|
|
|
|
|
(a) M is said to be strong-monotonic on L just in case, for each with and each , we have . |
|
|
|
|
|
|
|
|
(b) M is said to be strong monotonic on just in case M is strong-monotonic on each . |
|
|
|
|
|
|
|
|
(c) M is strong monotonic just in case M is strong monotonic on each . |
|
|
|
|
|
|
|
|
So, a strong-monotonic scientist, upon being fed a text for a language, outputs a chain of hypotheses such that if index j is hypothesized after index i, then . A consequence of this requirement is that if a scientist incorrectly hypothesizes that a particular element belongs to the target language, then it cannot revise this assumption by emitting a hypothesis that excludes this element. |
|
|
|
|
|
|
|
|
(a) |
|
|
|
|
|
|
|
|
(b) . |
|
|
|
|
|
|
|
|
We thus consider only the global version in the sequel. |
|
|
|
|
|
|
|
|
Wiehagen [200] suggested that the requirement of strong monotonicity is too stringent. He proposed the weaker notion of monotonic strategy, which simply requires that a scientist's successive hypotheses be more general only with respect to the target language. More precisely: |
|
|
|
|
|
|
|
|
5.50 Definition (Wiehagen [200]) |
|
|
|
|
|
|
|
|
(a) M is said to be monotonic on L just in case, for each with and each , we have . |
|
|
|
|
|
|
|
|
(b) M is said to be monotonic on just in case M is monotonic on each . |
|
|
|
|
|
|
|
|
Thus a monotonic strategy is allowed to correct its mistaken assumption that certain nonelements of L belong to L, but once it has correctly concluded that an element of L belongs to L it is not allowed to output a hypothesis that contradicts such a conclusion. The reader should note that we deliberately omitted the third clause in the above definition because the global version of monotonicity is equivalent to the requirement of strong |
|
|
|
|