Wednesday, July 26, 2006

"Brains" Moves

This blog has been quite successful, growing to an average of almost 100 unique visitors per day. I have decided to move it to its own domain name, using a more sophisticated blogging tool.

From now on, all new posts will appear only on the new website.

The blog is now at www.philosophyofbrains.com. Please update your links, etc.

And to entice you to visit the new website, I will post some important news as soon as I'm done with this.

Tuesday, July 25, 2006

News

Philosophers' Carnival #33 here.

Augenblick reports on various interesting things, including the following:

The "brain box," a new computer that attempts to mimick the fault-tolerant characteristics of the brain, is being built by scientists at the University of Manchester.

The first neurons to develop in the brain have been identified by researchers at Yale University.

Monday, July 24, 2006

New Philosophical Challenge

Check it out at Zeno's Coffehouse.

Tuesday, July 18, 2006

Some Philosophical Fun

Courtesy of Ben Ricker:

"The BBC web site posted several thought experiments that are/were in vogue in ethics and requested votes on what you would do. Some food for thought. Check it out here. After you vote, you see the total tallys."

Did Fodor know about Sellars?

It is sometimes noticed that Wilfrid Sellars's work in the 1950s is the origin of functional role semantics, contains the language of thought hypothesis, and has a lot in common with functionalism generally. So, it is natural to speculate the Hilary Putnam and Jerry Fodor, when they formulated functionalism in the 1960s, were influenced by Sellars. For instance, Dennett says that Putnam's functionalism was influenced by Sellars's work.

Putnam certainly knew of Sellars's "Empiricism and the Philosophy of Mind" (1956), which was heavily discussed at the time. But there is no evidence that he knew any other work by Sellars, or even that Sellars's work had a large influence on Putnam's functionalism.

As to Fodor, I know of no evidence that Fodor knew anything about Sellars's work. Fodor told me he doesn't remember knowing Sellars's work at the time.

I have discussed the evidence I could find about this in my "Functionalism, Computationalism, and Mental Contents" (in Canadian J. Phil.)

More recently, Bill Lycan told me he thought Fodor must have known of Sellars's work, because Fodor and Chihara, "Operationalism and Ordinary Language" (1965) uses Sellars to criticize Wittgenstein. Unfortunately, upon checking, I was unable to find any references to Sellars in the paper by Fodor and Chihara.

Does anyone know of more evidence bearing on whether Putnam or Fodor knew about Sellars's functionalism in the 1960s?

Sunday, July 16, 2006

Kim vs. the Subset View of Higher Level Properties

Jaegwon Kim is a prolific and influential writer on the topic of higher-vs.-lower level properties and mental causation. Many of his arguments may be seen as raising the following dilemma: either higher level properties reduce to lower level ones (i.e., their causal powers are identical to the causal powers of their lower level realizers), or higher level properties are epiphenomenal.

But it has always seemed to me that the "subset view" escapes Kim's dilemma. By subset view, I mean the view that higher level properties are a part of their realizers, in the sense that their causal powers are a subset of the causal powers of their realizers. (The subset view occurred to me years ago while reading Kim and listening to his talks. I believe a version of the subset view has been defended by Sydney Shoemaker, though I haven't read Shoemaker's work. I took the term from Gillett and Rives's recent paper in Nous. The subset view as I understand it seems consistent with different views of properties: either properties as individuated by their causal powers, or properties as constituted entirely by their causal powers.)

The subset view escapes Kim's dilemma because being a proper subset of something is not the same as being identical with something, and yet there is no reason why a subset of the causal powers of a realizing property (which is assumed to be causally efficacious) should be epiphenomenal.

But to the best of my knowledge, Kim has not discussed the subset view in print. I was interested his opinion, so I emailed him and asked: do you agree that the subset view is a legitimate alternative to reductionism and epiphenomenalism about properties? If not, why? If you reject the subset view, why do you?

