Dennett’s Special Pleading

Daniel Dennett has a review out in Artificial Intelligence of Douglas Hofstadter’s 2007 book I am a Strange Loop, and Marvin Minsky’s 2006 book The Emotion Machine.  Dennett admits that he is biased for these authors, that they mainly rehash their famous books of twenty years ago, and that they fail to meet the standards of all related academic disciplines, and so are rejected by those academics.  But, well, he just likes them anyway: 

I am much too close, personally and intellectually, to Doug Hofstadter and Marvin Minsky to write a proper academic review of these wonderful books … I travel in several quite different academic circles and I find that each gang has its particular way of not taking these thinkers seriously. The neuroscientists deplore the absence of rigorous experiments and the refusal of both Hofstadter and Minsky to canvass the relevant experimental literature thoroughly and explicitly. Where are the data? The philosophers of mind, at the other extreme, find few formal arguments and a frustratingly cavalier refusal by the authors to define their terms at the outset. Where are the proofs?

The cognitive psychologists, in the middle, are offended by the fact that neither Hofstadter nor Minsky sees the need to adjudicate between all the competing models and theories that have been painstakingly developed and defended, and instead offer their own impressionistic and oversimplified sketches. Where are the models that make testable predictions? The artificial intelligence crowd wants to see a running demo program. Where is the code? It’s all just speculation! And then there is the style, too clever and playful, apparently written for bright high school students, not professors and graduate students. ….

People shouldn’t try to do what they are no good at doing, and it is also true that this genre of free-wheeling "informed speculation" can be an utter waste of time in the hands of somebody who isn’t well-informed and hasn’t got a disciplined imagination. Very few people are equipped to do this kind of work well, but both Hofstadter and Minsky have had years of practice using the best available thinking tools. They haven’t just marinated their minds in Lisp and other computer languages; they have built and dismantled computer models of a wide variety of phenomena, and sympathetically explored the efforts of others. ….

As Hofstadter says ruefully in his preface, in spite of all the attention and praise lavished on his masterpiece, Gödel, Escher, Bach (1979) , its "fundamental message . . . seemed to go largely unnoticed" (p xiii). Minsky’s 1986 book, The Society of Mind, was also lauded but underestimated and misunderstood. The books under review attempt to tell the tales again, with improvements and embellishments, and I think they both succeed, at least for me.

Surely Dennett should wonder: how strongly can he count his personal favorable impression, relative to academic rejection and his personal biases?  How much should the fact that Dennett, Minsky, and Hofstadter are relatively popular writers count? 

GD Star Rating
Tagged as: ,
Trackback URL:
  • g

    Perhaps he does wonder that, but then considers that he’ll be providing more information to his readers if he just gives his impressions, and a summary of opposing views, as they are without pre-averaging them? (I suspect not, but I think that would be quite a reasonable way of proceeding.) Incidentally, it’s “Dennett”, with two “t”s.

  • g, thanks, spelling error now fixed.

  • cw

    I read GEB in high school, and to this day it is one of my favorite books, both for the ideas it contains and for the masterful structure and presentation. So I was surprised and disappointed that a large chunk of “I am a Strange Loop” consisted of reworking ground that Hofstadter had already covered. The most interesting idea in the book was the notion that our representations of other people consist of coarse-grained strange loops, that are essentially dispersed parts of a person’s consciousness. It’s a striking idea, and he talked about dispersed consciousness in GEB, but I question whether my representation of another person is actually a strange loop, because it does not necessarily point back to itself. Rather it points outward, to the person it represents. It’s possible that we could develop representations of people that are so complex that they could become self-referential, but it’s not a necessary outcome. (I realize this comment is not exactly on topic, but I wondered if anyone had any different thoughts on the book.)

  • Recovering irrationalist

    I read it 2 months ago, this review sums up my feelings better than I could.

  • It is quite possible that the standards of all relevant academic disciplines are no good and should be ignored. The books in question are probably closest to philosophy of mind in subject and tone, but I have no problem at all with ignoring the standards of that field, which has been notoriously unproductive (if the authors ignored philosophy of mind and just ended up recapitulating it, rather than being original, that would be a problem, but I don’t think that’s the case). AI was founded as an interdisciplinary field that originally had very loose standards, but has since congealed in a way that Minsky (one of the founders) doesn’t like, so he largely ignores those as well.

    Academic fields form around problems or areas of discourse that are well-defined, often where there are simple theories, formalisms, and simplifying assumptions that everyone can work from. But the theories are often wrong. The things that are easy to formalize may not be the things that matter. It’s like the drunk searching for his car keys where the light is rather than where he dropped them. So AI and philosophy have their standards but it is far from clear that those standards are helping people look in interesting places.

    And BTW, I believe economics has this problem in spades, but that’s a flame for another time.

  • I happen to love GEB. But I recognize that its “solution” to the problem of consciousness has its distinct limits. I have not had time to read the new book.

  • Mark

    Doug Hofstaders is quoted as saying something to the effect that analogy is almost everything with respect to thinking.

    My own analogy that I use to think about Hofstader’s ideas is a poker tournament. You have a defined rule book and you start out with a bunch of players and tables. The winners of each table move on, until you arrive at a winner. Note, you don’t need a optimal hand to win each table or the tournament.

    But how does the winner of the tournament inform all the tables that gave rise to it?

    That’s the tricky part. In order for you to have won this table you must have won the prior table and so forth. So how does evolution give rise to that?

    And how come I’m only aware of the end of the card game, the higher order of representation?