It seems to me that, in order for non-naturalists and naturalists to avoid talking past one another, we need to arrive at an understanding of normativity that is topic-neutral between non-naturalist and naturalist characterizations (just as we have a concept of phone that is topic-neutral between landline and cellular varieties).  But once we have a topic-neutral characterization in hand, it seems possible that this will allow a functionalist reduction of the normative to features of the natural world.

E.g., one might propose the following (I hope) topic-neutral analysis of normative statements:

S ought to φ iff

1. Importance Requirement: Some set of possible worlds is designated as important, or as more important than its complement.
2. Modal Requirement: S must/probably φ-es in the set of possible worlds designated as important.
3. Failure Requirement: S can not-φ in a set of possible worlds that includes the actual world.

Note: If you think further requirements are applicable, you can replace “iff” with “only if”.  The Modal Requirement is emphasized by linguists like Kratzer and von Fintel, who understand “ought” respectively as either a strong necessity or weak necessity modal.  The Failure Requirement is noted by Kant: the point is roughly that when there is no possibility of failing to φ, we would be inclined to replace the ought-claim with a will- or must-claim.  But what philosophers in ethics and meta-ethics have in mind for the most part when they talk about normativity is the Importance Requirement, and sometimes what they mean by normativity is just importance, not strong/weak necessity or the possibility of failure.  (So let’s call the concept of importance taken by itself as a narrowly normative concept, though I prefer to classify it as an evaluative concept.)

If the analysis is right, we may characterize normativity as the higher-order property of having, in part, the property of meeting the Importance Requirement.  Call this last property Importance.  We may then ask what in the natural world confers Importance.

Humeans would say: passions and desires (and perhaps also other elements of human psychology).  Take desires to be propositional attitudes, and propositions to be sets of possible worlds.  My desire that p selects the set of possible worlds at which p is true as important, or as more important than its complement.  The stronger the occurrent desire that p, the greater the importance assigned to p-worlds, other things remaining the same… but there are other ways of assigning priority available to Humeans (for instance, if the desire that ~p is a disposition that remains stable under critical and consistent reflection, or is integral to S‘s character, whereas the occurrent desire that p is unstable and uncharacteristic, then ~p-worlds are more important than p-worlds).

At this point I may even try to turn the tables on non-naturalists and non-Humeans, and ask them how there can be anything of importance in the world without someone taking it to be important.  Humeans try to account for all the importance assigned to things in the world in terms of what we find to be of psychological importance.  So the onus seems to be on the non-naturalists to explain how there could be things of (free-floating) importance in the world, but my impression is that they tend to set this burden aside by resorting to mysterianism.

Also, what I call “psychological importance” is already narrowly normative, or as I prefer to call it, evaluative.  That I find it psychologically important to φ does not just help explain why I ought to φ; it also helps in justifying why I ought to φ.  So, psychological importance is both a natural and a narrowly normative or evaluative property.  In this sense, my position is a form of non-reductive naturalism, and the kind of project I embrace is a piecemeal, bottom-up approach to identifying various sources of psychological importance and ways of weighting and channeling it.  Having identified the relevant sources and mechanisms (the cement and building blocks of full-blown, all-things-considered normativity), if we can then derive more complex norms that approximate the rich normative structures we already recognize, or come to see as what we ought to recognize, then I would count the project as having succeeded.

Richard may have read my position as a reductive version of naturalism because the stuff about psychological importance is nestled in a teleosemantic account of mental representations, which ultimately appeals to biological fitness to help identify and explain the proper functions of mechanisms that produce/consume representations.  But contribution to biological fitness only explains why it is the proper function of the desire-producing mechanism to produce representations that its consumer finds psychologically important to realize; it does not use (what I will call) the proto-normativity inherent in that proper function to provide justification for or criticism of our actions.  By the way, I think teleosemantics is a limited theoretical resource for identifying and accounting for the proto-norms of our biologically based psychological mechanisms.  We may also use the resources of evolutionary game theory to identify the proper functions of various weighting and channeling mechanisms.

Now to show my utter ignorance of logic.  I’m trying to rephrase Quine’s answer—i.e., “Everything exists”—in first order logic, but find myself stuck.  Here’s why.

On the one hand, I shouldn’t write: ∀x Exists(x).  For in first order logic, existence is not expressed by means of a predicate, but by means of existential quantification.

On the other hand, this doesn’t seem to work either: ∀x ∃x.  That doesn’t look well formed, since quantified statements have the form ∀x …x…, ∃x …x…, ∀x ∃y …x…y…, and so on.

Here’s a third option: ∀x x = x.  Or let me try: ∀x ∃y x = y.  But these are not what I wanted to say.  I want my statement to express the thought that everything exists, not the different thought that everything is identical to itself, or the thought that for anything there’s something identical to it.

So the question is, how do I state “Everything exists” in first order logic?  And while we are at it, how about also: ”Something exists”?

Here’s my proposal on how to admit quantification over absences with a clear conscience: paraphrase talk of absences in terms of relation-talk.  What we need is a three-place relation between: 

1. an absentee (the entity that’s absent),
2. locus₁ where it’s absent, and
3. locus₂ (that does not overlap locus₁) where it’s present.

That’s my basic proposal for explicating absence-talk in terms of relation-talk, eliminating the former in favor of the latter.  Absence-talk in counterfactual and manipulationist accounts of causation will require a more complicated modification of the basic proposal, but you get the idea.

P.S., this basic proposal is inspired by the Nyaya system of philosophy, one of the six Brahmanical systems in India.  The Nyaya philosophers admit absences into their ontology, which I find very curious.  My impression is that they take absence as a binary relation between the absentee and the locus where it’s absent.  But my further impression is that I’m probably misrepresenting and oversimplifying their account, and that a close study of the Nyaya position would be a fruitful, worthwhile endeavor.

• Eating Vegetarian Is Taking Global Warming Personally, by Kathy Freston

Freston writes:

A U.N. report from just this past November found that a full 18 percent of global warming emissions come from raising chickens, turkeys, pigs, and other animals for food. That’s about 40 percent more than all the cars, trucks, airplanes, and all other forms of transport combined (13 percent). It’s also more than all the homes and offices in the world put together (8 percent).

That figure (18%) and the comparisons are just staggering.  The author unfortunately does not explain why the mass production of animals causes so much of “global warming emissions”.  At first I thought it might be from the carbon dioxide produced in animal respiration.  But not really.  It’s the flatulent cows, pigs, etc., releasing all that methane into the air from their digestive tracts.  Here’s Noam Mohr’s lowdown on methane and animal agriculture from EarthSave.org, and his conclusion, in the article A New Global Warming Strategy:

By far the most important non-CO2 greenhouse gas is methane, and the number one source of methane worldwide is animal agriculture.

Methane is responsible for nearly as much global warming as all other non-CO2 greenhouse gases put together. Methane is 21 times more powerful a greenhouse gas than CO2. While atmospheric concentrations of CO2 have risen by about 31% since pre-industrial times, methane concentrations have more than doubled. Whereas human sources of CO2 amount to just 3% of natural emissions, human sources produce one and a half times as much methane as all natural sources. In fact, the effect of our methane emissions may be compounded as methane-induced warming in turn stimulates microbial decay of organic matter in wetlands—the primary natural source of methane.

With methane emissions causing nearly half of the planet’s human-induced warming, methane reduction must be a priority. Methane is produced by a number of sources, including coal mining and landfills—but the number one source worldwide is animal agriculture. Animal agriculture produces more than 100 million tons of methane a year. And this source is on the rise: global meat consumption has increased fivefold in the past fifty years, and shows little sign of abating. About 85% of this methane is produced in the digestive processes of livestock, and while a single cow releases a relatively small amount of methane, the collective effect on the environment of the hundreds of millions of livestock animals worldwide is enormous. An additional 15% of animal agricultural methane emissions are released from the massive “lagoons” used to store untreated farm animal waste, and already a target of environmentalists’ for their role as the number one source of water pollution in the U.S.

The conclusion is simple: arguably the best way to reduce global warming in our lifetimes is to reduce or eliminate our consumption of animal products. Simply by going vegetarian (or, strictly speaking, vegan), we can eliminate one of the major sources of emissions of methane, the greenhouse gas responsible for almost half of the global warming impacting the planet today.

UPDATE (January 29, 2008): The New York Times has an article relevant to this post, titled Rethinking the Meat-Guzzler.  The article mentions that “beef generates 24 times more carbon dioxide equivalent gases than the vegetables and the rice“, in the form of fossil fuels burnt in producing beef.  So the production of meat contributes more greenhouse gases than vegetables, not just in terms of methane, but also in terms of CO2.  (Hat tip: Richard Chappell)

To elaborate a bit, we have certain psychological abilities, namely the ability to maximize our utility by striving after what we most prefer, the ability to commit ourselves to long-term plans, and the ability to sympathize with the pains and pleasures that we ourselves do not currently feel.  I expect evolutionary game theory to explain why we have these three abilities.  More importantly for the purpose of this post, I claim that maximization, commitment, and sympathy are norm- and reason-giving faculties, which generate prudential as well as moral norms.  This is why I believe it possible to reduce morality to these psychological faculties without violating the is-ought gap.  For all I am claiming is that we can derive morality from our normative deliberations on morally neutral reasons given by these three psychological faculties.  And the best instrument of this reduction or derivation is the sort of contractarianism that Gauthier has developed. 

Philosophers nowadays recognize two broad varieties of contract theory.  One is contractarianism, and the other is contractualism.   Contractualism stems from Kant, and its contemporary proponents are Rawls and Scanlon.  Contractarianism can be traced back to Hobbes, and its greatest present-day advocate is Gauthier.  Perhaps the most important difference between these two variants is that while contractarianism clearly seeks to derive morality from a non-moral basis, contractualism arguably does not.  As Hampton (p.50) points out, Kant assumed the intrinsic value or dignity of individual persons as ends-in-themselves before conducting an inquiry into what people could reasonably agree to.  Ashford and Mulgan generalize this point when they observe that contractualism is “grounded in the equal moral status of persons”.  Hobbes, on the other hand, assumed that individual persons only have instrumental value or a market price in the state of nature, and sought to derive individual rights from rational self-interest.  Gauthier develops the Hobbesian line of thought in light of rational choice theory.  He seeks to derive morality from the application of rational choice theory to strategic contexts (i.e., game theory) involving utility-maximizing agents, without using any moral premisses in the derivation.  In contrast, Rawls for instance does use moral premisses in his derivation of the principles of justice from the original position.  For the original position makes use of the veil of ignorance, and no one would want to place themselves behind this veil unless they were already impartial.  So the assumption is a moral premiss requiring impartiality, which plays an essential role in deriving Rawls’s two principles of justice.

I hope we are now clearer about the distinction between contractarianism and contractualism.  Allow me then to recast the distinction in the following terms, which presents a starker contrast between the two variants of contract theory.  Contractarianism aims to reduce morality to non-moral constituents, by showing how these constituents figure in the derivation of morality.  On the other hand, contractualism only aims to show how certain aspects of morality derive from certain moral and non-moral constituents.  These are stipulative definitions, and although they have some continuities with the usual understanding of contractarianism and contractualism, I emphasize that contemporary philosophers do not have strictly these definitions in mind when they use these terms.  Now, according to these definitions, Rawls is a contractualist and Gauthier is a contractarian.  While Rawls shows how principles of justice can be derived from rational choice theory and impartiality, Gauthier aims to show how impartiality itself can be derived from rational choice theory.  The latter project seems intellectually more satisfying, although more ambitious.  But since Gauthier has already made so much progress in the project, I can pick up where he left off.

Some revisions do need to be made to Gauthier’s contractarianism, however.  Gauthier tries to derive morality as the product of agreement by utility-maximizing agents interacting with one another in strategic contexts.  I shall argue in a future post that Gauthier cannot succeed in this project so long as he relies on the exiguous basis of utility maximization.  This is roughly because maximization of individual utility sometimes requires that we violate our moral commitments, or disregard the dictates of benevolence.  But when we add that the agents are not only utility-maximizing, but also committed and sympathetic creatures, it becomes easier to derive morality as the product of agreement between such agents. 

Now, I know, you will raise the worry that commitment and sympathy seem to be moral features of agents.  To allay this worry, I can give you immoral and non-moral uses of commitment and sympathy.  As for commitment, a thief can be immorally committed to his lifestyle, even though burgling is a risky and difficult job.  And an artist can be non-morally committed to creating great works of art.  How about (Humean) sympathy?  Actors can non-morally sympathize with the characters whose roles they are playing.  And most people use sympathy in limited altruism towards the near and dear, which can result in immoral cases of nepotism, racism, etc.

Here is a further consideration that may silence your worry.  Prudence is the typical source of non-moral reasons.  The great moral skeptics or amoralists known to philosophical literature, such as Plato’s Thrasymachus or Hobbes’s Foole or Hume’s sensible knave, are non-moral but prudential creatures.  So it was thought by Plato or Hobbes or Hume that prudential reasons for being moral could convince them to be moral.  But think of what prudence involves.  It involves the ability to make long-term plans and stick to them even at the expense of foregoing immediate pleasures and suffering short-term costs.  In short, it involves the ability to make undeviating commitments.  It also involves the ability to simulate one’s future pains and pleasures, even though one cannot feel them now, so that they can now be taken into account in formulating a long-term plan.  And this ability is just what I mean by sympathy.  A utility-maximizing agent who lacks commitment and sympathy cannot be a prudential agent.  For he cannot stick to past plans if they require his failing to maximize utility in his present situation.  This feature of utility maximization is known as modular rationality.  A purely utility-maximizing agent, lacking commitment and sympathy, can make decisions only in short-lived modules, and cannot coordinate his life as a whole.

Prudence, then, requires commitment and sympathy in addition to utility maximization.  The norms of prudence can be reduced to the norms of commitment, sympathy, and utility maximization, roughly as follows.  Sympathy feeds anticipated future pains and pleasures into present preferences, commitment projects present preferences into future courses of action, and utility maximization aims at fulfilling present preferences thus informed by one’s past commitments and simulated future.  Since prudence is a non-moral source of norms and reasons, and can be reduced to commitment, sympathy and utility maximization, commitment and sympathy are a fortiori non-moral sources of norms and reasons.  So I can use commitment and sympathy along with utility maximization as a non-moral yet normative basis from which to derive the norms of morality. 

It is important to note, however, that I am not trying to reduce morality to prudence, or use prudential reasons to justify morality.  The reasons generated by maximization, sympathy and commitment are deployed in one way in the case of prudence, and deployed in another way in the case of morality.  One consequence of this is that I believe there can be conflict between prudence and morality.  In my opinion this is the right result.  Whether this conflict can be resolved by looking at the common sources of norms underlying both prudence and morality seems to be a complex matter I won’t examine in a long while.

Sources:

• Hampton, Jean (1991), “Two Faces of Contractarian Thought” in Vallentyne (ed.), Contractarianism and Rational Choice: Essays in David Gauthier’s Morals by Agreement, pp.31~55.
• I have provided links to the other sources, which are entries in the Stanford Encyclopedia of Philosophy.

Objection 2 suggested that the difference between the arrangements of microparticles composing into a computer running chess or kness can explain the systematic difference between the movements of knight-wised arranged microparticles in chess and kness.  This obviates the need to posit the laws of kness and chess regulating the movements of microparticles in addition to the laws of microphysics.

In response I had noted that even on this suggestion there should be a nomic correlation between microparticles arranged chess- or kness-program-wise and other microparticles moving systematically according to chess or kness rules.  And there must be indefinitely many nomic correlations like this, since there are indefinitely many conventional rules.  So much I have argued in the previous post, and in the present post I had intended to use multiple realizability considerations to argue that these nomic correlations themselves are best subsumed under irreducible and higher-level conventional rules than under base-level microphysical laws. 

But now it seems to me terribly misguided and confused to accord the status of law to rules of convention.  There’s a world of difference between laws which regulate the behavior of objects with natural necessity, and rules of convention which do not.  The  computer is a mechanism that operates with natural necessity in accordance with the laws of mechanics, designed to produce output that covaries with input.  The rules of chess or kness are part of the input, i.e., instructions that we feed to the mechanism on how it should process incoming data.  The mechanism then carries out these instructions with nomic necessity, not because its behavior is governed by these instructions, but because it is governed by the laws of mechanics.  In general, if computers and brains can be modeled abstractly as universal Turing machines, rules of convention should be understood as instructions fed to the machines as input, rather than as laws regulating the physical implementations of such machines. 

My final assessment, then, is that the thought experiment fails to establish downward causation.  As this is not the desired result, I am a bit disappointed.  At least I know now that an argument aiming to show downward causation should appeal to irreducible laws which operate with natural necessity rather than to rules of convention. 

Now let me explain why I think there need not be unacceptable conflict between downward causation and causal closure.  This is often presented as a matter of conceptual impossibility: if there is causal closure of the microphysical, then it is impossible or inconsistent to conceive that downward causation occurs. 

Let me suggest one conceivable way of reconciling downward causation with causal closure of the microphysical.  This is by adopting a pluralistic account of causation in the real world.  Now, there are many different accounts of causation available: e.g., various regularity accounts dating back to Hume, the counterfactual analysis due to Lewis, the recent manipulationist account of Woodward, probabilistic accounts, conserved quantity theories (Salmon, Dowe), etc.  I think the best account of the nature of causation in the real world will combine a nomic subsumption account with a conserved quantity account.  Namely, actual instances of causation will involve the propagation and transfer of universally conserved quantities (like mass-energy) in natural processes in accordance with laws of the form “in relevantly similar circumstances, similar things will happen”.  By laws here I don’t mean just laws of nature which are exceptionless, but also the invariances captured by ceteris paribus generalizations made in the special sciences.  We can also add qualifications to allow for probabilistic laws. 

If we accept such a pluralistic account of causation, we see that the claims of downward causation and causal closure of the microphysical are ambiguous, and that there is a disambiguation of the two claims which can reconcile both with one another.  When I accept the causal closure of the microphysical domain, what I accept is that the domain as a whole is one in which matter-energy (and other universally conserved quantities) are conserved.  But I deny that matter-energy is conserved only in accordance with microphysical laws.  The conservation is also in accordance with higher-level laws.  And when I accept that higher-level laws make a difference in the collective behavior of microparticles, I am not claiming that higher-level objects and properties generate new fundamental forces which inject additional energy into the microphysical domain.  I am only claiming that new laws emerge at higher levels which channel the fundamental forces in the microphysical domain in new ways.  So I deny what Terry Horgan says about causal emergentism, i.e., that it views higher level properties “as fundamental force-generating properties [which] generate new forces over and above those generated by the causal properties of physics, so that the net force affecting the distribution of matter is different from the net physical force.”  The net force is the same, there is a difference only in the nomically patterned ways of channeling the fundamental forces at the microphysical level.  To continue in this figure of speech, the microphysical laws are coarsely patterned, whereas the laws at higher levels are more finely patterned.  This figure of speech may not be very helpful, but it is the only way in which I can conceive the compatibility of downward causation and causal closure. 

Does the thought experiment show the possibility of downward causation?  Not by a long shot.  It needs to be bolstered by further considerations.  At least I hope my response to the first objection (which modifies the thought experiment to involve computers playing chess or kness in a Schroedinger’s Cattish setup) removes the worry about any question-begging appeal to the libertarian position on free will.  Let’s now move on to the second objection, which Aaron was quick to raise in his comment. 

Objection 2.  We need not resort to the different rules of chess and kness to explain the systematic difference in the movements of knight-wise arranged microparticles in chess and kness.  Take one of the two computers playing chess or kness, according as the quantum object’s spin is up or down.  Whichever game it is playing, it is running the game from a program that is stored in the computer, let’s say in its hard drive.  In the hard drive there will be a set of microparticles arranged chess-program-wise (call this set ‘C’), and another set of microparticles arranged kness-program-wise (call this set ‘K’).  It is the difference in the initial states of microparticles in C and K that accounts for the systematic difference in the movement of knight-wise arranged microparticles in chess and kness.

My Response.  Notice here that the reductive physicalist is transferring the burden of explanation from the laws governing the movement of knight-wise arranged microparticles to the initial states of microparticles arranged C-wise or K-wise.  Let’s dub this move initialization.  Let’s see what this move amounts to in terms of Hempel’s covering law model of explanation.  Let I₁, I₂, …, I_m be statements of initial states of microparticles, let L₁, L₂, …, L_n be statements of microphysical laws, L_C (or L_K) statements of the laws of chess (or kness), and I_C (or I_K) statements of initial states of microparticles in C (or K).  The explanandum, E_C (or E_K), is the rule-governed movements of knight-wise arranged microparticles in chess (or kness).

Where I propose:

            (1)  I₁, I₂, …, I_m

            (2)  L₁, L₂, …, L_n, L_C (or L_K) 

            —————————————-

             \   E_C (or E_K) 

The reductive physicalist instead proposes the following initialization: 

            (1)  I₁, I₂, …, I_m, I_C (or I_K)

            (2)  L₁, L₂, …, L_n 

            ————————————-

             \   E_C (or E_K) 

By this maneuver, the reductive physicalist hopes to explain the movements of knight-wise arranged microparticles in chess and kness with just the laws of microphysics and the initial conditions specifying the positions, velocities, and other properties of all microparticles involved. 

Let me respond to the move of initialization with a counter-move that I call legalization.  Whenever we find a recurring pattern B in recurring circumstances A for non-accidental reasons, we are warranted in positing the existence of a law to this effect: for any A and any B, if A then B.  If the law is not hedged by ceteris paribus clauses, it’s a law of nature.  But it could also be a ceteris paribus law, and there are such laws aplenty in the special sciences, or at levels of reality higher than the fundamental level of microphysics. 

Now, let a C-duplicate be an exact replica of C-wise arranged microparticles (in other words, an atom-for-atom duplicate of the microparticles arranged chess-program-wise in the hard drive of one of our computers in the thought experiment).  By supervenience, this C-duplicate, having the collective base level property of being arranged chess program-wise, must thereby also have the higher level property of being a chess program.  And the chess program must issue in movements governed by chess rules when it is run.  These chess movements, in turn, are subvened by movements of microparticles arranged chess-piece-wise (call these ‘C-movements’). 

Thus we have a recurring pattern of C-movements in recurring circumstances involving C-duplicates.  Since this is a nomic regularity, we may canonize it into law (call it C-Law).  The same legalizing maneuver applies to K-movements issued by K-duplicates, so that there is also K-Law.  So we now have:

            (1)  I₁, I₂, …, I_m

            (2)  L₁, L₂, …, L_n, C-Law (or K-Law) 

            ————————————————–

             \   E_C (or E_K) 

The above argument suggests that the move of initialization fails.  Recall that the reductive physicalist, in Objection 2, tried to shift the burden of explanation from Line (2) to Line (1) of the covering law explanation.  What I hope to have shown in the above argument is that Objection 2 fails to the extent that it relies on initialization.

No doubt the reductive physicalist will point out that C-Law and K-Law are different from the rules of chess and kness.  C-Law and K-Law are different, because they only make reference to arrangements of microparticles giving rise to movements of other microparticles.  In response, though, I wonder whether these laws are not just parasitic on the rules of chess and kness.  I point out also that there are an indefinite number of such laws, as many as there are conventions available or imaginable.  So is it plausible to maintain that there are all these laws on the level of microphysics?  Or aren’t these nomic regularities just shadows cast on the microphysical level of the conventional rules we invent, follow and impose on some higher level of reality?   

The reductive physicalist may respond that C-Law and K-Law are ceteris paribus laws (e.g., they will not obtain if the set of microparticles composing into the computer in question is lacking a proper subset that composes into the CPU), and that these laws can be derived as special cases of universal laws L₁, L₂, …, L_n obtaining at the microphysical level.  This will be the (I hope last) objection, to be discussed in the next post of the series.  But before we go on to that discussion, I ask you to reflect on the following.  These rules of convention, such as those of chess or kness or the traffic regulations, do not obtain as a matter of natural necessity.  These rules are arbitrary and artificial, invented for our own amusement or safety.  And we know them primarily because we create and enforce them, not because we have observed certain patterns of regularity to obtain (except in derivative cases).  Isn’t it remarkably strange, then, to suppose that these rules can be derived as special cases of the laws of microphysics?

As a non-reductive physicalist, I want to maintain the existence and causal efficacy of higher-level objects, properties, and events (that is, at levels higher than the microphysical).  And since I am also a physicalist, I am required to acknowledge the causal closure of the microphysical, i.e., that if a microphysical event has a cause, then it has a microphysical cause.  But there is a tension between these two positions: if higher level entities have non-redundant causal powers, then this leads to downward causation (the microphysical level is open to causal input from higher levels).  Downward causation, in turn, seems to violate the causal closure of the microphysical.  Kim‘s causal exclusion argument and Merricks‘s overdetermination argument exploit this tension in favor of reductive or eliminative physicalism. 

I believe that the tension ought to be resolved without giving up the causal efficacy of higher level entities.  Now I have a thought experiment which convinces me of their causal efficacy.  The basic idea behind the thought experiment is this.  Our social world is rife with conventions, and these conventions are arbitrary.  For instance, it is the convention everywhere to drive on one side of the road (right or left), but each country could just as well have adopted the alternative convention of driving on the other side of the road.  Hence the movements of hundreds of millions of car-wise arrangements of microparticles could have been vastly different.  This seems to indicate that our traffic conventions (in addition to microphysical laws) determine the movements of these microparticles, which is an instance of downward causation.

My thought experiment builds on the basic idea that the movements of microparticles this way or that way depend causally on the adoption of this or that rule of convention.  So, consider the game pieces on the chess board, moved about in accordance with the rules of chess.  These pieces are composed of microparticles, and the movements of these microparticles on the board are microphysical events which have microphysical causes.  Sure, these game pieces are moved about in accordance with the rules of chess.  But given the causal closure of the microphysical, the movements of the microparticles composing the pieces would be fully determined by the laws of microphysics. 

Now here’s the twist.  Take the game of kness, which is exactly like the game of chess, except that the knight can only jump over pawns (I suppose the clergy and the royalty might appreciate this rule).  Whenever we sit down to play a game of chess, we might decide to play kness instead.  And if we decide to play kness instead chess, everything else remaining the same up to the moment of the decision, then the movements of the knight-wise arranged atoms on the board would be systematically different from those in the game of chess. 

So, what accounts for this systematic difference between the kness game we play and the chess game we could have played in this scenario (and in the correspondingly different movements of knight-wise arranged atoms)?  It cannot be the laws of microphysics, since they remain the same regardless of whether we play in accordance with the rules of chess or kness.  It cannot be the initial positions, velocities, and other intrinsic/extrinsic properties of the microparticles at the start of the game, since we have stipulated these to be the same in our ceteris paribus clause, “everything else remaining the same”.  Therefore, the systematic difference can only be accounted for by the difference in the rules of chess and kness, which govern the movement of chess pieces.  These rules, when rigidly followed by the players, dictate the systematically differential movements of knight-wise arranged microparticles in kness as opposed to chess. 

Objection 1.  By now, of course, you will have noticed a gaping loophole in my ceteris paribus clause, which allows us to get around the conclusion just reached.  The physicalist who affirms the causal closure of the microphysical domain will surely insist that everything could not have remained the same up to the moment of our decision to play kness instead of chess.  There would have been differences in the microphysical states underlying the intention to play kness and the intention to play chess.  Moreover, the determinist would add that I am assuming, in that ceteris paribus clause, the libertarian position that free choice uninfluenced by physical causes and influencing physical events is possible.  This libertarian assumption begs the question against causal closure of the physical domain, and renders the argument in the previous paragraph circular.

My Response.  Now, I believe these physicalist responses are red herrings (I especially do not want to mix up the issue of the causal efficacy of higher level entities with the debate between determinists and libertarians on the free will).  The loophole can easily be mended, as follows.  Instead of human players, imagine two computers playing against one another in a room.  These are programmed to move about the game pieces on the board in accordance with the rules of chess or kness.  This eliminates the room for human error.  And let’s rig up the room to a setup like that of the Schroedinger’s Cat: a Stern-Gerlach device measures the spin of a quantum object in a superposition state, and according as the measured spin turns out to be up or down, the computers play chess or kness.  This allows us to get around the issue of human free choice.  Whether the computers play chess or kness depends on the probabilistic outcome of the measurement, everything else remaining the same up to the moment when the measurement takes place.  Then our old conclusion still stands.  The laws of microphysics and the initial states of the microparticles cannot account for the systematic difference in the movements of knight-wise arranged microparticles in chess and kness.  Only the different rules of chess and kness do that.  (There is, of course, a difference in the spin of the quantum object which triggers a game of kness as opposed to chess.  But this only explains why the computers play kness instead of chess.  It doesn’t explain the systematically different movements of knight-wise arranged microparticles in chess and kness.) 

In later posts in the same series, I hope to anticipate and address some further objections that a redutive physicalist is likely to make, and suggest a way of resolving the tension between causal closure and downward causation.  But quite likely I will face insurmountable objections.  So, more half-baked ideas on the way in the next few days!

All phenomena are empty, according to Nagarjuna.  That’s to say, any phenomenon lacks essence or independent existence; it exists not inherently, but only conventionally.  This is how Garfield explains the idea of emptiness of, say, a table:

…it does not exist “from its own side”—that its existence as the object that it is—as a table—depends not on it, nor on any purely nonrelational characteristics, but depends on us as well.  That is, if our culture had not evolved this manner of furniture, what appears to us to be an obviously unitary object might instead be correctly described as five objects: four quite useful sticks absurdly surmounted by a pointless slab of stick-wood waiting to be carved.  Or we would have no reason to indicate this particular temporary arrangement of this matter as an object at all, as opposed to a brief intersection of the histories of some trees.  It is also to say that the table depends for its existence on its parts, on its causes, on its material, and so forth.  Apart from these, there is no table.  The table, we might say, is a purely arbitrary slice of space-time chosen by us as the referent of a single name and not an entity demanding, on its own, recognition and a philosophical analysis to reveal its essence.  That independent character is precisely what it lacks on this view.  (Garfield, The Fundamental Wisdom of the Middle Way, pp.88~9)

As far as I can see, Nagarjuna argues for the emptiness of phenomena mostly by reductio ad absurdum.  For any phenomenon our names designate as a single entity, Nagarjuna purports to establish two claims: that (i) it only exists interdependently with other phenomena, and that (ii) it is mind-dependent and hence conventional.

