Monday, April 29, 2013

Beneath the surface

Next up, a review published in Nature of the latest book by Douglas Hofstadter. Interesting, but God it’s long.


Surfaces and Essences
Douglas Hofstadter & Emmanuel Sander
Basic Books, 2013
608 pages
ISBN 978-0-465-01847-5

I finished this review and stored the file in the 'Nature' folder on my desktop, then emailed it to the editor. Or did I? A file, after all, was once a sheaf of papers, and a folder a cardboard sleeve for holding them. A desktop was wooden, and mail needed a stamp (no, it needed a little piece of adhesive paper). But all I did was use an interfacing device (named for the most superficial resemblance to a rodent) to rearrange the settings of some microprocessor circuits. To see that almost everything we say and do refers by analogy to other things we or others have once said or done – which is the main point of Surfaces and Essences – there is no better illustration than the way we have constructed our computer software as a conceptual and visual simulacrum of the offices our grandparents knew.

On the one hand this is kind of obvious. Why (science fiction writers take note) would we invent new categories and labels for things when we can aid comprehension by borrowing old ones, even if the physical resemblance is negligible? What cognitive scientists Douglas Hofstadter Hofstadter and Emmanuel Sander set out to show, however, is that this sort of elision is not merely a convenience: all our thinking depends on it, from the half-truths of everyday speech ("that always happens to me too!") to the most abstruse of mathematical reasoning. I was convinced, and the ramifications are often thought-provoking. But when you have had authors telling you the same thing again and again for 500 pages, perhaps you’ll believe it whether it's true or not. I’ll come back to that.

Hofstadter is famous for his earlier, Pulitzer-prize-winning treatise on how we think, Gödel, Escher, Bach (1979). Fans of that dazzling performance might find this book surprisingly sober, but it is also lucid and, page for page, a delight to read. Whether there is any conceptual continuity between that and this vision of how we think is debatable, except perhaps that GEB’s delight in puns here becomes an assertion that pretty much all our cognition depends on punning elevated to analogy.

The claim that drawing parallels between one thing and another central in our thinking seems obvious in art: analogies are the bread and butter (there we go again) of the visual, literary and theatrical arts. (Of these, the authors seem curiously unconcerned about anything except poetry.) Yet Hofstadter and Sander are really inverting that usual picture: it is precisely because the brain seems to be an analogy machine that art is possible and meaningful.

They focus most on the use of analogy in language. Moving steadily from words to phrases and narratives, they show just how deeply embedded is our tendency to generalize, compare, categorize, and forge links. Individual examples seem trivial until you realise their ubiquity: tables have legs, melodies are hauting, time is discussed in spatial terms, and idioms are invariably analogical, if you get my drift. Thus the lexical precision on which dictionaries seem to insist is illusory – words are always standing in for other words, their boundaries malleable. This flexibility extends to our actions: we see that a spoon can serve as a knife when no knife is available. (Indeed, the spoon then becomes a knife – objects may be fixed, but their labels aren’t.)

These arguments can be carried too far. Is to extrapolate to make an analogy (I expect the future to be like the past)? Is a Freudian slip an analogy, or mere crosstalk of neural circuits? Is convention an analogy (why don’t we write mc2=E?). Can we, in fact, turn any mental process into an analogy, by that very process of analogy? These are not rhetorical questions, for one might at least examine whether the same neural circuitry is involved in each case. But a lack of interest in neuroscientific examination of their idea is another of the book’s odd lacunae.

In fact this intriguing, frustrating book seems to exist almost in an intellectual vacuum. Unless one combs through the bibliography, one could mistakenly imagine that it is the first attempt to explore the notion of analogy and metaphor in linguistics, overlooking the work of Raymond Gibbs, Andrew Ortony, Sam Glucksberg, Esa Itkonen and many others. And one is forced to take an awful lot on trust. Hofstadter and Sander describe, for example, the evolution of the concept of ‘mother’ in the mind of a child as he learns to generalize from experience. It all sounds plausible, but the authors offer no empirical evidence for the developmental pathway they describe.

Neither is there any real explanation of why we think this way. Isn't it perhaps, in part, a way of minimizing the mental resources we need to engage in a situation, to avoid having to start from scratch with every unfamiliar encounter, object or perspective? Is it an adaptive technique for making predictions? Are mirror neurons part of a built-in cognitive apparatus for analogizing ourselves into others’ shoes?

The lack of historical perspective is also a problem – it is as if people always thought like they did now. Analogy was arguably all we once had for navigating experience, for example in the Neoplatonic idea of correspondences: “as above, so below.” This “just as… so…” thinking remains at the root of pseudoscience: the Moon influences the tides, so why not our body fluids? So how do we distinguish between good and bad analogies?

There are gems of insight in here, but again flawed by the authors’ relaxed attitude towards evidence. An analysis of Einstein's thought is splendid, explaining what is missing from conventional accounts of the discoveries of light quanta, relativity and mass-energy equivalence, namely what qualities distinguish Einstein from his peers. These qualities are convincingly shown to be analogical: Einstein was able to take leaps of faith and make connections that postpone rigour and are certainly not self-evidently true. One would usually call this intuition - Einstein's friend and biographer Banesh Hoffmann did just that. But it is shown here to be intuition based on a conviction that different areas of physics were comparable. In other words, his intuition is not left ineffable but is taken apart so that the inner workings – some of them – can be seen. As a result, we see that Einstein's insights were very subtle and not self-evidently true. Analogies, the authors say, left Einstein like J. S. Bach on hearing a theme: "very quickly able to imagine all of its possible consequences." All very fine – but such a detailed account must surely be supported by Einstein’s own words. Almost none are offered; we get only fragments of Hoffmann’s commentary.

Maybe at least some of these questions are merely evidence of the fecundity of the authors’ thesis. But they’d have more excuse for not answering them if they did not fill so much space with endless examples to ram the point home: they never give one when 60 will do, and I’m not exaggerating.

Such things make me wonder whom the book is for. Academic linguists will be irritated by the absence of references to other work. Physical scientists aren’t indulged until page 450. General readers could have been given the basic ideas, with equal conviction, in half the length, and will occasionally get the feeling they have been led along and then dumped. The thesis suggests no obvious mode of further development, no manner of testing and probing. It remains stimulating, but less would certainly have been more.

1 comment:

@CLANicholson said...

I found one of Hofstadter's previous books - I Am A Strange Loop - equally unnecessarily drawn-out. The guy needs a decent editor.