|Home Page|Publications|Time's Arrow and Archimedes' Point|

David Deutsch is an Oxford theoretical physicist. He is best known as one of the pioneers of "quantum computing", which seeks to exploit the peculiarities of "quantum interference" to perform—in principle, and perhaps in practice—computations which could not be performed by classical means. However, Deutsch’s interests extend well beyond his home field. In The Fabric of Reality he defends radical conclusions about matters seemingly far removed from his home discipline.

Given the ambitious, trans-disciplinary nature of Deutsch’s project, and the fact that his book is aimed at the popular science market, the obvious comparison is with his Oxford colleague, Roger Penrose. Deutsch aims even higher than Penrose, however. Where Penrose argues for a novel connection between three of the hard problems of contemporary science (quantum mechanics, gravity, and consciousness) Deutsch goes one better, and offers us a four-fold synthesis: quantum mechanics, computation, evolution and Popperian epistemology.

Thus Deutsch has set himself a very ambitious target. Too ambitious, in my view, for the effects of magnification are just what one might expect: details get lost, cracks get larger, and key points go out of focus. And where there is room for doubt, Deutsch has an annoying tendency to err on the side of confidence. So for philosophical readers familiar with Penrose’s clear and carefully-argued books, The Fabric of Reality is likely to be a disappointment.

One example: Deutsch maintains that quantum interference effects can only be explained by the many worlds view (more on this argument later) and concludes that quantum mechanics gives us very strong evidence for the existence of what he calls the "multiverse"—the set of all universes allowed by the laws of physics. He goes on to argue that the multiverse makes sense of some notions which would be meaningless without it: free will, causation, and counterfactual conditionals. Concerning the last, for example, he says:

Suppose we agree. What is Deutsch himself to make of the counterfactuals we need to discuss the consequences of the laws of physics themselves, such as: "If charge were not conserved, the world would be a very different place"? The multiverse is no help, for by definition, it contains no worlds in which the laws are different. Deutsch’s view thus implies that these essential claims about how things would have been had the laws of nature been different "have no meaning".

The moral is that if we want to give truth conditions for counterfactuals in terms of possible worlds, we need the set of logically possible worlds, not the set of physically possible worlds. Indeed, Deutsch’s whole multiverse probably counts as a single possible world in David Lewis’s account, to which Deutsch later compares his own view:

In fact, what Lewis postulates is not a multiverse theory, in Deutsch’s sense.

The aspect of The Fabric of Reality of most interest to philosophers is likely to be Deutsch’s account of what he sees as the physical evidence for the many worlds interpretation of quantum mechanics. Deutsch thinks that the case for this view is extremely strong. His presentation of the crux of argument, as he sees it, is admirably non-technical and clear.

In Deutsch’s version, the argument involves a comparison of two-slit and four-slit versions of the familiar experiment involving photons. Somehow, opening a second pair of slits affects the photons that pass through the first pair of slits, preventing them from reaching portions of a screen on which they are otherwise detected. Unless we take refuge in instrumentalism, we must acknowledge that a real effect demands a real cause. And when we investigate the nature of that cause—taking account, for example, of the fact that it may be blocked by exactly the same class of things that would block "real" photons—we find that to all intents and purposes, it does comprise "intangible" photons, detectable only via their interference with the photons we detect by other means.

In effect, then, Deutsch thinks that opponents of the many worlds view face a dilemma: Either their view does not provide a realist explanation of the interference phenomena; or it amounts to the many worlds view, by another name. For example, he mentions the hidden variable theory proposed by David Bohm, "in which a sort of wave accompanies every photon, washes over the entire barrier, passes through the slits and interferes with the photon that we see." (p. 93) Deutsch notes that "Bohm’s theory is often presented as a single-universe variant of quantum theory", but says that this is a mistake:

However, this kind of point cuts both ways. If God is love then I am a theist, for I believe in love; but this way of getting me to accept theism doesn’t look like a good bargain, from a traditional theist’s point of view. (I owe the example to Geoff Sayre-McCord.) Similarly, a Bohmian might say that if believing in the reality of the wave function is all it takes to believe in the multiverse, then of course she believes in that—but wonder what all the fuss was about. A Bohmian can say this because she has a distinct story about what it takes to be real in the sense that the photons we observe directly are real. Bohm’s view admits a single particle with a definite trajectory, as well as the multi-branched wave that guides it. Accordingly, a Bohmian will say, there’s a big difference between our universe and all the others. Our universe alone comprises real particles of this kind.

Presumably, Deutsch would object that it is a disadvantage of Bohm’s view that it admits these two kinds of reality, and that the second is inelegant and unnecessary. The Bohmian would reply that on the contrary, the "definite trajectory" kind of reality does a crucial job—without it, there is nothing to explain why what we experience is a determinate classical world. It is not clear that a supporter of the multiverse can adequately respond to this, without having to abandon the claim that the difference between the multiverse and Bohm’s wave is merely terminological.

In any case, it should be apparent that even if one shares Deutsch’s strongly realist sympathies, the case for saying that any acceptable explanation of quantum interference will be equivalent to the multiverse view is far from being as clear cut as he takes it to be. A convincing case would need to lay out and defend the assumptions of the argument with a great deal more care. To finish, I want to mention a further crucial assumption of Deutsch’s version of the argument, to which again he fails to call attention.

In essence, the argument goes as follows: "With two slits open, some photons arrive at a point X. With four slits open, no photons arrive at X. Opening the second pair of slits thus affects at least those photons which would otherwise have arrived at X." The hidden assumption is that the initial state (or distribution of states) of the photons is the same in both cases. Without this assumption, it is a trivial matter to reproduce the two-slit and four-slit distributions, with objects of many kinds. We could do it using undergraduates and a wall with four gates, for example. When two gates are open, we give the undergraduates a graph of the two-slit distribution and the instruction: "Choose a gate a random; go through it, and then choose your target on the second wall at random, using the probabilistic weights shown on this curve." When four gates are open, we use the same instruction, but give them a graph of the four-slit distribution.

Of course, this only works because we know in advance how many gates will be open, and set the initial state of the undergraduates accordingly. Obviously, the photons couldn’t "know in advance" what sort of screen they are to encounter, and choose their initial state accordingly.

Obviously. But why, exactly? We take for granted that the state of the photons does depend on the fact that they have a particular history—we might get very different results if our photons didn’t all arrive from the same source, for example. Why do we assume that their state cannot also depend on their "history" in the other direction—on what kind of source they "come from" in the future? Unless we do assume this, we have no right to assume that the initial state is the same in the two-slit and four slit cases, and hence no right to assume that something must interfere with the four-slit photons, in order to prevent them reaching the point X.

Of course, to point to the existence of this loophole is a long way from showing that it may be exploited. Equally, however, to point out that the loophole has not yet been successfully exploited is a long way from showing that it cannot be exploited—which is the conclusion Deutsch needs, if he is to convince us that the multiverse provides the only viable realist interpretation of quantum phenomena.

In sum, Deutsch gives his readers a good sense of why he and many other physicists find the physical evidence for the many worlds view to be compelling. But he fails to show that the evidence should still seem compelling, in the light of a more careful examination of the underlying argument. That task requires a nose for fine distinctions and a respect for logical rigour and clarity, and these are not the talents on display in The Fabric of Reality.

Huw Price
School of Philosophy
University of Sydney