The Engines of the Soul, by  W. D. hart. (Cambridge: Cambridge University Press, 1988. Pp. xi + 190. Price 25.00.)

The Engines of the Soul.

Review Author:
Jim Edwards
The Philosophical Quarterly, Vol. 39, No. 157. (Oct., 1989), pp. 512-515.

Hart constructs a clear and subtle argument for a Cartesian thesis that we are non-material entities who could, logically but perhaps not causally, exist in a disembodied state. The basic argument for this thesis is also due to Descartes: viz., each of us can imagine being a disembodied consciousness; what can be imagined is possible, so we are, in the actual world, non-material entities only contingently lodged in our bodies. Hart's is a modified Cartesianism, for he insists that the non-material self is literally located within the skull, and that a disembodied self which perceived the material world from some particular point of view would be located in the material world at that point of view. Hart requires also that there be psycho-physical causal interactions, governed by principles of the conservation of psycho-physical energy.

The pleasing subtleties of Hart's argument lie partly in his discussion of the analogy between the roles of perceiving and imagining in the epistemology of modal claims. Also interesting is Hart's extended formal exploration of the prospects for psycho-physical causation. Hart starts from a non-Humean conception of causality: causal processes are marked by principles of conservation, some localized quantity being conserved along a causal chain, as momentum is conserved in standard billiard table examples. Science has developed the more general conception of energy to explain causal chains which span different kinds of phenomena, such as heat producing mechanical work in a steam engine. Some further generalization is called for by psycho-physical causal transactions.

Following the pioneering work of von Neumann and Morgenstern, Hart gives an abstract, axiomatic account of quantity. The various amounts of a quantity objects bear are represented by a function mapping those objects onto real numbers. Hart formally proves that his axioms are sufficient to ensure that any two such functions can be transformed one into the other by change of zero point and unit interval, as with the Celsius and Fahrenheit temperature scales, so that we may regard any function satisfying Hart's axioms as a measure of a real quantity. The stage having been set, beliefs and desires are to be exposed as such quantitative states so that conservation laws may proportion quantities of beliefs and desires to physical quantities in psycho-physical causal chains.

Hart's attempt to regiment degrees of belief and desire to suit his axioms proves instructive. For example, one axiom requires that our preferences be transitive, but the psychological facts seem otherwise. Imagine a thousand glasses of beer in a row, the extremes containing one and two pints respectively and adjacent glasses differing by one thousandth of a pint. Knowing all this and asked to choose between the extremes we may choose the glass with two pints, but asked to choose between any two adjacent glasses we may be quite indifferent, so our preference rankings are not transitive. Facing this example Hart claims that his axioms describe an ideally rational agent who really does want more beer rather than less. Hart's response misses the point of his own example. There is nothing irrational about preferring two pints to one whilst being indifferent to choices between adjacent glasses, nothing irrational in wanting more beer to less only if the difference is substantial, whilst not wanting more beef to less if the difference is orally imperceptible.

In general, when our beliefs or desires seem ill-fitted to Hart's axioms, Hart thinks his axioms 'a theoretical simplification or idealization akin to rigidity or the absence of friction in elementary mechanics' (p. 104). One sympathizes with the Galilean predicament of a pioneer of the psycho-physical, but the comparison is not apt. Real bodies in motion, only so rigid and moving against friction, have nevertheless real quantities of mass and velocity which obey conservation principles. The theoretical idealizations of elementary mechanics merely simplify by omitting some relevant quantities active in the actual causal chains. But Hart is aiming to show the intelligibility of actual, non-idealized, psycho-physical causal chains. So he needs to show that actual, non-ideal, beliefs and desires are quantified. To this end it matters not that the beliefs and desires of a mythical, idealized thinker are quantified. He needs to show that adding in other factors, the frictions and bendings of real life, does not destroy quantification nor adherence to conservation principles. So he should not admit that real beliefs and desires deviate from his axioms of quantity properly applied.

Hart claims to prove that we are non-material psychic substances. His argument is that each of us can imagine being disembodied, hence it is possible for each to be disembodied, and so each is a non-material substance. Let me agree with Hart that when asked to imagine myself being disembodied I am imagining something, rather than merely seeming-to-imagine something. And let me agree that I am imagining myself in some state, rather than imagining something that seems to be myself in some state. But am I imagining myself being disembodied rather than myself seeming-to-be-disembodied ? Surprisingly Hart does not ask this question, although he does explore at length related questions about imagining and possibility. But Hart does provide the materials to furnish an answer. Hart argues that the relation between what is perceived and what is actual is epistemologically parallel to the relation between what is imagined and what is possible. Accepting the parallel, we take our perceptions at their face value unless there is some theoretical reason for downgrading them, so I should take myself to be imagining myself being disembodied unless there is some theoretical reason to downgrade my introspective act to one of imagining myself seeming-to-be-disembodied. There need be no difference in the images involved.

Hart has in effect conceded that there is more to imagining than meets the inner eyeball. His highly abstract discussion of quantities of belief and desire was, he claimed, necessary to support his claim that he can imagine being disembodied. But clearly the mathematical detail of that discussion is not and cannot be incorporated into imagery - nothing about an image can record that energy, conceived of as a quantity with real-number continuity, is exactly conserved in precise proportion along psycho-physical causal chains. So his discussion of quantity is best interpreted as a theoretical warrant for interpreting his images as images of disembodiment. Had Hart discovered some mathematical obstacle to quantifying beliefs or desires, he would have shown we were only imagining ourselves seeming-to-be-disembodied.


However, Hart's illuminating discussion of causal chains does not, I think, provide a warrant for interpreting my images as myself-being-disembodied rather than as myself-seeming-to-be-disembodied. His causal speculations are highly speculative, as he repeatedly admits (for example, pp. 130, 152). Suppose those speculations are, in fact, wrong. If they are wrong we are only imagining seeming-to-be-disembodied, not being disembodied. I am not claiming that we cannot imagine ourselves in any contra-causal circumstances at all. But I cannot imagine myself and the causal laws which most directly and immediately affect me radically altered - whether or not I realize that they have been altered.

Take a parallel case. If we ask why atomic numbers rather than, say, atomic weights determine elementary kinds, the obvious answer is that knocking a proton off a nucleus radically changes the properties of an atom, transmutes it into a new element, whilst knocking a neutron off produces a relatively minor change of properties. It follows, I think, that if we imagine a possible world in which the causal laws are so different that charge is not the principal determinant of differences of behaviour between things but mass is, then the division into elementary natural kinds in that world will be by mass and not charge. Similarly, if the causal laws which most intimately affect me are radically different in an imaginary construct from those which actually obtain, whether I know this or not, then the imaginary construct cannot include me, a creature of a kind determined by whatever causal laws actually obtain. The best I can do to include myself is to imagine a world in which I seem to be disembodied and in which those radically different causal laws seem to hold sway.

Hart would not agree. Even though we know that it is not causally possible for gold to dissolve in water, Hart argues that we may be able to imagine a piece of gold dissolving in water by imagining a world with radically different causal laws. And even, 'we can imagine a rabbit [NB not merely a rabbit look-alike] popping into being from nowhere in an otherwise empty universe; so we have reason to think that possible' (p. 46).

On the other hand, Hart's causal speculations may be correct, and hence we really are imagining ourselves disembodied, and thus are non-material creatures. But whilst we do not know what causal regime actually holds sway, Hart's Cartesian argument is blocked because we do not know whether we are imagining ourselves disembodied or imagining ourselves seeming-to-be-disembodied.

University of Glasgow jim edwards

Stable URL: