Non-physical Things Exist!

This post is part of my series on physicalism.

Physicalism doesn't deny the existence of the non-physical. Indeed, the whole point of physicalism is to explain the non-physical in terms of the physical - and one doesn't try to explain things that don't exist.

I reject as false the view that every attribute is a physical attribute. (Poland, p.188)

Melnyk calls this "retentive realizationism." He reserves the option of "going non-retentive" if necessary, however. For example, he says, if it turns out that "life" cannot be explained as a functional property of physical systems, then the physicalist can deny that "life" exists, and replace it with some appropriate (physically realized) functional property that does exist. This is not as radical as it sounds. He points out that anyone objecting to this move would have to show that, in addition to such functional properties as being able to take in sustenance and transform it into parts of the body, and being able to reproduce, there is something called "life" that is real, while at the same time being above and beyond any such functional properties.

The difference is between explaining something and explaining it away. Take ghosts, for example. If you believe in ghosts, then you might explain them by saying, "Ghosts are the manifestation of spirits of the dead."  If you don't believe in them, you might explain them away: "Ghosts are psychological effects brought about by a deep emotional connection with someone who has died."

Physicalism is in the business of explaining the non-physical, not explaining it away. Consider, for example, a chemical compound that is made of (realized by) some combination of atoms, or an organ - say, a liver - that is composed of some collection of cells. To say that the compound, or the liver, is so composed is not to say that it doesn't exist! Rather, the physical realization allows us to explain why the compound has the chemical properties it does, why the liver functions as it does, on the basis of the underlying structure of atoms/cells and their, more fundamental, properties.

Computers and Minds

This post is part of my series on physicalism.

I like the computer program example so much I want to expand on it beyond what Melnyk says.

Take two computers, one a PC and one a Mac, that are both running some program (say, Windows 7). The two have different processors that must be programmed with different machine languages, so the actual physical events that occur in each computer will be quite different. Yet, if the programming and compiling has been done correctly, the two will be functionally equivalent. The two screens will look the same, and the same changes will occur when I click on the same icon on each.

Even if we were unaware of the details of the program the two computers were running, a careful investigation would reveal the existence of certain groups of physical processes that correspond to particular subroutines of the program in both computers. Thus, the physical-level description could be used to explain why the two computers were behaving in the same way. 

But the existence of a physical-level explanation doesn't invalidate an explanation at the level of the program itself. The same phenomena can be described in terms of IF-THEN statements and FOR loops. Realization physicalism says both types of explanation are valid.

This makes a terrific analogy for understanding the physical realization of mental phenomena like thoughts, emotions, or qualia. There is no way that the physical events occurring in my brain when I look at a red wall are the same as the events occurring in your brain when you look at the same wall. But I see no reason to doubt that there might be a functional equivalence of some sort between the events in the two brains, in a similar manner as between the two computers. It might even be possible, at some point in the future, to analyze the patterns of neural activity and identify groups of processes that correspond to particular aspects of the visual experience. These patterns could then be used to explain why the two people were having a similar experience.

Here, too, the existence of a physical-level explanation doesn't invalidate the explanation of mental phenomena at the level of mental events. ("The baby reached for the pacifier because it wanted something to suck on.") Physicalists don't deny the truth of mental causation any more than they deny that the billiard ball moved because it was struck by the cue ball. The existence of another description of both balls at the level of neutrons, protons, and electrons doesn't invalidate the causal efficacy of the cue ball. Rather, it explains it. And in the same way, the (postulated) physical realization of mental events explains (rather than denies) mental causation.

This is not to claim, of course, that science is at a point where it can prove the physical realization of mental phenomena. But the idea doesn't seem ridiculous on its face. The burden of proof, rather, lies with anyone who would claim that it is impossible for such mental phenomena to be physically realized.

A Sudden Realization...

This post is part of a series on physicalism.

The concept of realization is central to the physicalism program for both Poland and Melnyk. As it is more clearly laid out and (I think) more useful, I will follow Melnyk's version.

For Melnyk (p.26)

Every object is either an object of some physical kind or  a physically realized object of some functional object kind.

The same goes for properties and events: they are either physical or physically realized. The physical, as we have seen, is for Melnyk simply that which is describable in the proprietary language of fundamental physics. But what does it mean for something to be physically realized?

For Melnyk, higher-order types are functional types that are defined via an associated condition. A lower-order object (or property, or event) realizes a functional type if and only if it meets the associated condition.

This is a fairly abstract definition, and it would be really great to have an example here. Melnyk gives a few:

Examples of functional object kinds plausibly include can openers, digestive systems, and cells.... Examples of functional  properties plausibly include transparency, having currency, and being an analgesic.... Examples of functional event kinds plausibly include storms, births, and extinctions.(pp.21-22)

Unfortunately, he neglects to explain what the associated condition is for each of these examples. Perhaps the associated condition for being a can opener is "having the ability to open cans"?

