Dream a Little Dream

One of the other issues that came up at Miguel’s cogsci talk was that of the empirical testability of the HOT theory. Miguel suggested that we might have the following argument against HOT. Experimental evidence suggests very strongly that the dorsal lateral pre-frontal cortex is likely to be the home of HOTs. David has said several times that if we did not find activity in the DLPFC when we had evidence that there were conscious mental states about this would be very bad for the HOT theory. So t if we think that we have conscious mental states in our dreams and we accept the evidence that shows that the DLPFC is deactivated during REM sleep this would seem to count as evidence against the HOT theory. David seemed to think that there were basically two plausible responses to this argument. One copuld deny that there are conscious mental states during dreaming or one could argue that the HOTs have a summer home that we haven’t found yet. A lot of the discussion centered on whether or not we had any evidence that dreams are conscious in the way we think they are. David argued that we didn’t Miguel that we did.

David’s argument seemed to me to be the following. The evidence we have that dreams are conscious are the reports that people make when they are awake and remembering the dream. But it is equally consistent with this that the dreams were all unconscious and only seem to be conscious when we reflect on them in the morning. Miguel seemed to think that it was obvious that dreams were conscious. I suggested that perhaps the kind of work that Eric does on dreams suggests that our naive views about dreams are wrong. Pete suggested that we had good experimental evidence that dreams were conscious from teh kind of studies where subjects are given instructions of the sort that if they see a flashing object in the dream they should clap five times. During the discussion the phenomenon of lucid dreaming came up and David reported that in lucid dreaming the DLPFC was active and so lucid dreams count as conscious mental states.During REM sleep subjects then can be seen to make clapping motions. But is it clear that this counts as a report in the relevant sense? This activity could be the result of unconscious dreams just as well as the result of conscious dreams. In David’s terminology we can ask whether the clapping is an expression of their mental states or whether it is a report. If it truly counts as a report and there is no activity in the DLPFC then David’s view would be in trouble.

This got me to thinking; how could we devise an actual empirical test of these kinds of issues? Hakwan suggested an interesting conceptual approach earlier which led me to think about binocular rivalry. If you could just have subjects in a scanner looking at  stimuli that are known to induce binocular rivalry without having the subjects do any kind of reporting we could then look at the DLPFC and see if the activity there reliably correlates with the conscious percept. A quick search on this led me to this article which seems to get results that line up with HOT theory very nicely, though with scalp EEG and with a button push which is a confound…

The New New Dualism

Yesterday I attended Miguel Angel Sebastian’s cogsci talk entitled “The Subjective Character of Experiencre: Against HOR and SOR Theories” which was very interesting. Miguel was primarily trying to show that higher-order and same-order representationist theories of consciousness cannot account for the subjective character of an experience by which he means the thing that accounts for the experience being for the subject. His main complaint seemed to be that in order to account for this we need some notion of the self and so he suggested that we need a model where we have representations of teh self interacting with representations of objects and we thus end up with a representation of the form “x for-me”. There were several interesting themes of the discussion and if I have time I will probably come back to some of them but I thought I’d start with this one.

In response to the mis-match problem David has settled on the following view. The phenomenology goes with the HOT. The sensory qualities of the first-order state play no role –other than that of concept acquisition– in determining the phenomenal character of a conscious experience. So in the case of Dental Fear the subject has a first-order state with vibration sensory qualities and a HOT that they are in pain so their conscious phenomenology is like having pain for them. The first-order sensory qualities play a perceptual role in the mental economy of the subject so having them is important but they don’t play a role as far as consciousness is concerned. In fact even if there is no first-order state at all (as may perhaps be the case in Anton’s syndrome) the phenomenology goes with the HOT. Now in the cases where there is no first-order state one still counts as being in a conscious state. The mental state that is conscious is just the one that the HOT represents oneself as being in and so in this case the conscious mental state is a notional state, which is to say that it doesn’t exist. It follows from this that there are conscious mental states that have no neural correlates. We thus end up with a dualism about consciousness of a new variety. There are some conscious mental states that exist physically in the brain and there are other conscious mental states that exist only notionally as the content of a HOT.

