Is the universe significant?

In this paper, I consider the question of what it means to claim that the universe is significant, and the implications of this debate for a particular age-old argument, the Design Argument.

In section 1, I show that the success of the Design Argument depends on making this claim. In section 2, I show how the claim is very problematic. In section 3, I describe one way in which the claim could be revised so that the problems can be avoided. In section 4, I defend this revision from two objections. Finally, in section 5, I briefly discuss the implications of the revision for the Design Argument.

1. Refining the Design Argument

The Design Argument is an argument which has frequently been used in support of the claim that God exists. More precisely, what the argument shows, if it is successful, is that there is a purposive agent to whose acts we can attribute various aspects of our environment.

There are several versions of the argument. One traditional version runs:

Of course, most Design Arguments continue with the claim ‘…and this agent is God’. In this paper, I will not be concerned with this final step; I will simply be concerned with the part of the argument whose conclusion is that we should prefer the ‘purposive agent’ explanation to the ‘chance’ explanation.

The traditional version has been heavily criticised in recent centuries, but there are other versions which are not open to the same kinds of criticisms because they appeal to different kinds of evidence. For instance, more modern versions substitute ‘the universe’ for ‘life on earth’; that is, these versions point to certain characteristics of the universe, rather than about life on earth, as phenomena that need explaining.

One such modern version, which I will use for discussion in this paper, might be stated as follows:

Although this is a familiar version of the argument, I have called it a Crude Argument for a good reason. Probably no writer has ever used it in exactly that formulation, because it is very weak.

It is weak because its underlying premise is that all unlikely outcomes should be attributed to a purposive agent rather than to chance. We would be making a mistake if we thought that this premise was sound. To illustrate the fault, consider Roy’s Mistake:

Clearly, Roy is making a mistake. True, it was immensely unlikely that Roy would win. But someone had to win, and Roy’s name was just as likely to be drawn as anyone else’s. It is a mistake for Roy to conclude, just because he won, that the outcome of the lottery had been rigged. If this
conclusion were justified, then parallel reasoning would lead us to infer that every lottery ever held had been rigged in exactly the same way.

So, clearly, the premise on which the Crude Life-Containing-Universe Version is based must be revised. We must add another criterion to the conditions for special explanation: significance. For an outcome to need a special explanation, it must be more than just unlikely; it must be both unlikely and significant.

The plausibility of this extra condition can be demonstrated by considering Rory’s Mistake:

Why is Rory making a mistake? Well, if any one of the 999,999 incredibly rich people had won, that outcome would not have needed a special explanation. But the winner turned out to be poor, starving Rory. The odds of selecting the one entrant who desperately needed the money were a
million to one (well, 999,999 to one). This likelihood is so low, and the outcome is so significant, that we should very strongly suspect that the lottery has been rigged – that the outcome was somehow forced so that Rory, the one person who really stood to benefit from winning, actually has won. In this case, we should prefer the ‘purposive agent’ explanation to the ‘chance’ explanation.

Some writers have questioned this reasoning. They would object that Rory is not making a mistake (see, for instance, Scriven 1966, pp.121-124). I believe that these objections have been answered, most thoroughly by Leslie (1989). For the remainder of this paper, I will assume that these answers are correct, and that Rory’s Mistake is a mistake.

To take into account Roy’s Mistake and Rory’s Mistake, we should rephrase our argument as follows:

2. Criticising the Refined Argument

I have shown that the Crude Life-Containing Universe Argument must be revised to avoid conclusions that are too strong (Roy’s Mistake). This revision took the form of adding an additional premise to create the Refined Life-Containing-Universe Argument.

However, this opens the door to a new objection – one which simply denies the added premise:

To meet this objection, we must claim that a life-containing universe is in fact significant where the alternatives (lifeless universes) are not. In this section, I will discuss how this claim might be supported.

Perhaps it seems just obvious that a life-containing universe is significant. If so, we might try to communicate this intuitive obviousness by describing the alternatives. If the universe was not lifecontaining, it would be a universe devoid of life, perhaps because it collapsed in on itself shortly after its creation; or because its constituent parts flew apart too rapidly to allow any interaction of matter; or because the energy level in it was so low that chemistry never took place (see, for instance, Parfit 1998; Rees 1999).

Or we may claim that it would be possible for there to be no
universe at all. All in all, to parallel Dawkins (1986, p.9), we could point out that there are countless billions more ways for a universe to be lifeless than there are to be life-containing; virtually all of them are dull, cold and empty. When we consider these alternatives, how can we doubt that a lifecontaining universe simply is significant in a way that these are not?

The problem with this line of reasoning is that significance is a subjective property, and therefore what is significant from one viewpoint may not be significant from another. Remember Roy’s Mistake: of course, Roy’s winning was significant from his own viewpoint, but this was not enough
to demand an explanation other than chance. By the same reasoning, of course we think that a life-containing universe is significant. But we are life-forms in that universe, so we are not in a position to make any objective judgements. To insist that an aspect of the universe cannot be explained by chance simply because it seems significant to us would be to make the cosmic equivalent of Roy’s Mistake.