At any rate, we can now say what it means for something to be physically realized:

A token x of functional type, F, is physically realized if and only if (i) x is realized by a token of some physical type, T, and (ii) T meets the associated condition for F solely as a logical consequence of the distribution in the world of physical tokens and the holding of physical laws. (p.23)

Here, again, it would be great to have an example or two, but unfortunately Melnyk doesn't provide any. So let me try to interpret this statement.

A can opener is an object of functional type in that it is capable of opening cans when wielded by a human (with some conditions of the human's strength, size, and mental ability presumably required). Note that "can" and "human" are not definable in purely physical terms, so they are (I think) non-physical in Melnyk's view.

The type "can opener" is physically realized if there is some configuration of atoms that meets the condition of being able to open cans, and does so purely by virtue of the physical properties of the atoms of which it is composed.

The can opener is a bad example, because probably no one doubts that can openers are physically realized. Later, Melnyk gives a really interesting example: a computer program.

A computer program is about as non-physical as something can be. It's an abstract set of processes relating some inputs to some outputs. It's really a mathematical function of a particular sort, though we don't usually talk about it that way. You can, of course, write it down, or type it into your computer so that it is stored in memory, but that doesn't make the program physical any more than writing down your thoughts makes them physical.

A computer program is a great example of a functional type. A particular computer can be said to be running the program if the the physical bits of the computer act according to a certain pattern: they are "related to one another in mathematically specifiable ways." (p.40) In Melnyk's language, those mathematical specifications are the associated condition, and any computer that meets that condition is realizing the computer program.

This, I think, provides an ideal example of what physicalism is all about. It doesn't deny the existence of abstract, non-physical things such as computer programs. But it claims that all actual instances of such things are physically realized.

It's OK To Admit You Don't Know

In the previous post I pointed out some difficulties with the idea of basing physicalism on current physical theory. Poland takes a different approach: he says physicalism should be based on the true physics, whatever that may be. Can we do any better with this approach?

It is important for Poland's program that there be determinate physical bases for physicalism.

By "determinate bases" I mean classes of entities that are well defined: for any entity, there is a fact of the matter as to whether it is included in the bases of the system or not.... Vacuous or indeterminate content, therefore, undermines the significance of physicalist doctrine.... (pp. 147-148)

Yet, Poland claims that it is not necessary that we know what the true physical bases are. It is enough to have a general definition of physics, so that we can recognize it when we see it. By basing physicalism on a general characterization of physics, rather than on any specific physical theory, Poland hopes to avoid the problem of the changing nature of physical theory.

Although our knowledge of the physical bases changes with physical theory, the actual bases themselves do not. And although current theory provides the best estimate of what is in the domain of physics and thus in the bases, it neither provides the content of physicalist theses nor determines their fate. (p. 166)

This seems to me to be a reasonable approach. (Melnyk clearly doesn't think so, but his objection seems to me to miss Poland's point.) In fact, Melnyk and Poland seem to be making a similar point: that it is the cart of physicalism, not the horse of physics pulling the cart, that is the focus of their philosophy. Realization physicalism is about how higher-level theories are related to lower-level theories, not about providing the specifics of the realization for every specific case.

I wonder if it would be better to drop the "physicalism" and just call it "realizationism." After all, if we were able to demonstrate that mental phenomena are realized at the level of neurophysiological processes (say), that would be a more than sufficient accomplishment for the program. It would hardly be necessary to further reduce the neurophysiological processes to the fundamental physical theory of the moment to declare success in a physically-based explanation of mental phenomena.

At any rate, I now need to tell you what "realization" means. I will turn to this task next time.

What Are The Error Bars On That?

I have been summarizing the physicalist views of two authors, Jeffrey Poland and Andrew Melnyk. In this post, I give my own opinions about Melnyk's view. Caveat lector!

Melnyk thinks that physicalism should take as its basis the current consensus physics of practicing physicists. At first I thought this a reasonable approach, but the more I thought about it the less I liked it. I see two very serious problems with it:

• - What if the best current theories of physics are not even logically consistent with each other?
• - What if the ontological basis of current physics is itself problematic?

On the first point, I think many physicists would agree that General Relativity (GR) and current particle physics theories (especially the Standard Model - SM) are logically inconsistent. This is a big problem for Melnyk, because he says the attitude that we should take toward physicalism is the "scientific realist (SR) attitude," that is, to "assign the hypothesis a higher probability than its relevant rivals." (p. 227) But if two theories are logically inconsistent, then we can't assign their joint probability as anything other than zero.