What should our reaction to this be? When this first became clear at David’s Mind and Language seminar it prompt Steve Stitch to shout ‘he’s worse than a dualist!’ Miguel seemed to think that at the very least this is a cost of the theory and that if you can have a theory that explains all the data without it that is preferable. David refused to say that this was even a cost for the theory, in fact he seemed to suggest that it wasn’t even counter-intuitive. His reasons seemed to be as follows. I can have a thought about things which are not present and those notional objects can have properties. So, if I think about a squirrel I might think of it as brown, and bushy even if there is no squirrel around yet the squirrel has properties; it is brown and bushy. Thus it is simply a fact about intentional states like thoughts that their contents can be notional and that those notional objects can be said to have properties. If that is right then there is nothing fundamentally mysterious about notional mental states having properties. The second step in his defense seemed to involve an appeal to hallucinations. We hallucinate regularly enough for it to be a common-place of folk psychology. Why doesn’t it make sense to say that we can hallucinate mental states? On this line the notional state is just like my hallucination of a pink elephant: it seems like it is there from my point of view but it isn’t really there. This isn’t mysterious since that just simply means that I represent myself as being in a state that I am not in. Now given various theoretical assumptions this will indeed turn out not to be counter-intuitive and since those who do find it counter-intuitive will do so because of different theoretical assumptions I suppose I can see why David thinks that this is not a cost to the theory.

But suppose that one had different theoretical assumptions?  Suppose that one wanted to avoid this kind of existence dualism and so endorsed some kind of principle like this: For every conscious mental state there is a corresponding brain state. But suppose one also wanted to remain a higher-order theorist…what are the options? The most obvious option is to identify the phenomenally conscious state with the HOT. The HOT is not introspectively conscious –for that it would need to have a third order state targeting it– but it is phenomenally conscious. It is the state in virtue of which there is something that it is like for the subject and so it seems natural to identify the property of phenomenal conscious with having the HOT. Ned Block has argued that if one does this then one has falsified the higher-order theory. Why? The transitivity principle says that a conscious mental state is one which I am conscious of myself as being in but on the previous analysis we have a phenomenally conscious mental state (the HOT itself) of which we are not conscious of ourselves as being in (there is not third-order HOT) thus adopting this view falsifies the transitivity principle. But this may be too quick. This way of formulating the transitivity principle leads us to the view that the HOT transfers or confers the property of being conscious to the first-order state but as we have seen what the transitivity principle really says is that a conscious mental state consists in my being conscious of myself as being in some first-order state. That is, the transitivity principle is a hypothesis about the nature of conscious mental states. It is a mis-reading of the transitivity principle that takes it to postulate consciousness resulting in a relation between the first-order state and the higher-order state. That this is the dominant way of interpreting the transitivity principle is not in doubt; it most certainly is. However, it is misleading and cause way too many problems. I think higher-order theorists need to be more explicit about this mis-reading of the transitivity principle.

To me the second is the best option. However, lots of people seem to think that of one adopts a same-order theory one can avoid these kinds of issues. Since one takes the conscious mental state to be a complex of a first-order content and a second-order content that represents the first-order content we don’t have to worry about notional states. Bit it is far from obvious that this theory has any advantages over the HOT theory. First it is unclear why the higher-order content cannot occur without the first-order content. This seems like an empirical issue that can’t be settled by definitional fiat (I guess I think Anton’s syndrome might be a problem here). Second, even if it turns out that you can’t have one with out the other it is still not clear why there cannot be a content mis-match. Why can’t a red first-order state be coupled with a higher-order content that represents the first as green?

Does the Zombie Argument Rest on a Category Mistake?

re-reading Ryle’s “Descartes’ Myth” I was struck by the following passage

…the Dogma of the Ghost in the Machine does just this. It maintains that there exist both bodies and minds; that there occur physical process and mental process; that there are mechanical causes of coporeal movement and mental causes of coporeal movement. I shall argue that these and other analogous conjunctions are absurd…the phrase ‘there occur mental process’ does not mean the same sort of thing as ‘there occur physical process,’ and, therefore, that it makes no sense to conjoin or disjoin the two. (this is from page 37 in the Chalmers anthology)

I have always been sympathetic to the category mistake move and have viewed it as a precursor to the claim that it is simply question begging to treat mental terms as synonymous for ‘non-physical’. I also think that a lot of my complaints about the intelligibility of substance dualism originate in Ryle’s discussion of the origin of the category mistake.

Re-reading this today I started thinking that maybe one could use this kind of claim to cause problems for the Zombie argument. The first premise of the zombie argument employs the conjunction (P & ~Q) where P are all of the physical facts and process and Q is some qualitative fact like that I feel pain. If it is really logically  illegitimate to conjoin these terms then the zombie argument cannot even get off the ground. So what is the response that the dualist will make here? It seems to me that all of the examples of category mistakes involve concepts that have fairly straightforward conceptual entailment relations between them. So, a pair of gloves just is a left glove and a right glove and we can tell this just by analyzing the concept of PAIR OF GLOVES. The same can be said for teh University, and the battalion. But if course it is not obvious, to say the least, that the same is true for PAIN or SEEING BLUE. To many, myself included, it seems as though there are no conceptual entailment relations between my “pure” phenomenal concept of pain and physical processes (for me the ‘seems as though’ part is especially important).  But maybe it is at just this point that I myself, as well as the dualist, commit the category mistake!

