I leafed through this now 12-year old book, withdrawn from stock and sold by a public library. There's no biographical info in my copy - a page has been torn out possibly by some reader irritated by such trivia - but MIT clearly plays some part. The evolutionary strategy needed to produce such books seems to need [insert elaborate qualifying sentences here]: (1) Amiable relationships with a list of people; some mature in the field, e.g. Putnam; some famous for other reasons, e.g. Chomsky; some conceded to be exciting, e.g. Hofstader. (2) Writing style has to be adapted to a paper - an issue must be raised, thoughtful comments made at the appropriate level, which are then unanswered - definite answers are to be deplored, for one obvious reason. (3) Technical stuff must be referred to only in an overview sense. There is for example almost nothing about actual computers or hardware in this book, which seems odd. Nor is there awareness of the real structure of the brain. (4) The outside world must be referred to occasionally, and always in a way suggesting no criticism of authorities. (5) A certain palette of references from the past is needed, such as Turing, and the 'Cartesian Theater'. (6) It's permissible to puzzle over animal behaviour, but not to suggest animals have much the same abilities as people, but can't easily put the into action. (7) Padding can be provided by games, models, toys, and puzzles - the 'game' of Life is an example, though for some reason nobody ever speculates how changing the rules would change the game.



I don't think computer-generated papers are quite possible yet, but they're working on it. Whether any form of civilisation will survive to provide readership is another matter.

Скачать книгу бесплатно (pdf, 2.88 Mb)

---

Читать «Brainchildren: Essays on Designing Minds (Representation and Mind)»