Picture: William James. A while ago I discussed the way William James, who originated the phrase “stream of consciousness” nevertheless went on to embrace a radical scepticism about the very idea of consciousness. In an interesting recent paper William Lyons reviews the roots and significance of this apparent apostasy; I was interested in particular in an argument which he thinks helped form James’ view.

James was happy with thoughts and experiences, but denied the existence of any special reflexive sense of self. Hume famously said that when he turned his attention inwards on his own mind, he found only a bundle of perceptions; in the same spirit James says he detects nothing but ‘some bodily fact, an impression coming from my brow, or head, or throat, or nose’. In fact, he becomes convinced that when you boil it right down the breath is probably the core of what has been built up into a strange metaphysical/spiritual entity; I expect he had in mind the ancient Greek pneuma, both breath and spirit.  Maybe, he says, we could better talk of ‘sciousness’ to refer to our awareness; consciousness, the special self-awareness, is out. James was impressed by an argument for the impossibility of introspective self-awareness set out by Comte.   It points out that when you observe your own thoughts there is an awkward circularity involved.  Observation involves conscious attention, but in this case conscious attention is also the target. Necessarily then, there must be some splitting, some withdrawal from the target of observation; but then it ceases to be the conscious awareness we were trying to introspect. It’s like trying to tread on your own shadow.  You end up, at best, observing not your real immediate self, but a kind of fake or ersatz thing, an idea of yourself which you have generated.

I’m not absolutely sure this argument is watertight. Comte supposes that in order to observe our own thoughts, we have to stop observing whatever we were contemplating before.  If that’s so, is it a disaster? All it really means is that when we observe ourselves, we’ll find that we are currently observing ourselves. That’s circular alright, but I’m not sure it is necessarily disastrous. Moreover, can’t we think about more than one thing at once? Comte suggests that self-observation requires a kind of separation in the self, but aren’t we complex anyway, with several different layers and threads of thought often co-existing? Can’t it be that when we introspect we can observe the thoughts that were running along beforehand accompanied by a new meta level on which we’re watching the original thought and also watching ourselves watching?  It sort of seems like that when I introspect – more like that than like bundles or gusts of breath in an empty head.

Putting that aside, if we accept the argument there is another way out, apparently proposed by John Stuart Mill and adopted by James, namely that the required separation takes place over time. So instead of trying to observe my own mind at this very instant, what I’m really doing is contemplating the state it was in a few moments ago; I am in fact remembering rather than perceiving. James concluded that introspection is really retrospection. Far from being a special case of infallible perception, it is as flawed and prone to error as any other memory.

Well, yes; but then all our perceptions are subject a small delay, aren’t they? It may only take a fraction of a second for light to reach our eyes and work its way to the brain, but it’s not instantaneous; and before the perception becomes conscious at least a few more milliseconds of processing will surely intervene. It’s arguable in fact that all perception is retrospection; and we could still argue that self-perception is privileged in at least a weak sense because it all takes place within the brain, where the most serious sources of error and misinterpretation can hardly apply and the delays are presumably at their shortest.

But in fact I think the whole argument goes wrong from the beginning in assuming that when we talk about conscious self-awareness we’re talking about a redirection of the same stream of attention that we habitually direct towards the outside world. I think claims about the special qualities of self-perception actually rest on a view that it is not a product of perception, but inherent in it.

If we’re using a light to explore a dark cellar, we have to turn the beam on each object in order to perceive it. But then what about seeing the light?  Do we have to turn the beam of light on to itself – and if we do somehow manage that will we really be seeing the light as it is, or some strange reflected optical phenomenon? Obviously not: light is visible already without more light being played on it: and we are aware of our own perception without having to perceive it.

Of course we can turn our eyes inwards, and all the confusingly complex business of retrospection and perceiving imagined self-images are all perfectly possible, and part of the complicated overall picture. But the essence of it is that acts of perception inherently indicate the perceiving self as well as the object of perception. Perception of the self seems to be a special case, invulnerable to error, because we can be mistaken about the objects of perception but not about the fact that we are perceiving (as Descartes might have said).

I dare say, however, that some reasonable people would be content with sciousness, or rather would take simple awareness to be consciousness, without being unduly concerned with the self-reflecting variety. It seems as if James, by contrast,  would agree with higher-order theorists about the nature of consciousness, but differ from them in considering it impossible.


  1. 1. Paul Bello says:

    Cool post, Peter.
    I think in some sense, the argument for and against introspection fail in various ways because few folks are really (1) clear, and (2) agreeable on a standard definition of the term. Talk about introspection being intimitely tied up with the notion of “the specious present” get confused because no one can adequately define what the specious present actually is. One thing that does seem clear is that we’re able to do belief revision, for whatever that’s worth…and moreover, we’re able to do so consciously, meaning that we have to have some representation of having a false or low-confidence belief in the past, which is of course, some form of reflexive consciousness. I’ve started to develop a very crude mechanistic account of introspection and its relation to third-person mental state attribution by way of mental simulation that isn’t explicitly self-reflective in the way normally thought about in sentential models of cognition. Of course, I’d be happy to share ideas with any of you off-line, if interested.

  2. 2. Vicente says:

    Really cool indeed !

    The concern of Comte about the need to split the focus of attention in order to carry out introspection, and the way out proposed by J.S. Mill based on “retrospection” of short term memory salient details, makes me think that most analogies and metaphores usually used to try to depict consciousness are just completely wrong and basically misleading… (reminds me of the homunculus infinite regression)

    In the last page James says that consciousness is just “pure experience” and interactions between experience, and anticipating the natural question to be made he acknowledges:

    pure experience is something we know not at all

    As it usually happens we ended up in square 1, after that what else could be said.

