Customer Reviews

Goodbye, Descartes: The End of Logic and the Search for a New Cosmology of the Mind

5 star:		(4)
4 star:		(2)
3 star:		(2)
2 star:		(2)
1 star:		(2)

 

 

The astonishing thing about human communication is not that it sometimes fails but that failure is so rare. Given the complexities of context, facial expressions, tone, body movements, and grammar, all going in at least two directions, it is truly incredible that it works so well. As the author points out by example, he can write a sentence that no one else has ever created before, and yet there is no difficulty in determining what he means. Understanding human language is a situation where our obviously finite brains are capable of resolving an infinite number of scenarios. The examples given in this book make you appreciate just how much "computing" power there is in the human brain.
Many of the theories regarding the instinctive understanding of human language, independent of word order, are considered and often questioned. The gross shortcomings of Artificial Intelligence (AI) are also raised and used to demonstrate that there is now no effective model for how humans process data and make "rational" decisions. Despite all the original promise and hype, AI has been used to solve few problems and even some of the reported successes are clearly very weak when thoroughly examined. Therefore, the argument throughout the book is that there needs to be a new approach to the problems of cognition
The arguments are presented in a thoughtful, detailed, and understandable manner. There are times when the arguments do get technical, but they are few and can be skipped without disrupting the flow of the material. At the end, Devlin also argues for a radical rethinking of the last three thousand years of traditional reasoning that dates back to the Greek origins of logic. He uses the phrase "soft mathematics" to describe what he believes the answer to be. Unfortunately, or perhaps necessarily, he is quite vague as to what it is. Devlin only points out that it will be something quite different from the current rigorous reasoning.
Raising some profound and fascinating questions regarding fundamental shortcomings in understanding the most human of activities, Devlin is at his best. Whatever your field of interest or background, if you are interested in thinking about thinking, then you must decipher the squiggles that appear on these ages.

Published in Journal of Recreational Mathematics, reprinted with permission.

The previous reader makes the same error with the monty Hall Problem as do many. New Scientist has been running a web discussion on this problem in its "biteback" section (http:www.newscientist.com), after a strongly positive review of "Goodbye Descartes" brought a small deluge of letters from readers who, like the previous reader, had misunderstood not only the correct Monty Hall solution given in the book, but along with it most of the book's argument. Wise readers will decide for themselves who is "right" on this issue. --Keith Devlin

For more than 2000 years, philosophers and scientists have attempted to use symbolic logic to investigate the structure of language and, by extension, the human mind. Our speech and thought processes, they believe, operate according to underlying rules that are rigorously mathematical. Devlin argues that this approach is a dead end and that we should pursue new avenues of research.

Much of the book is a critique of symbolic logic. Invented by Aristotle, it was merged with algebra and became a branch of mathematics and its most recent applications have been in artificial intelligence (AI) as well as the liguistic of Chomsky. What these disciplines have in common - what is "cartesian" about them - is their attempt to "captur[e] patterns of reasoning...in a pure fashion, isolated from context" and even meaning. In this view, computers are the perfect logic machines, processing info by manipulating symbols without understanding what they are doing.

The failure of AI to meet its original goals demonstrates, in Devlin's view, what is wrong with this approach. AI (or an "expert system") lacks common sense, whatever its daignostic capabilities, and cannot make judgments when unforseen or ambiguous situations arise. Consequently, AI cannot operate outside extraordinarily narriow confines and hence are unreliable in many applications. Computers have also failed to produce a human-like language. This is proof, Devlin says, that the human mind is more than a logic machine ("Smart meat" as the WIred crowd might argue) that acts according to rigid subsystems of logical rules: context and meaning matter. These arguments are convincing and cogently argued.

Unfortunately, Devlin's arguments of where to go from there are far weaker than his analyses of past failures. The last third of the book is a loose jumble of idaes and speculation. He wants to create a "soft math" to incorporate context, meaning, and the qualitative into the study of the human mins, but does not get beyond saying we need it. THis is a research agenda, but too vague to be of much use in my opinion. Of course, maybe I am expecting too much and his next book will cover that!

Unfortunately, his writing style is repetitive and gets bogged down in elaborate proofs and thought experiments - just the type of arcane stuff that keeps (or bars) many of us from reading more by academics.

So this is a mixed bag. The ideas on the human mind are well worth the effort, but getting through it is not fun, at least for me.

