Norretranders Don’t look at I: me is doing all the work. To hell with grammar: that’s the point about consciousness, at least according to Tor Nørretranders. I must admit my heart sank just slightly when I first saw the title of his book The User Illusion. So many people want to denounce the self as an illusion these days! I think it’s actually rather hard to deny that my self has some real substance – for most purposes selfhood seems to be an inoffensive, if not essential means of distinguishing between matters bearing on one human animal rather than all the others. Some of the sceptical arguments certainly have their appeal, but I generally feel that the best of them question the nature, rather than the existence, of the self.

Be that as it may, Nørretranders puts together a good case. It has a strong central theme, though it draws on arguments from several different sources. It’s a wide-ranging book, in fact: in places it reminded me of Roger Penrose’s tendency to go off on fascinating but slightly peripheral expositions. There’s even a picture of a Turing machine, very similar to Penrose’s, with the infinite tape heaped up in lines of boxes which disappear over the horizon (I worry slightly about the tangles that are liable to arise here – wouldn’t it be better to have the tape hanging down into a bottomless void?)

The main perspective is of human beings as information processors, which means tackling the vexed question of what information really is and how it relates to reality. Nørretranders describes a conference where the participants boldly set out to get, as they put it “it from bit”, a fantastically optimistic aspiration. If, as Frege found, we can’t reduce maths to logic, how likely is it that we shall be able to reduce the actual existence of a particular apple to information?

Claude Shannon. of course, gave us a watertight way of doing hard calculations about information, but only by adopting a restricted definition, which relates it to order and entropy but excludes the whole idea of meaning, essential to the everyday conception of information. Quantifying information in this wider sense is formidably difficult. In ordinary human discourse, a few words can convey a tremendous quantity of information: in fact, in the right context a single symbol can speak volumes. An exclamation mark requires only a handful of bits, but when Victor Hugo and his publisher exchanged telegrams which read merely “?” and “!” a great deal was conveyed in both directions about the progress of Hugo’s latest book.

To me, such examples show that there is something fundamentally wrong with the wider quantification project: Shannon, I suspect, probably did about all that can be done. If I have agreed on a code or convention with my contact, a single exclamation mark can mean, not just many different things, but absolutely anything whatever: does that mean it conveys an infinite amount of information? In fact, Victor Hugo could go on deducing further information from his publisher’s message for an indefinite period: it told him, for example, that his publisher was still alive at the moment of composing the message: that he was alive a minute before that point, half a minute before that point, and… but you get the idea. The example may seem silly, but the problem is real.

Nørretranders takes a more optimistic view. Perhaps the real measure of information is not the content, but the work that had to be done to get that content; to whittle down the range of possible communications to the one we’ve chosen. In fact, and this is an idea which recurs throughout the rest of the book, perhaps the important thing is how much information we had to discard in arriving at our message. Nørretranders introduces the concept of exformation – all the extra contextual stuff which doesn’t get included explicitly in the message, but which the recipient can infer or recognise without difficulty. This is strikingly like the Gricean implicatures which Sperber and Wilson have tried to quantify, but that other line of enquiry into the same territory does not get discussed here.

A second broad theme concerns the limitations of consciousness: we think our conscious minds are in control, but in fact there is plenty of evidence that the bit of us that does the talking and writing doesn’t really do much else. Nørretranders quotes a wide range of evidence, mostly well-known to those of us who follow these things: blindsight, split brains, subliminal influences, the tendency of patients to produce a confabulated rationale for behaviour they weren’t actually in control of, and Libet’s finding that the brain is geared up to act before we have consciously decided on action. Nørretranders quotes a finding by Zimmerman that the maximum information flow of conscious sensory perception is a mere 40 bits per second, while the eye alone is sending ten million bits of information to the brain over the same period. All those millions of bits are surely there for some reason, but not, it seems, to support conscious decision making.

Nørretranders proposes a model in which the “me”, the unconscious mind, is working away with all the reams of information provided by the senses: when it needs to communicate with another me, most of the information is discarded, and the tiny remnant transmitted between the two “I”s, or the two conscious minds, of the parties involved: only a tiny amount of bandwidth is necessary: but the recipient’s “me” is able to recover the copious exformation which goes with the message. He suggests that the I is really similar in some respects to the “user illusion” which makes personal computers viable. We don’t need, and couldn’t cope, with knowing all the details of how our computer does what it does, how streams of digital numbers are moved around between specific locations in storage or various registers: instead we need a simple analogy with pieces of paper, wastebaskets, tools, and so on. The analogy can be pretty loose, or even downright misleading in certain respects, so long as it broadly conveys what is going on and reduces the formidable complexity of the actual processes to something we can manage.

I’m not sure that analogy is very precise. Generating an illusion of purposeful personhood seems a much more demanding business than making a complex computer process look like a simple real-life one. Who is the illusion for? Is it the unconscious me that is being fooled into thinking that there’s a conscious I in charge? How do you delude someone who isn’t conscious? Moreover, it’s surely I that think I exist, not me, so the illusion must actually be for I’s benefit: but if I’m illusory, how can I think anything, let alone fall victim to the illusion ofmy own existence?

