Brain Reading, Brain States, and Higher-order Thoughts

Recently there has been a lot of progress in brain reading; for instance Here is a nice piece done by CNN, here is a nice article on brain reading video games, and here is a link to Frank Tong’s lab, who may be familiar to those who regularly attend the ASSCor the Tuscon conferences. This stuff is important to me because it will ultimately help to solve the empirical question of whether or not animals, or for that matter whether we, have the higher-order states necessary to implement the higher-order strategy for Explaining What It’s Like so I am very encouraged by this kind of progress. The technology involved is mostly fMRI, though in the video game case it is scalp EEG. But though this stuff is encouraging fMRI and scalp EEG are the wrong tools for decoding neural representation, or so I argued in my paper “What is a Brain State?” (2006) Philosophical Psychology 19(6) (which I introduced over at Brain a while ago in my post Brain Statves Vs. States of the Brain). Below is an excerpt from that paper where I introduce an argument from Tom Polger’s (2004) book Natural Minds and elaborate on it a bit.

Polger argues that thinking

that an fMRI shows how to individuate brain states would be like thinking that the identity conditions for cricket matches are to pick out only those features that, statistically, differentially occur during all the cricket games of the past year. (p 56)

The obvious difficulty with this is that it leaves out things that may be important for cricket matches but unique (injuries, unusual plays (p 57)) as well as includes things that are irrelevant to them (number of fans, snack purchasing behavior (ibid)). The same problems hold for fMRI’s: they may include information that is irrelevant and exclude information that is important but unusual. Irrelevant information may be included because fMRI’s show brain areas that are statistically active during a task, while they may exclude relevant information because researchers subtract out patterns of activation observed in control images.

I would add that at mostwhat we should expect from fMRI images are picture of where the brain states we are interested in can be found not pictures of the brain states themselves. They tell us that there is something in THAT area of the brain that would figure in an explanation of the task but they don’t offer us any insight into what that mechanism might be. Knowing that a particular area of the brain is (differentially) active does not allows us to explain how the brain performs the function we associate with that brain area. We need to know more about the activity. Consider an analogy: we have a simple water pump and want to know how it works. We know that pumping the handle up and down gets the water flowing but ‘activity in the handle area’ does not explain how the pump works. Finding out that the handle is active every time water flows out of the pump would lead us to examining the handle with an eye towards trying to see how and why moving it pumps the water.

And, as I go on to argue, after examining those areas to find what the actual mechanisms are neuroscience suggests that it is synchronized neural activity in a specific frequency that codes for the content, both perceptual and intentional, of brain states. So, multi-unit recording technology (recording from several different nuerons in the brain at the same time) is the right kind of technology for looking at brain states. This is not to say, of course, that the fMRI and EEG technology is not valuable and useful. It is, and we can learn a lot about the brain from studying it, but it must be acknowledged that it is ultimatly, explanatorily, useless. To find higher-order thoughts or perceptions we will need to use advanced multi-unit recordings.  

Applying Frigidity

As commonly understood Kripke’s notion of rigidity is a property that some terms have and that others lack. I argue that there is no such property that is had by some terms and lacked by others; hence there is no rigidity as commonly construed . Recent discussions of rigidity have, I claim, forgotten the importance that stipulation plays in Kripke’s original account. In short the argument is that the truth-conditions of sentences with supposed rigid designators in them can vary depending on the stipulative act of the speaker. But if rigidity were a property of the terms themselves the truth-conditions should not vary! I introduce the notion of frigidity which is not a property that terms have, but something that we do and is a tool that we use to evaluate counter-factuals (Introducing Frigidity). We decide to ‘freeze’ the referent of a term and then try to evaluate counter-factual statements in terms of the constant referent. The ‘freezing’ is accomplished by a stipulative act on the part of the speaker.

Thus it follows that there are two ways to perform the thought experiment of frigid stipulation corresponding to taking one or the other terms flanking the identity sign as frigid and asking ‘what about that in another possible world?’ We decide that we are going to stipulate, trivially as Kripke says, that we want to find out about X in a possible world. So for water=H20 we can ask ‘what if H20, this very chemical substance, was in a world that was different from ours?’ If it turns out that H20 is not ‘watery’ that is OK. We can then also ask ‘what about water? Stuff that acts like this, fills our lakes and etc? What if we found a world that had watery stuff that was not H20?’ And that is OK as well. This has the advantage of explaining why people’s intitions vary about whether twater is water.