But quite often (though I don’t think always) Nagarjuna seems to argue for (i) by assuming the truth of (ii), which I take to be one of the desired conclusions of his arguments.  Let me show why I think this, first by parodying the troubling features of some Nagarjunian arguments, and then giving specific examples of such arguments from Nagarjuna himself.

Here’s the parody.  Nagarjuna’s arguments sometimes sound like this (I give a reductio argument):

Assume that husbands and wives exist independently of one another.  Then it follows that there can be husbands without any wives.  But that’s absurd!  It follows from there being husbands that there must be wives.  Therefore, to avoid this absurdity, we must give up the assumption that husbands and wives exist independently.

In the toy argument above, we get the contradiction by relying on the fact that “husband” and “wife” are correlative terms.  But surely this doesn’t mean that the person designated or described by the term “husband” must depend for his existence on the person designated or described by the term “wife”.  The parody argument establishes only dependency in the way we describe things (call this “de dicto dependence”), and not dependency in the way things really are (call this “de re dependence”).  The argument goes wrong in assuming that de dicto dependence (i.e., interconnections in our descriptions of things) establishes de re dependence (i.e., interdependence in the natures of things in themselves).

Once we distinguish between dependence de re and de dicto, it seems that some of Nagarjuna’s arguments establish only de dicto dependence, not de re dependence.  Here are some relevant passages from his Mulamadhyamakakarika (Garfield trans.).  Each of these passages seems to rely on de dicto dependence to establish de re dependence.

Chapter II, verse 4:

For whomever there is motion in the mover,

There could be non-motion

Evident in the mover.

But having motion follows from being a mover.

Chapter X, verse 9:

If fire depends on fuel,

It would be the establishment of an established fire.

And the fuel could be fuel

Without any fire.

Chapter XIII, verse 5:

A thing itself does not change.

Something different does not change,

Because a young man doesn’t grow old,

And because an old man doesn’t grow old either.

Chapter XIV, verses 5~6:

A different thing depends on a different thing for its difference.

Without a different thing, a different thing wouldn’t be different.

It is not tenable for that which depends on something else

To be different from it.

If a different thing were different from a different thing,

Without a different thing, a different thing could exist.

But without that different thing, that different thing does not exist.

It follows that it doesn’t exist.

According to Garfield’s commentary, Chapter X “Examination of Fire and Fuel” is especially important, because it is Nagarjuna’s response to those Buddhists who accept a one-way dependence relation (somewhat like supervenience) of conventional phenomena upon a non-conventional base level of atoms.  Just as the existence of fire depends on fuel but not fuel on fire, so conventional phenomena depend on the basic constituents of reality but not vice versa.  Against these Buddhists, Nagarjuna’s Chapter X is meant to argue for two-way dependence in the case of fire and fuel, thus showing by analogy the emptiness of the basic constituents of reality as well.  But his argument for this, as I’ve suggested, relies on de dicto rather than de re dependence of fuel on fire. 

Does Nagarjuna beg the question then?  That’s how it seems to me.  His arguments seem to beg the question against realists like the Nyaya philosophers, and against Buddhists who accept a one-way dependence relation.  Quite subtly he assumes what needs to be argued for.  He needs to argue that for every X there is a Y such that X is dependent on Y.  In arguing for this he assumes that de dicto dependence of X on Y establishes de re dependence of X and Y.  The assumption in turn presupposes that X is mind-dependent, and this presupposition is just a particular instantiation of the general conclusion that needs to be argued for!

That said, it’s quite likely that I am not giving the most charitable interpretation of Nagarjuna’s arguments.  Perhaps Nagarjuna does have an argument to show, in the Mulamadhyamakakarika or elsewhere, that de dicto dependence can establish de re dependence without begging questions.  But this argument, or the most charitable interpretation of Nagarjuna, escapes me.  If the reader is aware of where my ignorance lies, I ask for enlightenment.