The following is an excerpt from Kim's response (reproduced with permission):

"No, I don't think one can escape the mental causation problem by defining "realization" in the way you describe. From the start, this approach looked to me like an attempt to solve a substantive philosophical problem by definitions. Don't you think it sound too neat and too good to be true? One way to see the problem with it is this, I think: If you define a realizer in the way suggested by the "subset" view, how do you show--what does it take to show--that mental properties have physical properties as their realizers? That is, how does one show that the physical realizes the mental? The subset view looks plausible at first blush, I think, because it is presented with the unspoken assumption in the background (which we normally make under our more or less intuitive and unspecific notion of realization) that the mental is physically realized.

"Consider a mental property M. How does M get to have a physical property, P, as one of its realizers? According to the subset definition, the causal powers of M must be a subset of the causal powers of P. How is that possible? We may assume that most of P's causal powers are powers to cause other physical events but we can allow, at this point, that P's causal powers may include causal powers to cause nonphysical events as well. But for the present strategy to work for the mental causation problem, the causal powers of M must include at least some of P's physical causal powers. This amounts to the supposition that M has causal powers to cause physical events. How do we show that? Well, showing that that is possible, or showing how that is possible, is exactly the problem of mental causation. We seem to be back to square one, and very quickly, in a small circle!"

In the rest of his email, Kim also writes that from his point of view, the subset view as I define it counts as a form of reductionism, and is unlikely to satisfy die-hard nonreductive physicalists.

I agree with Kim that die-hard non-reductive physicalists will not be satisfied with the subset view as I have defined it.

But I am not a die-hard non-reductive physicalist. I am happy to say that the causal powers of higher level properties are physical. In fact, I am happy to say that all causally efficacious properties, higher and lower level, are physical (even those higher level properties, if there are any, that are not identical to (but are "parts" of) lower level properties).

I don't think the subset view is a way of defining our way around a philosophical problem. I find the subset view attractive for independent reasons: it seems to me that the subset view accomodates the existence, robustness, and other characteristics of scientific explanations and generalizations at different levels and does so better than identity-based reductionism. This could be the beginning of a long story, but I'll have to stop here for now.

Saturday, July 15, 2006

Computation, Representation, and Teleology

Curtis Brown, "Computation, Representation, and Teleology," presented at E-CAP 2006, June 2006.

I just found the online (long) abstract of Brown's talk. Brown defends two necessary conditions for computation: it must operate on representations (semantic condition) and it must have the function to calculate (teleological condition).

I agree with Brown that there is a teleological condition on computation, at least in the sense of the term that is useful to computer science and cognitive science, and I have argued for this in some of my papers. I'd be curious to know more about what Brown means by "having the function to calculate". Since "calculate" is usually taken to be a synonym of "compute", Brown's teleological condition sounds circular. Unfortunately, the abstract doesn't say what Brown means by "calculate".

As to the semantic condition, I have argued at length that there is no such condition--on the contrary, in my view, computation does not require representation. One way to see this is by defining computations in terms of strings of letters instead of what the letters represent (such as, e.g., numbers). Defininig computations in terms of strings may be impractical when one is doing applications, but it is theoretically insighful.

Brown responds to my view by saying that even when computation is defined in terms of strings, the inputs and outputs of the computation are still representations. The only difference is that they represent strings instead of numbers or something else. This is an original reply, but I suspect it misses my point.

Strings can be seen as concrete entities (strings of concrete physical letters, inputs and outputs of concrete computations) or as abstract mathematical entities (strings of abstract letters, inputs and outputs of abstract computations). Either way, strings may or may not be semantically interpreted, and if they are, they can represent many things (including themselves, of course).

Here is an argument that would support Brown's conclusion. Consider a concrete computation defined in terms of strings. At the very least, it represents itself, or some abstract counterpart to itself. Strings must be represented no less than numbers or anything else does.

Yes, but the point of having a mathematical theory of strings is precisely to study certain properties of the strings without any concern for what (if anything) the strings represent. And one can do the whole mathematical theory of computation purely in terms of strings rather than in terms of what the strings represent.

