The third way

Considerations of anthropic fine-tuning seek to explain the appearance of an otherwise puzzling link between the universe on one hand and life on the other. Why should there be a connection? What does the Universe know about life? What do the laws of physics care about consciousness?

The most obvious way to establish a link between life and cosmos is to postulate a 'life principle' (or, extending this to encompass observers, a 'mind principle'). Indeed, many scientists have suggested just such a thing. It is often claimed by astrobiologists that life is 'written into the laws of physics' or 'built into the nature of the Universe' [19]. Thus Sydney Fox, in his theory of biogenesis, claimed that the laws of physics and chemistry were rigged in favour of those reactions that lead to life [20]. Others, such as Christian de Duve [21] and Stuart Kauffman [22], have hinted that somehow chemistry favours life and can fast-track matter and energy to the living state.

John Wheeler, in his 'participatory universe' principle, has even claimed something along those lines for mind [23].

Is there any evidence for such a principle? The laws of physics, as we now understand them, do not offer much promise in this regard. The reason is not hard to find. Life is incredibly complex but the laws of physics are, in the algorithmic sense, simple. So life cannot be contained in the laws of physics. Contrast this with another state of matter: crystals. The structures of crystals are determined by the symmetries of the electromagnetic force, and so they are built into the laws of physics. Basic geometry underlies them. Given the laws of physics, the structure of, say, common salt crystals may be deduced from purely geometrical considerations. Crystals are simple and have low information content, concordant with the low information content of the laws of physics. But one could not predict the structure of, say, a bacterium, nor even its genome sequence, from the laws of physics, because the genome has very high information content. It was for that sound mathematical reason that Jacques Monod declared 'we are alone' and 'the Universe is not pregnant with life'. In his opinion, life is just a stupendously improbable accident [24].

The root cause of the difficulty goes back at least to the time of Newton and the deep dualism that pervades all of science: the dualism between eternal universal laws and time-dependent contingent states. Because laws are general, simple, low in information content and unchanging with time, most specific states of matter cannot be built into them. States of matter are generally local, special, complex, high in information content and time-dependent. So the very structure of traditional scientific explanation precludes our finding a direct link between the underlying laws of the Universe (as we at present understand them) and the emergence of an exceedingly specific and peculiar state of matter such as 'life' - still less an even more specific and peculiar state such as 'mind'. Therefore, if we wish to postulate such a link, then the traditional dualism of laws and states must go.

Aristotle did not make a sharp distinction between laws and states. By introducing different categories of causation, and specifically by including final causes, he could speculate on how the Universe might develop in a directed manner toward certain special states. For Aristotle, life was indeed built into the nature of the Universe through final causation. Such goal-directed or purposeful influences in nature are termed teleological by philosophers.

The assumption of a link between laws and product states such as life inevitably amounts to slipping an element of teleology into physics. This is very unfashionable, but I believe it is unavoidable if we are to take life and mind seriously as fundamental rather than incidental features of the Universe. And the bio-friendliness of the Universe suggests that they are fundamental. We need not be as crude as Aristotle, by nailing down the final state in advance and constraining the Universe to generate it; de Duve, for example, has suggested in the context of biological evolution that the general trend (e.g. from simple to complex, from mindless to mental) is law-like, although the specific details are contingent [21]. In my essay 'The physics of downward causation' [25], I have suggested that such a felicitous mix of law and chance might be generalized to cosmology, producing directional evolution from simple through complex states, to life and mind.

Obviously these are just words, whereas what is required are concrete mathematical models. To investigate the basic ideas, I have developed some cellular automaton models with the help of Neil Rabinowitz. Recall that, in a conventional cellular automaton system, one starts with a 1-dimensional array of cells, or pixels, each of which can be in one of two states: filled or unfilled ('on' or 'off'). An update rule is specified that determines whether a given pixel remains on or off, is switched from on to off or vice versa. This rule is based on the state of the near neighbours, and there are 256 possible simple local rules [26]. This system thus mimics the physics of a causally closed system subject to local dynamical laws. An initial state is specified, for example a random scatter of filled cells, and the array is evolved forward in discrete time steps. A variety of interesting behaviour results. Crucially, the conventional automaton retains the ancient dynamical dualism: the update rules are always independent of the states.

