I Am a Strange Loop [Paperback]

Let's start by stating a simple fact: nothing by Hofstadter can ever be anything but fascinating (even his terrible translation of Eugene Onegin had a very interesting introduction). Now we've got that out of the way, let's admit that this book isn't quite up to par with his others (of which my favourite, for the record, is Fluid Concepts and Creative Analogies).

There's not really anything here that we haven't seen before: we have Godel's theorem, self-engulfing camera systems and other paradoxes from GEB; science-fiction thought experiments from The Mind's I; the Careenium from Metamagical Themas; blurred souls and personalities from Le Ton Beau. We get the sense that Hofstadter is frustrated that people still don't quite 'get it', which is fair enough except that I and most of his core readership probably *do* get it.

Now, naturally this doesn't detract from the fact that it's a lovely read as ever (although I miss Hofstadter's playfulness, which seems to have diminished over the years). The chapters on Godel, particularly, are well-explained and do clarify the relationship Hofstadter sees between Godel and the brain. Also, he spends some time expanding on the themes introduced in Le Ton Beau, that a person's spirit is not just held in a single brain but spreads through those they influence. He gives this more rigour than before, likening it to a virtual machine on a computer, creating a (slightly imperfect) version of another program. And his discussions of levels of soulhood (framed in musings about his own vegetarianism) are thought-provoking, particularly the idea that the cut-off point for having a soul could be the ability to have a concept of 'friend'.

What I'd have liked to see was more speculation from Hofstadter's actual area of expertise. He gives the impression that representational power simply appears within a system as soon as it has enough stuff going on on the lower level (this particularly strikes you when reading about his Careenium metaphor), whereas in his actual research he knows perfectly well that it takes a lot of work to make real representation (and indeed he often berates other AI researchers who miss this point). He discusses theories of Dennett, Searle and other philosophers, but we've seen this before and it would be nice to see some mention of the things we have learned in neurology, psychology and evolutionary biology since GEB.

A Hofstadter book is an all-too rare event. Here's hoping we get another and it has more meat to it.

Having read the book I could not disagree more with the reviewer who gave one star. This book does not have any agenda in the slightest and I doubt very much the reviewer really read the book as if anything Hofstadter argues that consciosness exists to a varying degree depending on the sophistication of the brain possessed by the animal and so one could infer that he advocates meat eating! which he doesn't do either.

Although elements of the book seem abstract and sometimes a little off track, once fully read I began to appreciate the care the author has taken to tie all of his anecdotes and metaphors together to provide his wonderful vision of what a soul means.

In addition to all this, the book is peppered with great: scientific, musical, literary and everday analogies that really clarify the points he is trying to make.

Philosophy, to those who are disdainful of it, is a sucker for *a priori* sleights of hand: purely logical arguments which do not rely for grip on empirical reality, but purport to explain it all the same: chestnuts like "cogito ergo sum", from which Descartes concluded a necessary distinction between a non-material soul and the rest of the world.

Douglas Hofstadter is not a philosopher (though he's friends with one), and in "I am a Strange Loop" he is mightily disdainful of the discipline and its weakness for cute logical constructions. All of metaphysics is so much bunk, says Hofstadter, and he sets out to demonstrate this using the power of mathematics and in particular the fashionable power of Gödel's incompleteness theory.

Observers may pause and reflect on an irony at once: Hofstadter's method - derived *a priori* from the pure logical structure of mathematics - looks suspiciously like those tricksy metaphysical musings on which he heaps derision. As his book proceeds this irony only sharpens.

But I'm getting ahead of myself, for I started out enjoying this book immensely. Until about halfway I thought I'd award it five stars - but then found it increasingly unconvincing and glib, notably at the point where Hofstadter leaves his (absolutely fascinating) mathematical theorising behind and begins applying it. He believes that from purely logical contortion one may derive a coherent account of consciousness (a purely physical phenomenon) robust enough to bat away any philosophical objections, dualist or otherwise.