Negative reviews of this book "blame the messenger" -- Devlin does an extraordinary job of explaining the state of the art for his chosen topic, yet he is taken to task for not giving the problems he describes a nice, tidy, Hollywood ending. He is being honest. Such solutions don't exist.

His topic is an account of why artificial intelligence's assumptions have led to failure delivering the innovations of machinery seen in the Jetsons, innovations which seemed not all that far-fetched for the field in the intellectual "bubble" during its advent in the post-WWII era. Understanding the failure will touch on formal logic, mathematics, linguistics, behavioral decision theory (at the end of the book) and it's clearly unreasonable to expect Devlin will juggle them all judiciously. Trained as a linguist myself (with cursory knowledge in these other fields), I could take exception with simplifying assumptions / infelicities in his account of language's role, but such would be missing the forest for the trees.

More generally though, Devlin's critics are blaming the author for not ultimately curing for cancer -- assuming that were his topic -- in an analogous book whose aim is to illustrate why phrenology might be a wrong-headed place to look for a cure in the first place.

Some of the negative reviews of this book have good points. Devlin is less than scholarly, and (big surprise) he does not offer some kind of fully developed logic of the mind at the end. But jebus, what did you expect?

This is a great book, but you have to understand the style it is written in. It is written in an intellectual-fun style, attempting to provide a good, thought-provoking read. It does not provide iron-clad argumentation nor does it provide scholarly historical coverage. What it does provide is an easy to read whirlwind tour of the history of logic and linguistics, as well as hundreds of examples of situations where the logical approach to studying the mind has run into roadblocks.

I think it is a great book and well worth the read, especially for someone new to either logic or linguistics.

I notice that the other reviews of this book are all over the map. That distribution of opinion matches my own feelings. I found the first part of this book delightful and informative, as good as the author's previous book Mathematics: The Science of Patterns. The last part I found unreadable. Goodbye Descartes begins with a persuasive argument that intelligence is not computation and ends up speculating on what intelligence DOES involve. The latter is admittedly a next to impossible task, but the author's attempt to combine logic and sociology to form "soft mathematics" left me unimpressed. Overall, a disappointment.

Devlin's book is really superficial.
For example, it is difficult to believe that anybody can write something like this:
"This rule of deduction is still known today by its original Greek name, 'modus ponens' " (35).
'Modus ponens' cannot be its original Greek name, for it is Latin.
In fact the rule was named as such in the Middle Ages.
I suspect that Devlin cannot tell the difference between Latin and Greek.

My original review was erroneous. Many examples were counterintuitive, but it turns out they were correct. The book is an excellent exercise.

Devlin's historical development of the Cartesian influence on science in general and (in particular) as applied to the attempts at 'artificial intelligence' are enlightening and very well presented. This is yet another piece of the puzzle one may call 'modern science encounters its limits.' This puzzle, or pattern, includes other limitations on science coming to the fore, sometimes because of limits on knowledge (yes, there are such limits) or on the scientists themselves. Hence we have Kinsey's use of pedophiles to promote infant 'sexuality', Margaret Mead's 'fateful hoaxing', and the ever increasing, brilliant scientific challenges to 'natural selection.' It must be a true spiritual shock to many buckaroos of science that HAL is, and will remain, a film fantasy. Devlin's flaws are only apparent when he leaves his field - for example, the old, highly inaccurate Galileo vs. the Church story. (Actually, the Church accepted Galileo's theories long before other scientists of his day, notwithstanding Galileo's friendship-gone-sour with the pope, a much more accurate and amusing story than is recounted by Devlin.) Devlin explains at great length that he is not really a heretic when he proposes inevitable limits on the abilities of artificial intelligence, but he may discover that the old guard of Cartesian (read: Victorian) scientists are as difficult to deal with as the pope was for Galileo, especially if the gravy train is derailed. Overall, Devlin's discussion on the development of logic is worth the price of the book, and his enthusiasm for mathematics is appropriate and delightful to read.

I studied symbolic logic and linguistics in college, so it was enjoyable to take a trip down memory lane with this book as it covered those topics and tried to weave them together. I say "tried" because I finished the book unconvinced there is any relevant relationship between the two and unsure what Devlin's point was. The book seemed very muddled to me, but maybe I just wasn't paying close enough attention. Overall kind of boring.