FOM: Questions on higher-order logic JoeShipman at
Wed Sep 6 10:48:13 EDT 2000

Dear Bob,

This answer is indeed much more informative.  It means that using axioms of 
the form "phi is a 2nd-order validity" within the framework of the 1st-order 
language of set theory doesn't get you very far, basically because such 
statements are limited in logical complexity.

"An uncountable cardinal exists" is certainly important in "ordinary 
mathematics", for example to obtain standard counterexamples in general 
topology (see, for example, Munkres's introductory text "Topology", which 
relies heavily on the "minimal uncountable well-ordered set"). 

This doesn't quite do what I had hoped, which is refute "2nd-order logicism" 
by showing that non-logical axioms are necessary for an important 
mathematical theorem, since assuming "phi is a 2nd-order validity" within the 
framework of 1st-order set theory is not the same thing as assuming "phi" 
itself within the framework of 2nd-order logic.  Someone favoring an 
alternative foundation of mathematics via 2nd-order logic might therefore not 
find the argument "You can't show an uncountable cardinal exists" persuasive.

I find this situation a little confusing.  By working in SOL with standard 
semantics one can develop much of ordinary mathematics in a more natural way 
than via the usual foundation in first-order ZFC (or ZC, since the 
Replacement Axiom is not needed in the standard developments).  But 
apparently within set theory one needs more information than which sentences 
are second-order validities, there have to be axioms specifically about sets, 
such as the powerset axiom.

-- Joe Shipman

More information about the FOM mailing list