In celebration of my three years in the Blogosphere I will be reposting some of my earlier posts that I am particularly fond of. This piece was originally published May 10th, 2007.
In his youthful exuberance Rosenthal argued that for a first-order state to count as a conscious state the first-order state had to cause the higher-order state to occur. But he has come to explicitly reject this causal requirement. He now talks about the higher-order thought ‘accompanying’ the target state. It need not have any causal connection to the first-order state at all. What this amounts to is that there are at least two different ways of thinking about the relation between the first-order state and the higher-order state depending on whether you think intentionality is at bottom a matter of description and functional role and holism or a matter of word-world relations and causation, and compositionality. This leads us to what I have called Q-higher-order thoughts and K-higher-order thoughts.
A K-higher-order thought is a higher-order thought that is caused by its target state and so picks it out in a something like a causally complex-demonstrative way. Something like ‘I am, myself, in (dthat) red state.’ In order to count some first-order state as a conscious state it has to be the cause of the higher-order state that targets it. On the other hand a Q-higher-order thought need not be caused by the state that it represents in order to be about it and for us to be conscious of it. It picks out the target state purely by description. The Q-higher-order thought characterizes the first-order state in terms of its resemblances and differences to and from other sensory states like it. Something like ‘I, myself, am in a state that is more like pink than it is like blue and more like orange than it is like green…and etc’. So which of these should we prefer? I have been arguing here, and in response to Pete over at the brain Hammer that this kind of higher-order theory allows us answer the ubiquitous objection from the so-called empty higher-order thought, and more recently, that it gives us a nice response to Pete’s Unicorn argument against Higher-order theories. These, I think, are already powerful reasons to think this is the right way to cast the theory, but one may wonder what else speaks in its favor.
Rosenthal gives two very quick arguments against his former K-Higher-order view, both in a footnote that he added in 2005 (p56). His first argument is that requiring the causal connection between the first-order state and the higher-order state in order for the first-order state to count as a conscious state is theoretically unmotivated. The idea behind this is that the transitivity principle requires only that one be conscious of being in the first-order state; it seems to be silent on what actually causes you to become so conscious. However the causal antecedents of the higher-order state will seem to matter very much if one is influenced by the Grice-Kripke-Fodor picture of the mind. So the claim that the causal requirement is theoretically unmotivated by the transitivity principle is more a revealing fact about Rosenthal that about the higher-order theory. A causal theory of reference is itself powerfully motivated, and if it turns ut to be correct, then we had better have a higher-order thoery that incorperates it (that is, if we want to have a higher-order theory in the first place).
The footnote continues by pointing out that one reason why the idea that the first-order states causes the higher-order state is so intuitive is because it is a way of saving the Cartesian insight that there is an intimate connection between mental states and consciousness. If first-order states are in the business of causing higher-order states about themselves we could easily explain why so many philosophers have thought that being conscious is essential to being a mental state. It also explains why we are conscious of our mental states in an immediate, non-inferential way, which is required by higher-order theories. This looks like some kind of theoretical motivation, so what is it that he finds so problematic?
He argues that if we require that the first-order state cause the higher-order state in order for it to be a conscious state, we end up having to say that being conscious is the ‘normal condition’ for mental states. The reason that we do not want to say that being conscious is the normal condition for mental states is because it obscures the important fact that they may occur unconsciously and that that seems like a pretty normal condition for mental states to be in as well. If the normal condition of a mental state was conscious then it is only if some special causal mechanism intervened in the normal procedure that we would end up with unconscious mental states. But this is wrong because as we saw in the first part of the paper the transitivity principle predicts that any state can occur unconsciously. One is not more normal than the other.
But it is natural to think that some kinds of states are more normally conscious than others. For instance it is natural to think that the sensory sates and other kinds of states that we most likely share with other animals, do normally cause higher-order states about them. Or in other words, it is natural to think that in the case of the sensory states it is more natural to occur consciously, though there are plenty of times when they do not. In the case of thoughts and other more complex forms of mental phenomena it is natural to think that they would be less likely to have to occur consciously being newer perhaps and less in the business of day to day survival. And there are all kinds of stories we can tell about why that is the case and how it would be implemented in a complex system like neural representation. There may be filters, thresholds, feedback networks, both inhibitory and excitatory, and who knows what else. We can do all this without falling into the trap of thinking that the sensory states must always be conscious.