However Kripke (Kripke 1980) makes the claim that when it comes to mental kinds we cannot do this because in the case of pains and whatnot their properties are not separable in this way. But once we switch from rigidity to frigidity this is less obvious. We can hold the brain state frigid and ask ‘what is it like to have this brain state in a world that is different from ours?’ It may turn out that that very brain state is not like anything to have at all. On the other hand we can hold the sensation of pain frigid and ask questions about worlds with that sensation. It certainly seems logically possible that some of those worlds will have that sensation and yet not have any brain states at all!

This is just what Kripke’s bjection to the identity theory is. He says “this notion seems to me self-evidently absurd. It amounts to the view that the very pain I now have could have existed without being a mental state at all,” (p.147). Well, yes this is true if what he means is that the very brain state he is in and which is his pain might have existed but was not painful for the creature that had it. This is to do no more than admit that there might exist an unfelt pain. He is wrong if he means that a pained creature, one that felt pain, would not be in pain.

The Meaning and Use of ‘is True’

The first thing that we need to do is to make a distinction between the redundancy theory of truth, which is a claim about the use of the predicate ‘is true’ in a natural language, and deflationsim, which is a metaphysical claim about the nature of the property picked out by ‘is true’. Usually what you find is that people just use ‘minimalism’ and ignore this difference though they seem to think that redundancy is true and so therefore deflationsism is true (Blackburn is a classic example of this).

The main motivation for redundancy is a collapsing of the meaning/use distinction that is characteristic of Horwich and other neo-Wittgensienians. If the meaning of a word just is the way that that word is used, the function it conventionally plays in a public language game, then finding out how people use the truth predicate and abstracting the rule that defines its function (the T-schema) is finding out the essence of our concept of truth. But there are reasons not to conflate meaning and use (which I won’t go into here). While I do think that people often use the word ‘true’ as a way of communicating that they agree with either what they themselves, or someone else, has said this communicative use of the predicate ‘is true’ depends on its having the correspondence meaning. ‘True’, the English word, means something like ‘being in accordance with the actual state of affairs’ and so it is easy to see how I could use it to express agreement with what has already been said; to say that something is true is to say that it is really the way things are. So in conversation I am able to exploit that meaning in order to indicate that I agree with something that has been said, I am in effect saying ‘yes, that is in accordance with the facts’.

We exploit the meanings of words in this way quite often. Searle (Searle 1969/2001, p. 142) pointed out a similar phenomenon with ‘promise.’ Suppose a parent says to their lazy child “clean your room or I promise I will take away your cell phone!” It is very odd to think the parent is actually promising to do anything here since the thing promised is not something that the child wants the parent to do. In fact this kind of utterance is most likely a threat or a warning. Or consider a professor confronting a student suspected of plagiarism. The professor says “this passage is taken from Wikipedia” and the student says “I didn’t plagiarize! I promise I didn’t!” This doesn’t look like a promise either, how can you promise that you did not do something? This is rather an emphatic denial of the professor’s accusation. How is this possible? It is because the verb ‘to promise’ is one of the strongest indicators of commitment in the English language, and so we adapt it in these cases as a way of indicating that we are really committed. It would be very hard to explain, from Horwich’s view, how the predicate came to have the function of indicating agreement in the first place without appealing to the correspondence meaning that the word has. If this is right then one of the motivations for accepting deflationsim about truth falls apart.

What’s so Unobservable about Causation?

What is it that is supposed to be so unobservable about causation? This question has nagged at me since my days as an undergraduate (and there has recently been a lot of discussion of it over at Brains, which inspired me to post on it). It had always seemed to me that the causal relation was entirely observable. We even have evidence that we have been observing it since before we could talk. For instance from the cradle we witness the passing of the sun behind clouds and the subsequent darkness, we witness objects falling to the ground, we witness that motion is attended with sound, and etc. Nor are these examples special; cases like these can be multiplied indefinitely. We can see water extinguish flame. While walking about I kick a stone and see my foot cause the rock to begin its trajectory. I can see the acid turn the litmus paper red. Pointing a magnifying glass at a piece of paper on a sunny day will cause the paper to smoke and turn brown, eventually catching fire. A small dog biting at my ankles will cause a pinching sensation and perhaps annoyance! I take these to be examples of seeing A cause B, seeing A causing B and feeling A cause B.