So, of course, when you do the theory of strings, you need to represent the strings. But when you define computations in terms of strings, you can happily ignore what the strings represent, or even whether they represent anything at all. For all you care, they can be meaningless.

But, one might reply, once you have your computations defined over strings, don't they at least represent themselves (or some abstract version of themselves)? Sure, but everything represents itself (and many other things besides, depending on how it is interpreted). This notion of representation is not going to do the job that traditional supporters of a semantic condition on computation want such a condition to do (i.e., contribute to an account of mental representation).

Caveat: I haven't listened to Brown's presentation and I haven't read his paper. All I saw was the abstract linked to above.

New Directions in DNA Computing

Ehud Shapiro and Yaakov Benenson, "Bringing DNA Computers to Life: Tapping the computing power of biological molecules gives rise to tiny machines that can speak directly to living cells," Scientific American, May 2006 issue.

Jeff Dauer noticed the above article and kindly sent me the link.

Ehud Shapiro is a great guy who works on DNA computing at the Weizenbaum Institute of Science in Israel. I met him recently at an Israeli workshop on the Nature and Origin of Computation, where he presented his work.

DNA computing is computing that exploits the combinatorial properties of DNA and RNA molecules. Traditionally, the goal is to exploit the presence of illions of molecules together to generate massively parallel computations. Lately, this project seems to be losing steam.

Shapiro and his group are pioneering a new kind of DNA computing, aimed at creating a new generation of drugs that can be released within cells depending on whether certain conditions are satisfied. Very cool stuff.

Communication by Gaze Interaction

Anna-Mari Rusanen told me about this story. Kati Lepisto is a Finnish former model who is now almost completely paralized. She communicates by spelling words with her eye movements, and the best reader of her eye movements (and guesser of what she is trying to say) is her mother. Some neuropsychologists are developing a communication device based on simulating her mother. As Anna-Mari notes, this is a very interesting case of cognitive modeling. Unfortunately, the book that tells the whole story is in Finnish.

Friday, July 14, 2006

The Real-life Mary

Greg Frost-Arnold has two interesting posts (first, second) on Sue Barry, a real neuroscientist who recently acquired stereoscopic vision. Her story is told in the latest New Yorker.

Thursday, July 13, 2006

Progress on Mind Reading by Machines

Article in today's NYT.

New Blog

By Brit Brogaard. It's called Lemmings (the term was coined by Weatherson: Language-Epistemology-Metaphysics-Mind-ings). Knowing Brit, I expect her blog to be good.

Monday, July 10, 2006

Difficulties for Psychosemantics

When Bill Lycan visited the NEH Seminar in Mind and Metaphysics last week, he said the problem of intentionality is much harder than the problem of consciousness, because there are four terrible problems facing psychosemantics that no one even talks about:

1. Abstract concepts

2. Metaphors (according to Lycan, "nearly every thought you have is metaphorical")

3. The fact that we can use the same name for different things

4. Propositional attitudes other than belief and desire and their varying directions of fit

(The links are to handouts by Lycan on the topic.)

It's a bit of an overstatement to say that no one talks about these problems, but they certainly have not held center stage in discussions of psychosemantics.

Saturday, July 08, 2006

A Dilemma for Representationalism

(Strong, Reductive) Representationalism about phenomenal consciousness is, roughly, the view that the phenomenal properties of experience can be explained by a combination of representational and functional properties.

The literature is full of putative counterexamples to representationalism (e.g., examples of putatively different experiences that represent the same thing, or examples of experiences that allegedly represent nothing). These putative counterexamples are regularly met with replies that appeal to appropriate representational properties that explain the features of the example.

The relative easiness with which putative counterexamples to representationalism can be met by imagining appropriate representational properties raises the following worry: what are the criteria for attributing representational properties to an experience?

If the game of finding the representational properties of experience is too unconstrained, the theory becomes trivial. For everything can be interpreted to represent a lot of things. So it can't be enough that experiences can be interpreted as representational; there must be criteria for establishing that the proposed representational properties are the correct ones, and these criteria should be motivated independently of the various putative counterexamples to representationalism.

But as soon as we look for criteria, we can start questioning their plausibility. The first criterion that comes to my mind is what the subject of experience takes the experience to represent. But this is not going to be enough to provide the right representational properties for a representational theory of experience. At least in my case, I have lots of experiences that as far as I can tell do not represent anything. So the representationalist must claim that pace me, my experiences do represent the right stuff. How can the representationalist convincingly show that my experiences represent what she needs them to represent without trivializing her theory?

Thursday, July 06, 2006

Wide Representationalism (About Qualia)

Adam Pautz, "Sensory Awareness is not a Wide Physical Relation," Nous, 2006. (The link is to an extended version of the paper.)

Today, at the NEH seminar in Mind and Metaphysics, we discussed Pautz's paper, which is an attack on wide naturalistic representationalism about phenomenal consciousness. (Wide (or externalist) reductive representationalism is the following conjunction: phenomenal characters of experience are constituted (at least in part) by what they represent, and representational contents are constituted (at least in part) by natural relations between brains, their bodies, and their environments.)

Pautz constructs the following putative counterexample (I will simplify the argument a bit). Suppose two similar individuals A and B look at two exactly similar squares. Their nervous systems are slightly different, such that it is reasonable to conclude that they have slightly different experiences. But representationalism entails that the difference in A and B's experiences is a representational difference, and naturalistic externalism entails that the representational difference depends only on some type of causal/informational relation between the experience and what it represents. But the causal/informational relation between A and B's experiences are the same, hence, wide naturalistic representationalism entails the incorrect result that A and B have the same experience, hence wide naturalistic representationalism is false.

In response, one could reject either externalism or naturalism or representationalism. One could also reject realism about the represented properties (I won't take the time to explain this move). But it's not even necessary to do any of that.

(In this paper, Pautz wishes to reject the externalist part only, while retaining the representationalist part. Pautz's argument is actually a step in a longer argument for dualism.)

The weakest bit in Pautz's argument is that he suggests he is refuting wide naturalistic representationalism regardless of what psychosemantic theory it employs, but his argument relies on features possessed by a subset of psychosemantic theories, namely, causal/informational theories. So Pautz's argument at most shows that causal/informational psychosemantic theories have a problem when applied to the contents of experience.

In psychosemantics, there are three kinds of naturalistic externalist theory: (1) theories that assign contents on the basis of relations between internal states and environmental inputs (e.g., Dretske, Fodor); (2) theories that assign contents on the basis of relations between internal states and behaviors (e.g., success semantics); (3) theories that do both (e.g., Millikan, Harman). Some theories (sometimes called two-factor theories) take into account what happens within the brain in addition to the relations between internal and external stuff. If successful, Pautz's argument refutes one-factor theories of type (1), but it does not seem to do much against two-factor theories of type (1) and theories of type (2) and (3).

Pautz does mention two-factor theories, but says it's not clear how to make such theories work. (Maybe, but this has nothing to do with his main argument.) He also mentions theories of type (2) (he seems to think Millikan's theory is of type (2)) and says he doesn't think they are going to work either, but doesn't really argue for it.

Lycan has a commentary on Pautz at the online philosophy conference, in which he calls Pautz on his lack of argument against theories of type (2) and (3). In the same place, Pautz has a reply in which he argues against theories of type (2) (still seeming to think that Millikan has a theory of type (2)). He says he doesn't think these theories work as theories of the contents of phenomenal experience, because they can't deliver contents with the right degree of fine-grainedness. (Once again, this has nothing to do with his original argument in the paper.)

In other words, when pressed, Pautz resorts to arguing that psychosemantics is hard. This is true and well known, and except for one-factor theories of type (1), it's independent of Pautz's main argument.

Wednesday, July 05, 2006

Philosophers' Carnival # 32

Here.