Whoa…I’ll have to come back to that because now I’m off to Miguel’s CogSci talk

Outline of the Case for Agnosticism

Via Leiter’s blog I found out about this Slate article on Agnosticism. I guess I agree with Leiter and a lot of the commenters that the piece is overly polemical and doesn’t address the arguments of either side. Being a card carrying agnostic myself I thought I might chime in.

I found myself in general agreement with the rhetorical position of the piece. People often associate agnosticism with either intellectual laziness (we haven’t thought about the issue long enough to have a view) or a certain kind of intellectual cowardice (we don’t have the guts to say what we really believe; i.e. we are secretly atheists/theists and are just too cowardly to admit it). Both of the charges are misguided. Agnosticism is simply the honest recognition that we do not have decisive reasons for thinking that there is, or isn’t, an all-knowing, all-powerful, morally perfect being. I don’t, however, think the case for agnosticism consists in simple demanding that someone explain why there is something as opposed to nothing, though that is part of the case.

In short the situation seems to me to be this: There is equally compelling evidence on both sides of the issue. Given that we like evidentialsim in some form and that we agree that it is an illegitimate move to count belief in God as properly basic (if we do we must count any belief as properly basic in some noetic structure) it really seems to me that a rational disinterested person should conclude that there is an equal chance that there is a God and that there isn’t.  Since there are pretty compelling cases to be made on both sides people tend to find their antecedent beliefs easy to justify and so we get fervent believers on both sides when really we should all just admit that this is an unresolved question. You may have placed your bet on one side or the other but that is all there is to it; a bet.

The A Priori Case

  • The Ontological Argument: This much maligned and misunderstood argument has been the subject of countless attacks, defenses and reformulations. I am convinced that the “existence isn’t a predicate” attack doesn’t work. The basic reason that all versions of this argument are inconclusive is just that our intuitions about the totality of the space of possible worlds is extremely unreliable. We may be able to coherently talk about particular possible worlds –though this is hard in itself– but when we try to conceive of the entire space of possible worlds, as we must if we are to conceive of a necessary being existing in some possible world, we loose our grip on what is going on. This is, of course, the very same problem that the parody ontological arguments face. Just as we cannot trust our intuitions about what objects necessarily exist so too we cannot trust our intuitions about which don’t.
  • The Logical Problem of Evil: I sometimes hear people say that Plantinga’s response to the logical problem of evil involving trans-world depravity has successfully answered the logical problem. But the obvious problem with this argument is that it assumes that God has no control over whether the creaturely essences he instantiates have the contingent property of trans-world depravity. This just seems wildly implausible to me. The possible world where all creaturely essences have morally significant free will and always freely choose to do what is right is conceivable and we can further conceive that these creaturely essences have trans-world sainthood which is the contingent property of always freely choosing to do the right thing.
  • The Logical Problem of Omniscience: Perhaps less discussed is the logical problem of omniscience. The problem here is that God’s foreknowledge is logically incompatible with His own free will. This is distinct from the traditional problem of free will and omniscience in that the claim is that there is a formal contradiction entailed by the set of claims that God is all-powerful, all-knowing, and morally perfect. Here again we see the same problem as we did with the other a priori arguments: the opposition has different a priori intuitions and we do not have any way to adjudicate between them.

The a priori cases are therefore inconclusive.