    I found very interesting the part on consciousness defined as a very fundamental level of knowledge, being conscious states vehicles of knowledge. What does he really mean by level of knowledge?

    Does anybody feel that James was proposing some kind of pansychism?

  3. 3. Peter says:

    Thanks, chaps.

    That’s a tantalising description of your ‘crude’ account Paul – I’d be interested to hear how it works.

  4. 4. Paul Bello says:

    Hi Peter,
    You can find a short 6-page paper here:

    I’m not settled on the details of what’s in here, but I like the overall direction. Of course, any comments or suggestions would be quite welcome. This is an interesting yet confusing approach for many on the first read. It is Meinongian in nature, and doesn’t advert to explicit relations between believers and propositional content in the way that traditional epistemic logics do.

  5. 5. Vicente says:

    Paul, very interesting paper, I have a few concerns, maybe because I don’t completely understand the approach, I am not sure that when we guess others minds contents (mindreading), the concept of simulation can really be applied (I don’t know), anyway, I just want to make the remark about the difference between to have an experience and to report about that experience, completely different “processes”. To me, this fact makes the whole approach quite feeble.

  6. 6. Paul Bello says:

    That’s a good question. One thing that’s not made explicit in the paper (and should be), is that this is totally non-verbal in nature. The model only purports to explain nonverbal instances of belief attribution. The semantics of belief reports are quite another matter, and it’s an open question how the latter interface with the former.

    Actually, the belief report issue is exactly what I’m starting to work on now. That’s a very large and very complex literature.

  7. 7. Vicente says:

    Paul, thank you for the clarification, now I understand it better. Good luck, it looks you are facing a hard job, but any results you get might shed light on this difficult issue.

  8. 8. John says:

    James does not make sense without considering his idea of ‘time’:

    “In short, the practically cognized present is no knife-edge, but a saddle-back, with a certain breadth of its own on which we sit perched, and from which we look in two directions into time. The unit of composition of our perception of time is a duration, with a bow and a stern, as it were — a rearward — and a forward-looking end.”.

    He describes durations as arrangements of events out there, amongst the objects within perceptual experience.

    (See Radical Empiricism and New Empiricism).

    I am sorry to be churlish Peter, but your blog seems to be devoid of any understanding of the problem of time and (the separate problem) of change. Whether it is the timing of mental events or the relationship between the static states of computers and the fluid, extended present of our experience your articles adamantly refuse to draw attention to the possibility that time is the central factor and our ignorance of the nature of time the cause of the apparent mystery.

  9. 9. Peter says:

    Thanks John, a useful point.

    On the wider deficiency you mention in respect of time: it’s not really an adamant refusal on my part – I dare say ignorance, stupidity and failure to pay attention are nearer the mark. Happy to take up any suggestions for better coverage of time (or I’d be pleased to feature a piece by you, if that doesn’t seem a cheeky suggestion).

  10. 10. Vicente says:

    John, even more, mental events have no time, unless you have a mapping to neural correlates that can have an associated physical time, which is not the case in general. Think of the paradoxical time of dreams. James was a good philosopher but a bad physicist in what to time concerns.

  11. 11. John says:

    Peter, thank you for your offer. I am off on holiday in a couple of days but would be interested to pursue this in February.

    Vicente, your example is demonstrative of some of the difficulties with the idea of time. As McTaggart spotted (but failed to explain), we can all too easily start speculating on the “rate of time” (seconds per second) if time is equated with change.

  12. 12. Vicente says:

    John, time cannot be “equated” with change. Change rate can be equated or measured against a change rate standard or pattern, e.g. 1 year is 1 turn around the sun… there you have a clock. This McTaggart seconds per second looks like Marx Bros joke.

    The problem is that the speed of light is finite and constant, and that creates a big complexity to fix and relate events (cause-effect) in a time axis (special relativity) to make it worse, time is intrinsically associated to the space fabric, spacetime, and then time measurement depends on the space metric… hell… general relativity….

    Then you have time direction related to the statistical evolution of thermodynamic magnitudes, mainly entropy… etc etc…

    The point is that phenomenological experience seems to have a psychological time, very much related to the perception of the physical world, but this psychological time seems to be quite decoupled from the physical time, it looks more like the time parameter in a computer simulation, when they say real time environments, something is not true.

    By introspection you can check that mind time is funny, incredibly elastic… it is not a physical time in general. In my opinion, we cannot apply the concept of time to our inner world, just notice that it changes, but there is no real clock.

    I think James needed to use some time concept to support his conscious flow model (of course, no time no flow by definition), but he didn’t handle the concept very well.

  13. 13. John says:

    Vicente on mental time: “it looks more like the time parameter in a computer simulation”

    Yes, I would agree that the rate at which events change in our experience can be decoupled from the rate at which a standard clock ticks in the physical world in the same way as a clock in a DVD on “fast forward” is decoupled. Change is not time.

    The fact that change is not dimensional time can be illustrated by considering an atomic clock. Atomic clock ticks are correlated with distances in dimensional time but the simple occurrence of the change that is a “tick” event is not a measure of dimensional time. For example, I could gather together a mass of radioactive material and select decays at 0.1ns, 1ns, 100microsec etc and display these as “clock ticks” – these would not correlate linearly with changes in dimensional time.

    But dimensional time is also not the whole of “time” because dimensional time could exist without change (cf: block universe). So what selects the stream of events that is time passing?

    There is a fairly cogent discussion of dimensional time at Wikibook on Special Relativity.

Leave a Reply