It’s certainly true that a great many mental processes unfold unconsciously, but it’s far from clear that these unconscious processes fit together to form a coherent unconscious me. Nørretranders may have done himself a disservice by quoting so many exotic examples: the split brain cases suggest the unconscious me must be confined to one hemisphere of the brain, but the other cases, such as the blindsight ones, seem to contradict that. Most surprising of all in this respect is the way Nørretranders invokes and apparently endorses Julian Jaynes’ remarkable theory of the bicameral mind. According to Jaynes, we were actually not conscious until some point in ancient history: the part of our mind which we now perceive as our consciousness was previously interpreted as the voice of God or gods. But Jaynes has his preconscious people dealing with language readily: in fact the Aeneid is an example of the products of a bicameral mind (most implausibly, if you ask me, but still). That seems to fit very uneasily with an unconscious me which has run up the illusion of consciousness specifically in order to cope with language and other low-bandwith forms of communication.

It seems more likely that a whole series of unconscious processes are operating more or less separately. It also seems to me that the “I” as discussed by Nørretranders is drawn much too narrowly – it really amounts to no more than the part of us that does the talking. But I have had the experience, in stressful situations, of listening to my speaking part yak mindlessly on, while with my conscious mind I only wanted to shut up. When I play tennis, I don’t think in words about where to hit the ball, or even very explicitly in other ways: but to say my tactics were unconscious would be going much too far (even for me). I suspect a more accurate model of how things work would be a complicated mess of unconscious processes feeding into a conscious level which is itself complex, with different levels which are clearly distinguishable, though still operating essentially as a unity.

That said, Nørretranders gives an insightful account, and covers an awful lot of ground. I don’t know what me thought, but I enjoyed it.


  1. 1. Eric says:

    Interesting stuff. I wonder if he sells Shannon short. What he calls ‘exinformation’ does carry information. For instance, the claim that ‘Johnny was murdered’ carries the information that Johnny is dead. It also carries information about what isn’t explicitly said. Whatever probabilities are distorted by that message signify a transmission of information. E.g., the likelihood that Johnny was shot is higher, given that he was murdered, than if we didn’t know he was murdered. Similarly for all the !’s and ?’s in messages transmitted between two parties.

    Finally, while I think Shannon information is not sufficient for semantic meaning, it is probably necessary. Too many people (except Dretske and a couple of others) have thrown out the baby with the bathwater. Dretske, OTOH, has tried to show what needs to be added to information to get semantic meaning.

  2. 2. peter.hankins says:

    Yes. I must read up on Dretske properly (not the first time I’ve said that).

  3. 3. Lakefall says:

    “How do you delude someone who isn’t conscious?”

    Oh, that’s very simple. I don’t know exactly what Tor Nørretrander is saying, but I’m pretty sure I have reached a general understanding how this thing called the consciousness works. I think I can even explain what generally happens in peoples’ minds when they fail to understand it. Their problems are they think about the consciousness atomically and they do not recognize a mirror when they see one. Let’s look at a moment in the life of “me”, the thing some people mistakenly call the subconscious.

    As I see it, the “me” system is quite a central nexus in the mind. It receives inputs from all over the place, from the senses, from memory, and from this mysterious thing called “I”. It probably has hard coded ways of understanding these inputs. It doesn’t need to understand what “I” is to know how to deal with it. “Me” is a general problem solver. It looks at its inputs and does its best to figure out what is going on. However, as we take the point of view of “me”, we should realize it isn’t self-aware by itself. “Me” does not know it is “me” and when it thinks about things it has no idea anything is being done. So, while we already know “me” isn’t conscious, let’s tease “me” with a problem: which one is conscious, “I” or “me”?

    “Me” looks around at all it’s inputs and immediately identifies the direction it needs to look at. It thinks the answer to this question lies in the direction of the input from which it regularly hears “I” telling it what “I” thinks about things. “Me” knows from both an underlying assumption and experience, that while “I” is human and makes mistakes, the information coming from “I” is generally well reasoned. It does not really need much thought to reach the conclusion that I is most likely to be the smart and the conscious one. In fact as far as “me” can tell, the very term consciousness refers to “I”, so the issue is solved before it began, right? But maybe something is being missed? What about “me”?

    Soon “me” receives information from “I”, saying “I” has considered the nature of consciousness and has reached the conclusion “I” is conscious, but this is not absolutely certain and also some additional thought about “me” is needed. “Me” takes out it’s notebook and writes down “I” is almost certain “I” is conscious, but has not given “me” enough thought. So, as “me” needs to be looked into, “me” wonders where “me” is and what has it been doing. It again takes a look at it’s inputs: there is “I”, there is the external world, and there are the memories, but where is “me”? “Me” cannot see “me” in any direction. It’s almost as if it is sitting on top of itself. 😉

    Oh well.. here comes information from “I” again. “Me” dutifully writes down how “I” has been looking at the inputs and has seen the external world, the memories and even itself, but has not found “me”. “Me” is left with the problem of figuring out why “me” is so elusive to “I” and also how can “I” be so fantastic it can look at itself in the eye (pun not intended) without even having a mirror. In fact, when you really think about it, isn’t it.. impossible?

  4. 4. reason says:

    Hi there! This is my 1st comment here so I just wanted to give a quick shout out and say I genuinely enjoy reading your articles.

    Can you recommend any other blogs/websites/forums that deal with the same
    topics? Thanks a lot!

Leave a Reply