Kristor has published an important piece on the ontological priority of wholes to parts. I had been fumbling around with related ideas, but had not gotten nearly as far. I still haven’t in fact. Kristor seems to possess a sort of metaphysical vision that I lack. Or maybe he just doesn’t show us his work. Either way, it often takes me much laborious thinking to cross the distance of one of his “therefore”s. What follows will be longer and cover less ground than the above post, but perhaps it will help others like me whose minds need to move in small steps. There are some considerations that strongly favor the ontological priority of parts, what I shall be calling “atomism”, but I will argue that they leave some room for the reality and even priority of wholes.
Two things I’ve gathered from modern analytic philosophers. First, causality isn’t really in the laws of nature as we know them, the equations allowing inference in either direction. Second, in emergent orders, the small scale causes the larger scale, i.e. parts are ontologically prior to wholes. The temperature of a gas is caused by the kinetic energy of its molecules, not the molecular motion caused by the macroscopic value of temperature. My thoughts are caused by the firing of neutrons, rather than my thoughts causing those neutrons to fire. Now, a mischievous non-philosopher like me might wonder, if we are to ontologically flatten correlations, always putting the left hand and right hand side of an equation on an ontological level–anything else being premodern mysticism–why this should not also apply to emergence. This has tended to be my own way of thinking: let us drop ontological priority altogether, abandon our atomist dogmatism.
One might ask if we could even reverse the direction of ontological reduction, pose a wholist dogmatism that makes the whole the entire cause of its parts. This runs into two huge obstacles which together constitute the strengths of the atomist position we wish to overcome. The first problem is that of degrees of freedom: the independent variables needed to specify the state of a system. The whole will have one set, the collection of parts another. The map between them is seldom one-to-one, but usually one-to-many: many microstates correspond to the same macrostate, a degeneracy quantified by the entropy. Thus, a parcel of gas can be described by a few thermodynamic variables, whereas the description of the state of its constituent molecules, in terms of classical phase space or (even worse) a multi-particle quantum state vector, contains enormously more information. One would expect that if A is the cause of B, the degrees of freedom in B should be able to be explained through A, while the reverse may not be true. Generally, the properties of wholes can be obtained from those of parts by averaging/coarse-graining away microscopic degrees of freedom, but there is no inverse process to recover the remaining microscopic information from the properties of the whole. There can’t be, or else the microscopic system actually would not have more degrees of freedom; it would have the same number, and the remaining variables would not really be independent.
Well, could we be daring and assert that this is true? Suppose an isolated collection of molecules actually doesn’t have a definite state vector (not just isn’t in an eigenstate of this-or-that observable, which is old news). All the real information is in a density matrix with coefficients set by macroscopic variables and equilibrium conditions. This won’t work either, because the atomist can now pull out his second killer argument. “You go around monkeying with quantum mechanics because you want to say that tables and chairs are more real than protons and electrons, but watch out–it doesn’t end there. You are part of an ecosystem, of a planet, of a galaxy. Logically, the cascade ends only with the universe itself. If no degree of freedom is real except what affects the overall state of the universe, than your own existence or nonexistence is inadmissible as real information. You don’t exist!”
So, atomist reductionism might work, but a “retaliatory” reduction in the other direction won’t. However, the above work has not been wasted, because it has set us up to pose a more modest wholism. Say we have whole A with N degrees of freedom (call them a_1, a_2, a_3, …, and denote the whole set by a_i) made of B, the set of all A’s parts, having M>N degrees of freedom (call them b_i). One can determine a_i from b_i. One cannot determine b_i from a_i, but one can constrain b_i from a_i. Given a set of a_i, the admissible b_i must lie on a subset, some sort of hypersurface, in the state space of possible b_i. We have thus divided the b_i space into equivalence classes [a_i], and there is (by construction!) a one-to-one map between a_i and [a_i]. We now have the symmetry we need, and we can appeal to our prior metaphysical commitments to decide which way, if either, to assign the direction of causality. Atomism would say that b_i determine [a_i] determine a_i. Moderate wholism would say that a_i determine [a_i], while the remaining degrees of freedom on the hypersurface [a_i] are determined by the parts, from their remaining intrinsic actuality.
In fact, there is nothing artificial about this way of thinking. As an example, we may say that our gas has a determinate total energy, so that the state of the molecules must lie on a particular surface in phase space; this is just what we do in the familiar microcanonical ensemble. Our atomist prejudices lead us to assume that this energy is an effect of the individual molecules’ independent motions, but taking macro-quantities as fixed from above is actually a better description of what one does mathematically in this and the other ensembles. As another example, it is quite natural to separate out center of mass motion from the relative motion of parts. Neither “determines” the other. By Galilean invariance, we may say that the former is determined at a wholistic level with no violence to the evolution of the remaining, microscopic degrees of freedom. (A possible implication: variables at the macro-level may often appear as symmetries / conserved quantities at the micro-level.)
To have a meaningful whole, it must be possible to describe the evolution of the whole strictly on its own level. The micro-level must average / renormalize away. Thermodynamics and hydrodynamics are only the simplest cases of this. Biology and psychology also have their proper orders of intelligibility. I can identify reasons for my actions. On the other hand, my neurons are controlled by the laws of physics. Is my reasoning an illusion? Are my neutrons pushed around in violation of physical laws by psychic forces? Both of these are implausible. However, presuming my brain is a properly-ordered whole, certain degrees of freedom of my neuronal state will “factor up” to form my macroscopic mental state, whose evolution only depends on its prior states and not on where in particular my neurons are in their hypersurface of possible states given my macroscopic mental state. Thus, although my brain may be more than just a computer, moderate wholism can apply even to computers. One can understand a program from its own logic without having to think about electrons moving in semiconductors. That doesn’t mean that the algorithm is psychically pushing those electrons around. The electronic degrees of freedom that don’t “factor up” don’t determine the ones that do. They’re separate degrees of freedom. Similarly, it’s no problem that I don’t feel particularly constrained as a part of the universe as a whole because my own state matters so little to its.
I have not argued that moderate wholism is true, only that it could be true. It would break decisively with atomist metaphysics in that wholes would not fully derive from their parts. As causally prior where their realm intersects with that of their parts, wholes would have their own distinct existence. They would not just be a coarse-grained description of a micro-world considered to be ontologically prior in a categorical sense.
Which I think has something to do with what Kristor was saying.