|
|
|
|
|
5.68 Claim Suppose M is recursive-reliable. Then for all , there exists a such that and . Moreover, there is a recursive function t such that, given s , as above. |
|
|
|
|
|
|
|
|
Proof (Claim 5.68): If such a does not exist then M converges to M( s ) on all functions consistent with s . However, M( s ) can be an index for at most one function consistent with s . This contradicts the fact that M is recursive-reliable. The moreover clause is obvious. |
|
|
|
|
|
|
|
|
Recall that . Fix a recursive-reliable scientist M. We show that M fails to identify . By Kleene's recursion theorem (Theorem 2.3), there exists an e such that j e may be defined as follows: let s 0 be such that , and . Clearly, . Also, . Thus, . |
|
|
|
|
|
|
|
|
In the course of identification a scientist may conjecture a hypothesis, abandon it after seeing more data, and revert to the abandoned hypothesis once again. A sensible strategy would be to consider only those scientists that never go back to an abandoned conjecture. |
|
|
|
|
|
|
|
|
(a) M is decisive on f just in case, for all l, m, with l < m < n and , we have . |
|
|
|
|
|
|
|
|
(b) M is decisive on just in case M is decisive on each . |
|
|
|
|
|
|
|
|
(c) M is decisive just in case M is decisive on . |
|
|
|
|
|
|
|
|
(d) ![0119-016.gif](0119-016.GIF) |
|
|
|
|
|
|
|
|
(e) ![0119-017.gif](0119-017.GIF) |
|
|
|
|
|
|
|
|
We shall now see that scientists can adopt a decisive strategy without loss of inductive competence. |
|
|
|
|
|
|
|
|
5.70 Proposition (Schäfer-Richter [167]) [Ex]decisive = [Ex]class-decisive = Ex. |
|
|
|
|
|
|
|
|
Proof: Clearly, . We show that ![0119-019.gif](0119-019.GIF) |
|
|
|
|