Friday, June 11, 2004

Dreams and Sensations

Jonathan Ichikawa has a very interesting paper, Fire and the Cogito: Dreams and Imaginings, where he argues convincingly for the counter-intuitive thesis that we don't really have sensations or form beliefs when dreaming - we only imagine that we do. I could probably go along with him on the 'beliefs' issue (at least, I wouldn't be surprised if beliefs in dreams were equivalent to our beliefs about fiction, however it is that they get explained), but I'm more doubtful about the sensations. I should begin by admitting that I don't really know anything much about this field, so all I'm going to do is highlight a few bits of Jonathan's essay which conflict with my own (possibly unreliable) intuitions.

Consider visual imagery. I can close my eyes and imagine seeing, for instance, a tree. I do not have the visual sensations(R) associated with seeing a tree, but I have the associated sensations(I). Laboratory experiments have demonstrated that subjects sometimes mistake one for the other.

Isn't sensation a purely subjective phenomenon? At least, that's how I'm inclined to think of it. If you believe that you are subjectively experiencing a sensation, then you are subjectively experiencing that sensation. I think it is possible to have visual sensations when imagining. I can (sometimes, not always) imagine a colour so well that I actually believe I'm seeing it - which implies (doesn't it?) that I was subjectively experiencing the visual sensation.

So I don't think there is any subjective difference between real sensations(R) and imagined sensations(I) that seem real. Either we experience a sensation, or we don't. (Though I'm not entirely sure what to do about weak visual imagery, e.g. if you try to imagine a tree, but don't actually experience seeing it. Perhaps the sensation you have is not an imagined visual sensation, but a sensation of imagining. I'm not sure how plausible this is - it may be a fatal flaw in my objection here.) You can, of course, differentiate based on the source of the experience, as is done in the 'laboratory experiments' Ichikawa cites. But that doesn't mean the experiences themselves are qualitatively different. People sometimes mistake sensations R and I, because all the subjects are really doing is guessing at the cause of their sensations (period), and it's no surprise that we're fallible at that!

(Update: See the first two comments of this FBC post for discussion of a similar issue. I guess I would have to agree that most visual imagery does not involve genuine sensations, but I'm not convinced that this is always the case. But, to be honest, I'm getting a bit confused about it all!)

To bolster the plausibility of his argument, Ichikawa appeals to the obvious invalidity of such inferences as:
I dream that p.
Therefore, p.

I dream that I have mental state m.
Therefore, I have mental state m.
[e.g. let m be severe depression, or superintelligence]

But these are invalid because they involve objective claims, which depend on matters external to our subjective conscious experiences. This makes them poor analogies for the inference from 'I dream that I have a sensation of X' to 'I have a sensation of X', which still strikes me as eminently plausible - perhaps even tautological.
What about actual conscious states? Consider volitions. Sosa has suggested an ethics-based reductio against that inference:

If I merely dream that I form an intention to succumb to temptation, I haven't done anything wrong. If I actually formed that intention, I would have done something wrong. Therefore, the inference, ‘I dreamed I formed intention X, therefore I formed intention X,’ is invalid.

I have some doubts about the second premise. Just as a general principle, I would have thought that we are not to be held morally accountable for our dreams. So even if S forms an immoral intention while dreaming, that is not to say that S himself has done anything morally blameworthy.

At least, that seems plausible if volitions are understood here has being nothing more than a sort of subjective experience. If, however, they are supposed to involve some external/objective notions also (e.g. the tendency to influence our actual real-life behaviour) - as is probably the case - then I'd agree that we don't form actual intentions in dreams. However, this concession says nothing about purely subjective experiences (such as I understand sensations to be), so I don't think this example helps Ichikawa's case after all. More relevant is the following attempt:
Let e be the experience as if a lion were chasing me. Then if I
have e in a dream, I would be epistemically required to form the
belief that a lion is chasing me, and prudentially required to form
the intention to run away. Failure to form that belief and intention
would constitute irrationality. But I am not irrational in failing to
form the belief and intention, therefore the inference is invalid.

I'm not entirely convinced by either of those premises. As before, I just don't think we are responsible (epistemically nor morally) for anything which happens in our dreams. Furthermore, I think that it would quite often be accurate to describe my dreaming behaviour as "irrational". Jonathan pre-empts the first response:
Dreaming is not like being forcibly drugged with respect to rationality and epistemic blameworthiness. Since rationality is important to me, it is appropriate for me to take measures to ensure that I do not find myself in circumstances in which I am likely to be irrational. I fail epistemically and am also blameworthy if I voluntarily ingest drugs, knowing that they will cause me to acquire false beliefs. Not so for dreaming – my interest in being a rational agent does not require me to take steps to avoid dreaming.

I don't think this counter works, however. His last sentence is true, not because we are rational while dreaming, but rather, because it doesn't matter that we are irrational while dreaming. This limited irrationality does not adversely affect our waking life, so we have no duty to avoid it. Ichikawa seems to be suggesting that if the orthodox model of dreaming were true, we would have a rational duty to avoid dreaming. This strikes me as highly implausible, even downright bizarre. If anyone really had a desire to avoid all irrational thoughts, even when sleeping, I'd be tempted to accuse them of irrationality!

The phenomenology of Lucid dreaming seems especially problematic for Ichikawa's lack-of-agency hypothesis (quoting Walton):
Dreams are spontaneous, undeliberate imaginings which the imaginer not only does not but cannot direct (consciously). This helps to explain why dreaming is often such a powerful experience, why dreams tend to be more compelling, more “realistic” than daydreams in which the imaginer either directs the course of his imaginings or deliberately refrains from interfering.

This should imply that when we have more control, as we do in lucid dreams, our confidence in the 'realism' of our sensations should diminish. But the very opposite is the case. Lucid dreams seem even more real than normal dreams.