Swampman

Swampman is a (imaginary) physical duplicate of Donald Davidson created by a freak accident by a lighning bolt hitting a swamp. Questions: do swampman's parts (e.g., the parts shaped like a heart, liver, brain, etc.) have functions? Does swampman have intentional states? Does he have qualia? These questions have been heavily discussed in the literature. (The original examples along these lines were discussed by Boorse, Stich, and others in the 1970s.)

Today Bill Lycan said that references to swampman should be eliminated from the literature and he should never be mentioned again. His reason is that intuitions diverge. Some people think the answers to the questions about swampman are obviously yesses, other people think the answers are obviously no's (at least at first; after functioning for a while, swampman will acquire functions, etc. (this is Ruth Millikan's view)). Given this divergence, asks Lycan, what's the point of insisting one way or the other?

It reminds me of something that I think Jerry Fodor wrote somewhere: "Intuition mongering strikes me as vulgar".

Alleged Counterexample to Representationalism

Bernhard Nickel, "Against Intentionalism," forthcoming in Philosophical Studies.

Today, at the NEH seminar in Mind and Metaphysics, we discussed Nickel's forthcoming paper. Nickel proposes a counterexample to representationalism, i.e., the view that the phenomenal aspects of experience are represented features of what is represented by the experience.

The counterexample is a tic-tac-toe board in which all squares are blank. If you look at it, at different times it looks as though different sets of squares in the board are more "prominent" than others. In other words, you see different patterns of squares on the board (though the actual board doesn't change). Sometimes the prominent squares form a cross, sometimes an X, sometimes a T, etc. Alleged upshot: the phenomenal aspects of experience are different but what is represented is the same, hence, representationalism is false.

I'm sorry to say that no one in the seminar (15 participants, 1 TA, Heil, and Lycan) seemed very moved by this alleged counterexample. (This includes those, like me, who are skeptical of representationalism.) Lycan said he found the counterexample interesting and thought it's not obvious what the best reply is, but he also listed eight possible replies! Others offered alternative replies. A plausible one, by Ralph Kennedy, is to say that different concepts are applied to the board in the different cases (analogously to the treatment of the dot array, an example previously discussed in the literature and mentioned by Nickel).

Undoubtedly, more alleged counterexamples to representationalism will be proposed, and more replies will be given. How fruitful is this enterprise?

Monday, July 03, 2006

Teleofunctionalism Uber Alles?

This week, Bill Lycan is visiting the NEH Seminar on Mind and Metaphysics. The main purpose of his visit is to discuss representationalism about qualia.

(According to representationalism, as Lycan formulates it, qualia are represented features of what is represented by a phenomenal experience (e.g., the redness of a tomato quale is the represented redness of the tomato represented by the quale). Representation does not exhaust the nature phenomenal experience, though: there is more to experience than qualia, including what it's like to have the qualia. (Unlike others, Lycan distinguishes between qualia and what it's like to have them.) Lycan's account of what it's like, and generally of non-representational aspects of experience, is functional.)

I asked Lycan what kind of functional account he appeals to for non-representational properties of experience--specifically, whether he still subscribes to the teleofunctionalism that he defended in his 1987 book (Consciousness, Cambridge, MA: MIT Press). I also asked him whether he was aware of any competitors to his teleological formulation of functionalism.

He said he is still happy with what he said on functionalism in the 1987 book, and that he wasn't aware of any other formulations of functionalism that are still on the market. He said, "I seem to have won."

Does anyone have thoughts on this? Is teleofunctionalism the only surviving formulation of functionalism? Is there anyone who defends alternative formulations these days? (What about Cummins's functional analysis; does anyone consider that the basis for an alternative formulation of functionalism?)

Upgrade (7/7): I asked Lycan what he thinks about causal formulations of functionalism (Shoemaker, etc.) and he said they shouldn't be called functionalism because they don't specify which causal relations are relevant among the many that obtain; I asked him about Cummins's functional analysis and he said it leads to anti-realism about functions (because there are so many Cummins functions and which ones we focus on depends on our perspective; this is also Eric Thomson's point in the comments), whereas a genuine functionalism requires realism.