Note, with another irony, his industry here: to express the physical parameters of a material thing - a brain - in terms of purely non-material apparatus (a conceptual language). In the early stages, Professor Hofstadter brushes aside reductionist objections to his scheme which is, by definition, an emergent property of, and therefore unobservable in, the interactions of specific nerves and neurons. Yet late in his book he is at great pains to say that that same material thing *cannot*, by dint of the laws of physics, be pushed around by a non material thing (being a soul), and that configurations of electrons correspond directly to particular conscious states in what seems a rigorously deterministic way (Hofstadter brusquely dismisses conjectures that your red might not be the same as mine). Without warning, in his closing pages, Hofstadter seems to declare himself a behaviourist. Given the excellent and enlightening work of his early chapters, this comes as a surprise and a disappointment to say the least.

Hofstadter's exposition of Gödel's theory is excellent and its application in the idea of the "Strange Loop" is fascinating. He spends much of the opening chapters grounding this odd notion, which he says is the key to understanding consciousness as a non-mystical, non-dualistic, scientifically respectable and physically explicable phenomenon. His insight is to root consciousness not in the physical manifestation of the brain, but in the patterns and symbols represented within it. This, I think, is all he needs to establish to win his primary argument, namely that Artificial Intelligence is a valid proposition. But he is obliged to go on because, like Darwin's Dangerous Idea, the Strange Loop threatens to operate like a universal acid and cut through many cherished and well-established ideas. Alas, some of these ideas seem to be ones Douglas Hofstadter is not quite ready to let go.

The implication of the Strange Loop, which I don't think Hofstadter denies, is that a string of symbols, provided it is sufficiently complex (and "loopy") can be a substrate for a consciousness. That is a Neat Idea (though I'm not persuaded it's correct: Hofstadter's support for it is only conceptual, and involves little more than hand-waving and appeals to open-mindedness.)

But all the same, some strange loops began to occur to me here. Perhaps rather than slamming the door on mysticism, Douglas Hofstadter has unwittingly blown it wide open. After all, why stop at human consciousness as a complex system? Cconceptually, perhaps, one might be able to construct a string of symbols representing God. Would it even need a substrate? Might the fact that it is conceptually possible mean that God therefore exists?

I am being mendacious, I confess. But herein lie the dangers (or irritations) of tricksy *a priori* contortions. However, Professor Hofstadter shouldn't complain: he started it.

Less provocatively, perhaps a community of interacting individuals, like a city - after all, a more complex system than a single one, QED - might also be conscious. Perhaps there are all sorts of consciousnesses which we can't see precisely because they emerge at a more abstract level than the one we occupy.

This might seem far-fetched, but the leap of faith it requires isn't materially bigger than the one Hofstadter explicitly requires us to make. He sees the power of Gödel's insight being that symbolic systems of sufficient complexity ("languages" to you and me) can operate on multiple levels, and if they can be made to reference themselves, the scope for endless fractalising feedback loops is infinite. The same door that opens the way to consciousness seems to let all sorts of less appealing apparitions into the room: God, higher levels of consciousness and sentient pieces of paper bootstrap themselves into existence also.

This seems to be a Strange Loop Too Far, and as a result we find Hofstadter ultimately embracing the reductionism of which he was initially so dismissive, veering violently towards determinism and concluding with a behavioural flourish that there is no consciousness, no free will, and no alternative way of experiencing red. Ultimately he asserts a binary option: unacceptable dualism with all the fairies, spirits, spooks and logical lacunae it implies, or a pretty brutal form of determinist materialism.

There's yet another irony in all this, for he has repeatedly scorned Bertrand Russell's failure to see the implications of his own formal language, while apparently making a comparable failure to understand the implications of his own model. Strange Loops allow - guarantee, in fact - multiple meanings via analogy and metaphors, and provide no means of adjudicating between them. They vitiate the idea of transcendental truth which Hofstadter seems suddenly so keen on. The option isn't binary at all: rather, it's a silly question.

In essence, *all* interpretations are metaphorical; even the "literal" ones. Neuroscience, with all its gluons, neurons and so on, is just one more metaphor which we might use to understand an aspect of our world. It will tell us much about the brain, but very little about consciousness, seeing as the two operate on quite different levels of abstraction.

To the extent, therefore, that Douglas Hofstadter concludes that the self is that is an illusion his is a wholly useless conclusion. As he acknowledges, "we" are doomed to "see" the world in terms of "selves"; an *a priori* sleight-of-hand, no matter how cleverly constructed, which tells us that we're wrong about that (and that we're not actually here at all!) does us no good at all.

Neurons, gluons and strange loops have their place - in many places this is a fascinating book, after all - but they won't give us any purchase on this debate.

Olly Buxton