(Some might argue with my characterization of these theories as inconsistent. However, it is easy to see that, historically, there have been many times when our best physical theories were logically inconsistent: for instance, around 1900 Maxwell's electrodynamics was inconsistent with Newtonian mechanics. So physicalism must at least recognize the possibility that, at some given time, a fundamental inconsistency might exist in the physical theory basis.)

On the second point, I would say that we are in exactly this situation with respect to quantum mechanics. The ontological status of the quantum wave function (or state vector) is a matter of considerable dispute. Does it represent something physical, or does it represent our state of knowledge? The latter is my own view, but this seems to create a huge difficulty for physicalism: If all large-scale phenomena are based on (realizable as) quantum phenomena, and quantum phenomena are only understandable as probabilities of certain large-scale phenomena (output of sensors, counters, and other experimental apparatus), then physicalism is in very great danger of circularity.

I don't know what to do about the second problem - this seems like a very serious problem to me. But I have a suggestion about the first. I think Melnyk is simply wrong about the SR attitude being the view of practicing scientists. After all, how could we physicists simultaneously endorse GR and the SM while acknowledging their inconsistency? I think scientists' (or at least physicists') attitude is better described as the "good approximation (GA) attitude." That is, GR provides a good approximation to how the universe works at the largest scales, and the SM provides a good approximation at the smallest scales. Physicists - in spite of what they might themselves say - are not actually interested in whether a theory is true. They are interested in whether it works.

In fact, they have to be. Let me explain. I'll ignore GR for the moment and pretend that all we need for a fundamental theory of physics is a theory of particles and their interactions. Now, the Standard Model consists of some equations that contain various parameters that must be experimentally determined: the speed of light, Planck's constant, the electron's charge, the masses of the quarks, and so forth. Let's suppose that there is a true theory of the universe that is exactly the equations of the SM with some values of those parameters. Let's call that theory SM-true. Our current theory has the same equations, but with some "best fit" values of those same parameters. Call that theory SM-bf.

Now, the probability that SM-bf and SM-true are the same theory is precisely zero. There is zero chance that the specific values that we have deduced from experiment are identical to the true values. It is like trying to hit an infinitesimally small bull's-eye with an infinitely thin dart. (Mathematically, the true value is a set of measure zero in the space of possible parameters.) Certainly, the relevant rivals of our SM-bf would include other SM-like theories with other values of those parameters. But the probability of each of those rivals is zero, too! So the SR attitude is useless in deciding among these rivals.

If you go look up the values of those parameters, you will see them listed with experimental uncertainty after the value. These are what we sometimes call the "error bars." What those uncertainties mean is this: we have no confidence that the parameter takes on the exact value listed, but we have high confidence that the true value lies within the range specified by the value and the experimental uncertainty.

I think endorsing a theory by taking the GA attitude should mean that we believe the universe will behave approximately as described by the theory, where "approximately" means "within the range of expected outcomes as determined by the range of experimental uncertainty in the parameters of the theory, as long as the values of relevant external parameters are within a specified range." Here "external parameters" refer to physical values that are characteristic of the particular situation in question, rather than parameters appearing in the theory. External parameters might include relative velocity, center-of-mass energy, and so forth.

By taking the GA attitude, we can endorse a theory even if we are sure (by reason of logical inconsistency, for example) that it is not a true theory. Thus, we can say that GR is a good approximation at large values of some external length parameter, and SM is a good approximation at small values of that parameter, while not claiming that either one is (or is an approximation to) a true theory.

Melnyk mentions the possibility of treating physicalism as approximately true (p. 225), but dismisses it on the grounds that notions of approximation to the truth are "notoriously hard to explicate satisfactorily." But, given the difficulties of Melyk's own account, and given the fact that physicists have to deal with approximations to the truth all the time, and have developed tools for doing so quantitatively, I think that this must be a more promising avenue than his own.

Physics is Fundamental!

This post is part of a series on naturalism.

In realization physicalism, as described by Poland and Melnyk, everything that exists is either physical, or is realized by a physical property or system. As a physicist, I at first found this starting point very attractive. Of course, physics is at the base of everything! And of course, we should take it as our starting point for our philosophy! (I'm starting to think there's a serious problem with this idea, but I'll save that for a later post.)

But what does it mean for something to be physical? Here Poland and Melnyk part company in an intriguing way.

We start with an apparently fatal dilemma for physicalism. Suppose that by "physics," we mean "physics as currently understood by practicing physicists." Then we have a problem: current physics, as any physicist will admit, is incomplete at best, and inconsistent at worst. Sure, it does a terrific job of approximating what the universe does, but there's very small likelihood that it is a true description of the world. Historically speaking, the best physical theory has turned out, over and over again, to be incorrect. New theories replace old theories all the time: what justification do we have to think that our current best explanation is any different?

