Posts tagged ‘self-organised criticality’

Picture: Etch-a-Sketch. An interesting New Scientist piece recently reviewed research suggesting that chaos has an important part in the way the brain functions. More specifically, the suggestion is that the brain operates ‘on the edge of chaos’, in self-organised criticality;  sometimes it runs in ways which are predictable at a macro level, more or less like a conventional machine; but at times it also goes into chaotic states. The behaviour of the system in these states is still fully deterministic in a wholly traditional, classical way, but depends so exquisitely on the fine detail of the starting state that the behaviour of the system is in practice unpredictable. The analogy offered here is a growing pile of sand; you can’t tell exactly when it will suddenly go through a state shift – collapse – although over a long period the number of large and small collapses is amenable to statistical treatment (actually, I have to say I’ve never noticed piles of sand behaving in this interesting way, but that just shows what a poor observer I am).

The suggestion is that the occasional ‘avalanches’ of neuronal firing in the brain are useful, allowing the brain to enter new states more rapidly than it could otherwise do. Being on the edge of chaos allows “maximum transmission with minimum risk of descending into chaos”. The arrival of a neuronal avalanche is related to the sudden popping-up of an idea in the mind, or perhaps the unexpected recurrence of a random memory. There is also evidence that the duration of phase-shifts is related to IQ scores – perhaps in this case because the longer shift allows the recruitment of more neurons. The recruitment of additional neurons is presumed in such cases to be a good thing (I feel there must be some caveats about that), but there are also suggestions that excess time spent in phase-shifts could be a cause of schizophrenia (someone should set out a list somewhere of all the things that at one time or another have been put forward as causes of schizophrenia); while not enough phase-shifting in parts of the brain to do with social behaviour might have something to do with autism.

One claim not made in the article, but one which could well be made, is that all this might account for the sensation of free will. If the brain occasionally morphs through chaos into a new state, might it not be that the conclusions which emerge would seem to have come out of nowhere? We might be led to assume that these thoughts were freely generated, distinct from the normal predictable pattern. I think the temptation would be to frame such a theory as an explanation of the  illusion of free will: why we feel as if some of our decisions are free even though, in the final analysis, determinism rules. But I can also imagine that a compatibilist might claim that chaotic phase shifts really were freedom. A free act is one which is not predictable, such a person might argue; however, we don’t mean unpredictable in practice – none of us is currently able to look at a brain and predict the decisions it will make in any given circumstances. We mean predictable in principle; predictable if we had all the data plus unlimited time and computing power. Now are chaotic changes predictable in principle or not? They occur within normal physical rules, so in the ultimate sense they are clearly deterministic. But the difficulties are so great that to say that they’re only unpredictable in practice seems to stretch ‘practice’ a long way – we might easily need perfection of measurement to a degree which is never going to be obtainable under any imaginable real circumstances. Couldn’t we rather say, then, that we’re dealing with a third kind of unpredictability, neither quite unpredictability in mere practice nor quite unpredictability in principle, and take the view that decisions subject to this level of unpredictability deserve to be called free? I think we could, but ultimately I’m disinclined to do so because in the final analysis that feels more like inventing a new concept of freedom than justifying the existing one.

There’s another issue here that affects a number of the speculations in the article. We must beware of assuming too easily that features of the underlying process necessarily correspond directly with phenomenal features of experience. So, for example, it’s assumed that when the brain goes quickly into a new state in terms of its neuronal firing, that would be like a new thought popping up suddenly in our conscious minds, an idea which seemed to have come out of nowhere. It ain’t necessarily so (though it would be an interesting question to test). The fact that the brain uses chaos to achieve its results does not mean that the same chaos is directly experienced in our thoughts, any more than I experience say, that old 40Hz buzz starting up in my right parietal, or whatever. At the moment (not having read the actual research, of course) it seems equally likely that phase shifts are wholly outside conscious experience, perhaps, for example, being required in order to allow subordinate systems to catch up rapidly with a separate conscious process which they don’t directly influence. Or perhaps they’re just the vigorous shaking which clears our mental etch-a-sketch, correlated with but not constitutive of, the sophisticated complication of our conscious doodlings.