So why think that the causal relation is unobservable? I take the canonical text on this to be Section seven of Hume’s Enquiry. For instance in paragraph six of that section he says,

When we look about us towards external objects, and consider the operation of causes, we are never able, in any single instance, to discover any power or necessary connexion; any quality, which binds the effect to the cause, and renders the one an infallible consequence of the other. We only find, that the one ball does, actually, in fact, follow the other. The impulse of one billiard-ball is attended with motion in the second. This is the whole as it appears to the outward senses. The mind feels no sentiment or inward impression from this succession of objects: Consequently, there is not, in any single, particular instance of cause and effect, any thing which can suggest the idea of power or necessary connexion. (Hume 1999)

No doubt it was passages like this that lead to Kant’s rather rude awakening from his dogmatic slumber. It seems to me to have plunged us into one. In the above passage, which is completely typical of that section, Hume seems to be saying that since we cannot see that the connection between these events is necessary, which is to say that we cannot see why what happens is the ‘infallible consequence’ of the first event as opposed to some other event, there is no way that the idea of causation could be taken from anything perceptible in the outside world. It follows from what he says that the causal relation is not observable; something that makes no impression on the human senses would have to be. The intuition that we have to see the necessity of the causal relation in order to see it at all has guided discussions of the observability of the causal relation. It seems to me that Hume here fails to make a distinction between what we observe and how we analyze what we see. What we observe is a relation; it is when we turn to how to characterize that relation that necessity is involved.

As C.J. Ducasse pointed out, Hume’s claim about the unobservability of causation would be true only under “the assumption that a ‘connection’ is an entity of the same sort as the terms themselves between which it holds…” (Ducasse 1993, p 131). Hume’s mistake was to look for the sense impression of a relation. This we will not find unless relations exist in the same way as the things they relate. But the relation between two objects is not itself a third object! Consider two objects, one to the left of the other. We do not see the relation ‘to the left of,’ but rather see one thing as being to the left of another. It does not even make sense to ask ‘what does the relation ‘to the left of’ look like?’ If asked this question all we could do is describe the position of the objects along with the definition of ‘left,’ and ‘right.’

Any further questioning about what ‘left of’ looks like would betray a category mistake. That is, mistaking a relation for an entity of the sort that it holds between. By way of comparison imagine Hume arguing that ‘left of’ is unobservable. I mean all we see is object A and Object B no where is there an impression of ‘left of,’ and so strictly speaking our idea of ‘left of’ is meaningless. But this is ridiculous! The argument seems to work against any relation and so proves too much.

When we see one ball hit the other we observe that the two balls are related, one causes the other to move, but we do not see that the relation is necessary even though it is in fact a necessary one. Just as we can see X moving without seeing that X is going 22 MPH, or we can see X being 3 ft away from Y without seeing that X is 3 ft away from Y. Despite the fact that I do not see that X is three feet from me I do see the relation ‘3 ft away from,’ I just don’t know that it is this relation. We see relation X qua relation while not seeing that X is such and such a relation. So on the view I am advocating we see A and we see B and we see A cause B but whether or not we see THAT the relation is necessary is a different question. 

As ‘Corny As I Want To Be

As some of you may know, I have been mounting an offensive against Pete Mandik’s Unicorn argument against higher-order theories of consciousness. We have been having quite a bit of discussion over at the Brain Hammer (Me So ‘Corny) on whether or not my proposed answer works or not, and so I thought I would take this opportunity to sum up the debate so far.

The Argument

Pete’s argument is actually quite simple. Here is the way that he puts it:

First, some quick and dirty definitions of my targets:

[Higher-order Representationalism] – The property of being a conscious state consists in being a represented state.

P1. Things that don’t exist don’t instantiate properties.

P2. We represent things that don’t exist.

P3. Representing something does not suffice to confer a property to that thing.

C1. Representing a state does not suffice to confer the property of being conscious to that state (so [higher-order representationalism] is false).

There is another conclusion (C2) that first-order representationalism is false, but I already knew that and so will ignore it.

Two Ways to Kill a ‘Corn

Now it is not secret that I think that is a bad argument that rests on several misunderstandings of the higher-order theory. It is not a threat to Rosenthal’s version of higher-order theory because he would deny the assumption needed to get P3 and hence C1. Here is the way I put it in Kripke, Consciousness and the ‘Corn.