But if physicalism is based on a theory of physics that is not true, then physicalism cannot be true, either.

Suppose, though, that we don't base physicalism on the current theories of physics. Then what do we base it on? Some future physics, that (we hope) will be an exactly true description of the world? Even if we had any expectation that we might some day reach that lofty goal, we do not know today the content of that future theory. So physicalism has no determined content: everything reduces to a physics that we know nothing about.

This is known as Hempel's Dilemma, and Poland and Melnyk grasp different horns.

Poland asks why we should believe that physicalism only has determinate content if there is a specific physical theory on which it is based. He points to determinism: philosophers have no trouble accepting that determinism is a meaningful concept, even if it does not refer to a specific deterministic physical theory. Poland argues that, similarly, physics is a meaningful term, even if we don't have a specific physical theory in mind.

If there is a true physical theory that correctly describes the reality that current physical theories purport to describe, then, regardless of whether we ever hit upon such a theory, it and the reality it describes exist and constitute the physical bases required by physicalist theses. (p. 162)

Physics, according to Poland, is

the branch of science concerned with identifying a basic class of objects and attributes and a class of principles that are sufficient for an account of space-time and of the composition, dynamics, and interactions of all occupants. (p. 124)

Poland thinks the dilemma is a false one. He thinks the cart of physicalism can be hitched to a horse called "physics," so defined, rather than to any specific physical theory.

Melnyk firmly declares that physicalism should be defined in terms of current physics. He also agrees that current physics has very little chance of being true. Thus, physicalism so defined has very little chance of being true. He then makes a rather strange move: he says that we can nevertheless endorse physicalism. A physicalist can "comfortably live with the result that physicalism has a very low probability."

To be a physicalist is to take the same attitude - whatever that attitude is - toward the hypothesis of physicalism that those who have broadly realist and antirelativist intuitions take toward what they regard as the best of current scientific hypotheses. (p.225)

Thus, "physicalism is viewed as no more and no less than a scientific hypothesis," (p. 226) and so should be held to the same standard as scientific theories.  These theories do not have a high probability of being true - they are constantly being refined and replaced, after all - but they are "the best we have so far." So, too, for physicalism.

But what happens when physical theory changes? Then, Melnyk says, physicalism as currently defined will have to change, too. The new physicalism will not strictly speaking be the same as Melnyk's physicalism (which is defined in terms of today's physics), but it will be a closely related view that retains the same structure as Melnyk's physicalism. Melnyk allows us to change horses without changing the cart of  physicalism.

I will leave you to ponder these two approaches: next time I will give my own take on them.

Hauser’s Mistakes

Thanks to Matt of Atheism: Proving The Negative, I have learned that Harvard's Marc Hauser, whose book I blogged about recently, is being investigated for possible fraud involving his research. Since I spoke highly of his book, I felt I should pass on the information.

A Gripe

We interrupt your irregularly scheduled discussion of naturalism to complain a bit.

Why can't philosophers learn how to use examples? When I'm teaching my physics classes, I can hardly say five sentences without feeling the need to throw in an example to illustrate whatever it is I'm talking about. There's nothing like a good, clear example to show what the abstract terms mean and how they are used in practice. But philosophers are capable of going on and on for hundreds of pages of the most abstract stuff without a single example.

I noticed this while reading about free will. Practically the only example I ran across was Austin's putt. (Certainly the most over-analysed event in sporting history. That's another thing about philosophers: once they get hold of an example they worry it to death.)

In the two books on physicalism that I'm reading it's just as bad. Melnyk is the better of the two: he occasionally describes a useful example. I'll report on one or two of these in upcoming posts.

But Poland goes on about "function" and "causality" and "instantiation" and "tokens" and so forth, never pausing to give an illustration of what these terms mean in real life. And when he does give an "example," it's so abstract as to be nearly useless. For instance, in describing how the same sort of explanation is used for natural regularities as for exceptions to those regularities, Poland gives the following "example". (And no, I am not making this up. It's on p. 220.)

For example, if an exception to a causal regularity is a case in which the antecedent, but not the consequent, attribute is instantiated, some other attribute being instantiated instead, then such an exception is explained via an account which clarifies the relations between the attribute(s) realizing the antecedent non-physical attribute and the attribute(s) realizing the non-physical attribute that replaced the expected consequent non-physical attribute and which explains how those relations between the physically-based attributes realize the relation between the two actually occurring non-physical attributes.

Thank you so much, Professor Poland.

We now return to your irregularly scheduled discussion.