As a first departure from the conventional prescription, we decided to start with a random input state and tried switching between two different rules either randomly or periodically. The results of one interesting case are shown in Figs. 28.1-28.3. This features two automata, designated 87 and 90 according to Wolfram's classification scheme [26]. Applied on its own, rule 87 leads to structured, but relatively dull, quasi-periodic spatial structures that move across the array at uniform speed (Fig. 28.1). Rule 90 merely perpetuates the random noise (Fig. 28.2). Thus, individually, rules 87 and 90 do not lead to interesting dynamical behaviour. However, when the rules are interspersed, the story is very different. Figure 28.3 shows the outcome when rule 90 is applied and interrupted every seven steps by rule 87. The upshot is the evolution of a form of organized complexity from disorganized, or random, input. Although there is nothing explicitly teleological in the set-up, a form of directionality - order out of chaos - is discerned. With a bit of experimentation, this rule-interspersion technique can be used to combine order and chaos in a suggestively creative manner, getting 'the

Fig. 28.1. Rule 87 cellular automaton with random initial state. Time runs downward.
Fig. 28.2. Rule 90 cellular automaton with random initial state.

best of both worlds' - the unpredictability and novelty of chaos with the coherence of order. Our results are reminiscent of Parrondo's games [27], in which two games of chance, each of which when played individually have an expectation of loss, when combined can lead to an expectation of gain. Parrondo's games show that, counter-intuitively, two losses can make a win. Figures 28.1-28.3 appear to be a cellular automaton analogue.

Fig. 28.3. Rule 90 interrupted every seven steps by rule 87 cellular automaton with random initial state.

To incorporate fully my 'third way' idea, we must alter the automaton rules so that they depend explicitly on some aspect of the state. To take a very simple example of state-dependent laws, the rule may be chosen to be A if the total number of filled pixels is even and B if it is odd. Alternatively, some statistical measure, such as the entropy or complexity (defined by some prescription) may be used as the discriminator of the rules. Whatever choice is made, the behaviour of a group of pixels now depends not only on the state of the neighbouring pixels, in analogy with conventional physical laws, but on the global state too. This is therefore an explicit form of top-down, or whole-part, causation [25]. Although our work is at a preliminary stage, the hope is that simple mathematical models might capture the elusive notion that certain complex states are favoured by acting as attrac-tors in the product-space of states and laws. This idea could be placed in a restricted multiverse context by considering how some universes, or regions thereof, generate their bio-friendly laws in an evolutionary sense, and thus become observed. So biology does not actually select a pre-ordained universe; rather, physics and biology co-evolve under the action of a (precise) principle operating at the multiverse level, in such a manner that teleologi-cal behaviour emerges. So this is a theory in which life and mind, goal and purpose, arise in a law-like manner from a dynamic universe (or multiverse). The key feature is that there is a causal link between laws and product states

(in contrast to Darwinian evolution, where mutations and selection events form causally disjoint chains). Thus life is neither a statistical fluke in an indifferently random set of laws/universes, nor is the Universe designed in an ad hoc way for life. Instead, life and mind, laws and universes, are common products of an overarching principle.

If I were to pick a symbol to characterize this set of still rather woolly ideas, it is that of a self-consistent, self-supporting loop. It has some elements in common with Wheeler's idea of a loop in which nature and observer are mutually enfolded [23]. I have described it as a 'turtle loop' in the context of the famous 'tower of turtles' metaphor [28].

As a final illustration of an implicit loop, consider the fact that the mathematics describing the underlying laws of physics is a product of the human mind. The mental realm occupies a conceptually higher level than the physical realm of particles and fields to which this mathematics applies. Why should something created at this higher level apply so famously well [29] to the physical realm? Why should 'software' apply to 'hardware' ? More specifically, the concept of what constitutes a computable function (software) is based on the idea of a classical Turing machine (hardware). As stressed by David Deutsch [30], the existence of such a physical device depends on the specific nature of the laws of physics. Thus the concept of computability depends on what the physics of the particular world allows to be computed. So the laws of the Universe permit the existence of physical systems (human beings, Turing machines) that can output the mathematics of those very same laws. This remarkable self-consistent loop is by no means guaranteed [28]. (It also constitutes a further example of why human beings are more than mere observers, which I considered in Section 28.3.3. Human beings are also 'computers'.) There could be many universes with computable laws that do not admit physical systems which can actually output the computable functions describing those laws. Or there could be universes with non-computable laws [31]. Since there is an intimate connection [12] between Turing machines and self-reproducing machines (i.e. life), we glimpse a link between life and laws.

Was this article helpful?

0 0

Post a comment