Lastly, Ichikawa suggests that this imagination model casts doubt on the certainty of Descartes' "I think, therefore I am".
for anyone who took himself to be asserting the premise could ‘step up’ a hypothetical level and ask, how do I know I’m not just thinking(I) that?
Imaginings, unlike beliefs, are not the kinds of things that can be certain.

I haven't properly studied Descartes, so it could be that my understanding of him is mistaken. But I thought the point was that if I am having any mental experience whatsoever (i.e. "think" understood broadly, so it would include imaginings as well), then a necessary pre-condition for this is that I must exist. (Otherwise, who is doing the imagining?) So I don't understand how the possibility of unwitting imaginings are supposed to cause any problems here.

Overall though, it really is a great paper. I enjoyed reading it immensely. I actually think that Jonathan's collective arguments are a lot stronger than the few confused objections I've raised here, but I just wanted to take this chance to try to sort through them. (With any luck, he might leave a comment and let me know how to clear up those confusions! I guess he'd disagree that sensation is a purely subjective phenomenon? Well, second-guessing won't do any good, I'll just wait and see...)

Update: Jonathan has a new blog post summarizing his view, here.


  1. [Copied from old comments thread]

    Jonathan wrote:Thanks very much for the comments, Richard. A few points:

    Isn't sensation a purely subjective phenomenon?I have some sympathy with this view -- I think it takes a little reflection to appreciate why I reject it. Remember the blind woman discussed in my paper? Intuitively, it seems like (1) she was blind, (2) blind people cannot have visual sensations, and (3) she took herself to have visual sensations. If these are all correct, then one can falsely take oneself to have visual sensations.

    But that doesn't mean the experiences themselves are qualitatively different. People sometimes mistake sensations R and I, because all the subjects are really doing is guessing at the cause of their sensations (period), and it's no surprise that we're fallible at that!There's more to it than just causes, but it's not easy to explain what exactly it amounts to. It seems likely to me that people who experience hallucinations actually experience sensations, even though they aren't caused in the normal way.

    Wittgenstein had a lot to say about imagery and sensation, and I've only just begun to touch on it. Interesting stuff.

    Just as a general principle, I would have thought that we are not to be held morally accountable for our dreams. So even if S forms an immoral intention while dreaming, that is not to say that S himself has done anything morally blameworthy.The argument runs like this:

    (1) If I intend to commit murder, then I have done something morally wrong.
    (2) In my dream, I intend to commit murder.
    (3) I have not done something morally wrong.

    From (1) and (3), we get (4): I have not intended to commit murder.
    And from (4) and (2), we get: There is no valid inference from (2) to "I intend to commit murder.

    It sounds like you want to deny (1) -- you think that if I really do form the intention to murder, but I'm dreaming at the time, then I'm not morally blameworthy. We agree, of course, that if I have a dream in which I form the intention to murder, I am not blameworthy. I'm trying to explain this fact. Your account owes an explanation of in what morally significant way dreams are different from reality here. I offer one: we don't really form intentions when we dream.

    The same response applies to the lion belief case.

    There's room for a lot more clarity on the cogito issue - that's probably its own paper, and maybe it'll get written this summer. The short response is, Descartes was interested in evaluating his *beliefs*. In order to do that, he must know what they are. You're right that there is a valid inference from "I have some mental state or other" to "I exist", but I contend that Descartes cannot achieve certainty about the latter without certainty that he has beliefs. It's very sketchy, I know.


    I wrote:Thanks for the clarifications. The more I think about it, the more I'm attracted to your (/Sosa's) view.

    Just a couple of quick counter-responses:
    i) I'm not sure why you'd want to deny that blind ppl can have visual sensations. They certainly don't have visual perceptions, but I don't see why they couldn't have hallucinations (for example). After all, we dream with our eyes closed, right? ;)

    ii) You're right that I want to deny premise (1) of the 'blameworthy' argument. I'd suggest that dreams are relevantly different because a) we don't have the same sort of rational control as we do in our waking lives AND b) Morality is essentially a public/social matter, and dreams are inherently private - any intentions we form in dreams do not carry over into our waking lives, which explains why they are not morally relevant.

  2. Just saw this in your "best of" list.

    I'll try to be delicate. But isn't the nature of sexual dreams pretty compelling evidence for sensation in dreams? Would the body respond the way it does if the sense subsystems of the brain not engaged?

  3. I'm not sure. I think Jonathan could argue that "imagination-correlates" can still account for this, especially since we know that imagination tends to activate very similar brain regions to perception.

    To support this, note that if you imagine a naked woman, that can lead to a similar physiological response as if you were seeing her -- even though you're not really having any visual sensations at all.

  4. Doesn't that avoid the issue of whether there are sensations in our imaginations? It seems to me you're merely providing a compelling reason for believing there are.

  5. Intuitively it seems that we can draw a distinction between sensations and mere visual imagery. If you imagine a woman, you aren't really having a woman-like visual experience. You just pretend or imagine that you are. (Though I do think it's possible for imagination to yield real sensations, as I explain in the main post.)

    But perhaps you were suggesting that the physiological response would in turn cause (rather than result from) real sensations. That sounds right, but then it occurs outside of our imaginations, so it isn't surprising.

    The question, then, is whether such responses would cause real sensations while dreaming. I'm not sure the prima-facie case here is any stronger than the usual one that activation of visual areas while dreaming implies visual sensations. It seems at least possible that both might simply be imagination-correlates.

    Alternatively, perhaps you were arguing that the physiological responses must have resulted from genuine sensations. But once we see that imagination alone can cause physiological responses, as in the naked lady example, I think this argument loses much of its force.

    (But perhaps I am missing your point. If so, you may need to explain it in more detail - delicacy be damned.)


Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)