    The A Posteriori Cases

    • The Cosmological Argument: The family of arguments here all suffer from well known problems. I won’t here rehearse them but it is clear to me that there is a stalemate here resulting from a clash of intuitions about infinite chains, what counts as an explanation, and the epistemological status of the principle of sufficient reason.
    • The Teleological Argument: Again there is well known and entrenched positions on both side of this issue. For myself I find the fine-tuning argument the most compelling and specifically in its evidential form. That is, fine-tuning gives us some evidence for God but it is defeasible. Of course it is possible, though highly unlikely, that this all happened by accident so the fine-tuning argument cannot prove that God exists but it does provide (defeasible) evidence for the existence of God as long as one accepts the claim that some fact F counts as evidence for a claim C just when F is more likely to have occurred given C. People who like the fine-tuning argument thus spend a lot of time justifying this principle. I find it fairly persuasive as an independent principle and so find fine-tuning to be persuasive empirical evidence for the existence of a God, though it is defeasible.
    • The Evidential Problem of Evil: Unlike the logical problem of evil this is the problem of whether or not the existence of the actual amount of evil in the world is evidence for the non-existence of an all-powerful, all-knowing, morally perfect being. Using the same kind of reasoning that we did in the fine-tuning case we seem to be led to thinking that the existence of evil is string evidence for the non-existence of God. Even if one accepts that God must allow some evil in the world the shear quantity of evil in the actual world (whereby evil is just the suffering in the world considered over the history of sentient creation) is much more likely to exist in a world where there is no God. All of the standard reasons that God may have for allowing evil do not license the amount of evil we actually find.
    • The Argument from Religious Experience: It seems to me that we should count experience as a justifier solely to the extent that the experience is repeatable and public to the extent that it can be had by different people at different times. If one accepts this then it would only be legitimate to appeal to religious experience if it were a wide-spread and acknowledge phenomenon. For instance, if we all saw God descend from the sky and smite the devil then it would be ok to say we believe in God because of a religious experience. This *may* have been the case in the past if we take the various holy texts at face value. My hunch is that it was not the case then either but we can leave that aside. More problematic is the moral argument against private religious experience. What compelling moral reason can we give which would justify God’s hiddenness from us? If we take religion at face value God used to be present but now He is missing. Why? if one denies that God was ever present in the way various religious texts say then we still have to wonder why that is the case. Why would a morally perfect being leave us alone?

    Thus we again end up with a tie. We have two hopelessly stalemated arguments and two compelling lines of evidence pointing in opposite directions.

      Natural Metaphysics Blowing Through the Air

      In November 2009 the Center for Ethics and Values in the Sciences at the University of Alabama Birmingham hosted a conference entitled Does Scientific Naturalism Exclude Metaphysics? The speakers were Michael Friedman, Andrew Melnyck, Ron Giere, Mark Wilson, Don Ross, Daniel Dennett, J T Ismael, James Ladyman, and Paul Humphries. The conference was video taped and the videos are now up on YouTube  here courtesy of Sarah Vollmer and her graduate student Morgan Anders who are also in the process of making a short documentary film on the issues raised.

      The conference focused on Ladyman and Ross’ new book Everything Must Go: Metaphysics Naturalized where they argue, first that scientism is true and second that a lot of contemporary metaphysics, even from philosophers who claim to be naturalists, physicalists, and scientismists (Armstrong is cited as an example), relies on a fundamentally misguided and outdated conception of scientific reality as consisting of little billiard balls flying around in space banging into each other, you know basically the idea that Democritus had 2,500 years ago. Scienticism is the view that science, in particular physics and the methods it employs, is the only real way to know about the world. A priori reasoning on this view is no good, especially when it is detached from science or especially when it employs this outdated model of the scientific model. They argue that the proper role of metaphysics is that of elucidating the connections between the various special sciences so that they form a unified picture of the nature of reality. This is a task that falls to no specific science and so can be called metaphysics (they cite as an example the claim that chemistry unified is physics and physics unified is metaphysics).

      My own reaction is to be sympathetic to the criticism of philosophers who try to derive conclusions about the actual world from current a priori reasoning. Given the track record it is far from clear that a priori intuitions are a good guide to the nature of the actual world. They are however a fine guide to the possible worlds. A priori reasoning fills out the space of possible and impossible worlds and science then locates the actual world in that space. The “fictional world” that occupied David Lewis, David Armstrong, as well as philosophers like Locke, Hume, and Kant, is a perfectly respectable possible world and is interesting in so far as it is a live option, which roughly means that it hasn’t been ruled out by scientific inquiry. The main thrust of Ladyman and Ross can then be seen as an argument that science doesn’t bear this picture out and so naturalistically minded philosophers should stop thinking about one set of possible worlds. But nothing in the argument suggests that a priori reasoning about a different set of possible worlds wouldn’t be useful. In fact we need the a priori reasoning about possibilities to make sense of the empirical data and this is the way we will ultimately identify the the mind with the brain, for instance. .

      What we get from this kind of picture is a two-dimensional view on which a priori reasoning gives us the primary intensions of statements and science gives us the secondary intensions, or to put it in more Kripkean terms, the job of science is to reduce the epistemically possible to the metaphysically possible. This is still an empiricist position broadly construed since the claim is that for beings like us the only way to know about the actual world is via empirical means. In fact i would count this as a scientismist position. This is perfectly consistent with the claim that an ideal agent who knew all of the facts would be in a position to know about the actual world in an a priori manner.