- #36
drnihili
- 74
- 0
With regards to the paradox of deductive logic. The most famous presentation of this is from Lewis Carroll (aka C.L. Dodgson) in What the Tortoise said to Achilles. It's reprinted in one of Hofstadter's books, I forget which one though. THe problem has very little to do with incompleteness in the Goedelian sense. Instead it's primarily a problem with the philosophical notion of a rule.
Any logical system must use rules to allow the transition from one statement to another. But these rules can also be written explicitly as a statement of the logic, typically as an axiom. Systems using "natural deduction" have lots of rules and no axioms. Axiomatic systems have lots of axioms and very few rules. The problem comes when one insists that every rule be reduced to an axiom. It turns out that's just impossible to do and still have anything resembling a logical system.
One way of trying to resolve the problem is via metalogic. In metalogic you prove things about a logical system, for example that any inference made in it is valid. In a sense though this merely puts the problem off for a bit. Metalogic must of course make inferences, and those inferences are subject to the same sort of objection which can only be resolved by appeal to meta-metalogic. And the list goes on.
Interestingly, this issue receives very little attention and is generally considered to not really be a problem. It's just accepted that no reasonably powerful logic system can show it's own soundness in the relevant sense. Every starting point can be questioned, but we do have to start somewhere.
The problems with inductive logic are far more trenchant. Perhaps the best read on those is Goodman's The New Riddle of Induction
Any logical system must use rules to allow the transition from one statement to another. But these rules can also be written explicitly as a statement of the logic, typically as an axiom. Systems using "natural deduction" have lots of rules and no axioms. Axiomatic systems have lots of axioms and very few rules. The problem comes when one insists that every rule be reduced to an axiom. It turns out that's just impossible to do and still have anything resembling a logical system.
One way of trying to resolve the problem is via metalogic. In metalogic you prove things about a logical system, for example that any inference made in it is valid. In a sense though this merely puts the problem off for a bit. Metalogic must of course make inferences, and those inferences are subject to the same sort of objection which can only be resolved by appeal to meta-metalogic. And the list goes on.
Interestingly, this issue receives very little attention and is generally considered to not really be a problem. It's just accepted that no reasonably powerful logic system can show it's own soundness in the relevant sense. Every starting point can be questioned, but we do have to start somewhere.
The problems with inductive logic are far more trenchant. Perhaps the best read on those is Goodman's The New Riddle of Induction