Оtologies of the high-level sciences

The high-level sciences are ontologically pluripotent. They make great use of physical individuation, that is, individuation according to the spatiotemporal wedding cake conception: even in EPA, the bearers of probability are typically physically discrete entities such as molecules, animals, or people, and the ultimate aim of models is to track the statistical movements of such things.

Yet at least as important are what you might call distributed ontologies, that is, individuations into entities—things and reifications of the tendencies of things, such as causal dispositions and enion probabilities—whose presence is determined by configurations of fundamental physical facts that overlap, so that the same facts contribute to many such entities. Enion probabilities are this paper’s paradigms of distributed ontology.

The entities distinguished by the first, spatiotemporal kind of ontological decomposition—the enions themselves, the lynxes and hares—are physically independent but interact with one another in many complex ways. The entities distinguished by the second, distributed kind of ontological decomposition—the enion probabilities, such as the probability of a certain hare’s death over the course of a month—are physically overlapping, but stochastically independent and therefore easily aggregated.

These two ontological schemes are not rivals, but rather work together within a single modeling technique in population ecology, serving up a compositional theory that solves the problem of aggregation. The wedding cake ontologists are right to think that spatiotemporal individuation has been essential to creating compositional theories of the high-level sciences, but wrong in thinking that it has been sufficient. To tackle the sciences of complex systems, we need what is, in a mild sense, ontological pluralism.

Can the same be said of other kinds of compositional theories? Let me give you two examples.

The first is spectral analysis in wave theories of various aspects of nature. In the high-l evel sciences, there are sound waves, ocean waves, waves on the strings of musical instruments, seismic waves, and more—where in each case, the wave is a movement of an underlying medium. In spectral analysis, the medium’s movement is decomposed into waves of different frequencies, as when the motion of a vibrating string is decomposed into a tone and various overtones. These waves coexist in the same medium—in the same volume of air, or earth, or water—and indeed in the same movements of that medium; consequently, the fundamental-level matters of fact on which different waves in a spectral decomposition depend are largely identical. The waves form a distributed ontology.

For the predictive and explanatory purposes of many wave models, the only aggregation required is the addition of the effects of these different frequencies, which is accomplished by the straightforward process of linear superposition. The wedding- cake alternative, in which a model keeps track of the movements of different parts of the medium—different segments of a vibrating string, or different volumes of air or water—is far more difficult to implement. A distributed ontology brings compositional modeling within reach.

(Our theories of the forces that come together to create waves are, however, often sensitive and combinatorially complex. As a result, wave theories will, for certain predictive purposes, suffer from a complexity explosion. That is why quantum chemistry is computationally so difficult.)

Another, more speculative, example is belief/desire psychology. Most of us would guess that the facts underlying beliefs and desires—the facts that make it the case that I believe that there is rabbit for dinner or that I desire to wash it down with a glass of wine—are to some extent distributed across the brain in an overlapping way. The propositional attitudes comprise a distributed ontology.

Belief/desire psychology is also a remarkably effective compositional theory of thought (Dennett 1987). The principles of composition are quite familiar to us, but the relation of the whole to the underlying facts remains, for now, opaque. It seems that belief/desire psychology solves an aggregation problem by way of a distributed ontology, then, but we cannot as yet be sure.

Some philosophers have suggested that beliefs and desires not be taken ontologically very seriously at all (Churchland 1981; Dennett 1987), in part, I would guess, because their distributed nature lends them a certain insubstantiality, a lack of proper placement within the great wedding cake of science. Perhaps contemplation of the role of distributed ontologies elsewhere, in wave theories and in EPA, can solidify the attitudes’ status, both scientific and metaphysical.

< Prev   CONTENTS   Source   Next >