(x,y1...yn)} is a term.
>
> The intended meaning for {x|phi(x,y1...yn)} is the set of all quine
> atoms that satisfy phi.
>
> Now add extra-logical symbols =,e to represent equality and
> membership, and add identity axioms stated in the usual manner for
> first order logic, add the following axioms:
>
> Axiom of Singletons: y e x -> y e y
>
> Axiom of Membership: y e y & z e y -> y=z
>
> Comprehension axiom schema: n=1,2,3,...;if phi is a predicate symbol
> and {y|phi(y)} do not occur in phi(y) then
> [y e {y|phi(y)} <-> y e y & phi(y)] is an axiom.
The main tool here is that comprehension axiom which is not written
correctly:
for n=1,2,3,...if phi(z,x1..xn)is a formula in which {y|phi(y,x1..xn)}
do not occur, and in which z,x1..xn are the sole variables occurring
in it, then
z e {y|phi(y,x1..xn)} <-> z e z & phi(z,x1..xn)
where phi(z,x1..xn) is obtained from the formula phi(y,x1..xn) by
replacing each occurrence of the variable symbol y in it by the
variable symbol z.
it would be nice if we can further restrict this schema to explicitly
define only finite sets.
/
Also a shred of an idea that occurred to me is that if it would be
possible to use special kinds of schemes in which variables can only
be substituted simultaneously across separate sentence of the
scheme,lets call it super-scheme
Super-scheme of DOMAIN:
If simultaneous substitution of x is assumed across sentences in the
below scheme, then one of those sentences is an axiom.
for n=1,2,3,.........
x= X_n.
To understand that, one could consider the whole schema as a one unit
although composed of different sentences, so if x is replaced by r for
example in sentence x=X_1 to become r=X_1, then when n=2 we must have
the same replacement of x by r to obtain r=X_2 and vice verse, and the
same goes across all values of n. and when we replace x by s for
example then this must be done simultaneously across the whole schema.
I don't know perhaps this is equivalent to some infinite quantifier
free first order language actually since what discriminate sentences
is simultaneous substitution.
This enforces X_1,X_2,... to range over the whole domain.
/
Another shred of idea, is that I'm seeing the set construction in
general is extensive, when we say for example the set of all oranges
in my hand, this formally translates to a statement that is not only
involved with the oranges in my hand but also to refute all other
objects from being members of that set, so there is virtually a scan
of the whole domain made just to grasp a simple notion of for example
two oranges in my hand, while the set itself is finite, the way it is
defined formally is potentially infinite, that is if not actually so,
and actually for the case of finite sets it looks as if the rejection
load is much more than the acceptance, which seems to be a long tale
that is not needed really. In ordinary day life we grasp finite sets,
heaps, collections, conglomerates, etc.. in a finitary way of
reasoning, and not by this exhaustive extensive manner.
Why we can't have a finite way of defining sets like saying for
example:
Construct Set X by the following Process:
Put K in X
Put L in X
.
.
.
Put J in X
STOP
This is easier procedure to define the set X, it is finite, the whole
scan of what is other than the members of the set which is what caused
the potential infinite process of defining sets traditionally, I say
the whole of that scan will be replaced by one word that is STOP,
which conveys what we want actually.
I don't know how to put this in appropriate formal context, but if one
succeeds along that line then we can deal with finite sets without
being involved in any procedure that range over the whole domain.