[T]his argument does not threaten Rosenthal’s version of higher-order theory because for him the higher-order thought does not ‘transfer’ or ‘confer’ the property of consciousness to the first order state. For him the property of being a conscious state consists solely in my representing myself as being in a certain state. The first-order state is not changed in any way by the higher-order thought. The only thing that has changed is that the creature is now aware of itself as being in the state.

Now it may be counter-intuitive to say that the higher-order state in no way changes the first-order state, but intuition is not argument. Also, the transitivity principle commits you to this claim as I detailed in Explaining What It’s Like, and as Rosenthal is well aware of. Here is his response to the problem posed by P2 (The interviewer is Uriah Kriegal)]

Ephilosopher: Professor Rosenthal, let me raise one final difficulty for your theory. According to your theory, what it is like for the subject to be in a conscious state is determined by how that state is represented by the second-order state. But what happens when there is a misrepresentational second-order state, with no first-order state at all? It seems your theory commits you to saying that, in such cases, the subject is under the false impression that she is having a particular kind of conscious experience, when in fact she is not. Doesn’t that strike you as absurd, though?

David Rosenthal: Answering this question requires a lot of care in how we put things. We can get a feel for what’s at issue by considering a case that actually occurs. Dental patients sometimes seem to themselves to feel pain even when the relevant pain nerve endings are dead or anaesthetized. The widely held explanation is that these patients feel sensations of fear and vibration as though those sensations were pain. We certainly have no trouble understanding this explanation. But how should we describe what’s happening specifically in terms of the patient’s conscious states? It’s undeniable that the patient is in some conscious state, but what kind of conscious state is it? From the patient’s subjective, first-person point of view, the conscious state is a pain, but we have substantial independent reason to say that there simply is no pain. How we describe this case depends on whether we focus primarily on the state of which the patient is actually conscious or on the way the patient is conscious of it. The trouble is that these two things come apart; the patient is conscious of sensations of fear and vibration, but conscious of them as pain. So it’s not at all absurd, but only unexpected, that one be conscious of oneself as being in a state that one is not actually in. It’s worth noting that this divergence between the state of which somebody is actually conscious and how that person is conscious of it has practical importance. The area of so-called dental fear is of interest to dentists and to theorists because patients who understand what’s happening readily come to be conscious of their sensations as sensations of vibration and fear, which is not especially bothersome. How one represents one’s experiences does determine what those experiences are like for one. Is this really the kind of case you asked about? You asked about what happens when one has a higher-order thought that one is in a state that doesn’t occur. But maybe we should treat the dental case rather as a higher-order thought that misdescribes its target; it misdescribes sensations of fear and vibration as a sensation of pain. But I think it will never matter which way we describe things. When a higher-order thought occurs, there are always other mental states, as well. So whenever a higher-order thought doesn’t accurately describe any state that actually occurs, we can say either that it misdescribes some actual state or that it’s about some nonexistent state; it won’t make any difference which way we characterize the situation

So on Rosenthal’s view there simply is no difference between saying that the HOT represents a state that does not exist and saying that it misrepresents a state that does exist. So Rosenthal’s versionof higher-order theory is completely unaffected bythe unicorn argument.

Even so, it does commit him to saying some strange sounding things, but there is another way to think of the relation between the higher-order state and the first-order state, and gives rise to the distinction between what I call K-HOTs and Q-HOTs (Two Concepts of Transitive Consciousness). A K-HOT is caused by the first-order state that it represents, whereas a Q-HOT simply ‘accompanies’ the first-order state it reporesents. Rosenthal used to endorse K-HOTs but has since moved to Q-HOTs, but as I argued in ‘Two Concepts’ there is no reason to abandon K-HOTs and the give us a second, more convincing, way to kill the ‘corn. Here’s how.

A K-HOT represents its target state via the concepts at the disposal of the creature in question in just the same way that Rosenthal has spent so long arguing is the case. The difference is that the K-HOT is (theoretically) required to be caused by some first-order state or other and it is that causal link that determines what first-order state the higher-order state is about. So, K-HOTs will NEVER represent a first-order state that does not exist, it will rather ALWAYS represent (or misrepresent) a state that does in fact exist. So the property of being represented is none other than the property of causing a higher-order state. This means that while it may be true that WE represent things that do not exists, K-HOTs do not. So again, P2 and P3 are blocked.

So whether you link Quine and Q-HOTs or Kripke and K-HOTs the unicorn is no threat to higher-order theories. Of course having said that I think there are reasons to prefer K-HOTs butthat is another story.