We might try to find objective ways in which this universe could be regarded as significant. We might say that the universe, by virtue of being life-containing, is highly complex and its matter is highly structured. (“An earthworm’s brain is more complicated than a lifeless galaxy” – Parfit, 1998.) As a rule, lifeless universes notably lack this kind of intricacy. Again, we might argue that this universe is relatively stable, while most other possible universes are either radically unstable or rapidly self-destructive. And many other characteristics might be singled out for this purpose: we might point out that, by virtue of being life-containing, the actual universe contains more activity, consciousness, moral reasoning or whatever than its non-actualised lifeless rivals would contain. For the most part, these qualities are objective ones – they can be identified and measured without the need for value-judgements. Is this how the universe can be regarded as significant?

But this suggestion fails, too. These criteria (complexity, organisation, stability, activity, consciousness, moral reasoning) may well be objectively measurable. But, in laying down these particular criteria as ‘what makes a universe significant’, we have already made our value judgement. Why should it be these properties that indicate a universe’s significance, and not other properties possessed in abundance by non-actualised lifeless universes – simplicity, emptiness, chaos, speed of decay…?

We may indeed be able to demonstrate objectively that matter in our actual universe is more complex and highly structured (for instance) than in most or all other possible universes. But this would only highlight a difference between our actual universe and other possible ones; it would not imply anything about the significance of that difference. We can objectively demonstrate a difference between X and Y, but we cannot objectively demonstrate that the difference is significant. In Roy’s lottery, no doubt all the potential winners were different in many ways from one another, but that was not enough to allow us to attribute the outcome to anything other than chance.

Significance is not an objectively measurable quality. That is, every claim ‘X is significant’ can be followed with the question, ‘Significant to whom?’, or ‘Significant with regard to what?’. So it appears that we must grant the objection. We cannot claim that a life-containing universe is simply significant in an objective sense; all we can claim is that it seems significant to us. And we have already seen that this is not enough for us to conclude that an explanation other than chance (such as a ‘purposive agent’ explanation) is needed. If we did jump to that conclusion, we would be making the cosmic version of Roy’s Mistake.

3. A useful sense of ‘significant’

In section 1, we showed that the Crude Life-Containing-Universe was faulty because it proved too much. In particular, it said that any unlikely outcome suggested the intervention of a purposive agent, which led us to make Roy’s Mistake. We introduced a ‘significance requirement’ into the argument to exclude such cases: we said that only significant outcomes required explanations other than chance.

But in section 2, we saw that the significance requirement can never be suitably met for the universe – we can never show that the universe is significant in the right way. At most, we can only demonstrate the universe’s subjective significance, and we know from Roy’s lottery that subjective significance is not enough.

This has left us with a problem. While our Crude argument proved too much, our Refined Argument now proves too little. Its ‘significance requirement’ is useless because it can never be met.

As a result, we now find it hard to justify why Rory’s Mistake is really a mistake. If the outcome of Roy’s lottery was not suitably significant, why should the outcome of Rory’s lottery be suitably significant? Surely, all that can be said about the outcome of Rory’s lottery is that it, too,
seems significant to us. Why should the situation in which one poor man is saved from starvation be more significant than the situation in which an already wealthy man becomes wealthier? As before, the reply that the former result has some intrinsic quality, such as moral value or equality, does not help, because the question can then be asked: why should such intrinsic qualities be more significant than any others? What is significant, in the relevant sense, about moral value or equality? The best we can say is that morally valuable outcomes such as Rory’s seem significant to us. But to say that this is the reason we should infer the existence of a purposive agent is to make Roy’s Mistake.

There is, then, a glimmer of hope for the Refined Life-Containing-Universe Argument. We have rejected it on the basis of an objection. But it now appears that the objection in question drives us to too strong a conclusion in the other direction, because it forces us to claim that the outcome of Rory’s lottery does not require any explanation other than chance either. This is false, so we should dismiss the objection.

There is, in fact, a useful sense in which the outcome of Rory’s lottery can be said to be significant: it is significant to the benevolent being which we postulate as the purposive agent who rigged the lottery. In other words, it is plausible to think that a poor man winning the lottery is the kind of result that a benevolent lottery-rigger might want to achieve.

Generalising, I suggest that the ‘significance requirement’ in any Design Argument is not useless after all. There is a useful sense in which it can be interpreted, namely:

When I say that X is ‘within the nature and power’ of an agent, I simply mean that X is the kind of thing that agent might plausibly be expected to do and would be able to do. So what this says is simply that any outcome is evidence for the existence of a particular agent if the outcome is of the kind that the agent in question might plausibly have brought about.

This is why the outcome of Roy’s lottery is not evidence for the existence of a lottery-rigger. There is nothing about Roy’s outcome that picks it out as the kind of outcome that a lottery-rigger might want; Roy has no distinguishing characteristic that picks him out as the kind of person that a lottery-rigger might want to cause to win.

But Rory’s lottery is different. There is something about Rory’s success that picks it out as the kind of outcome that a lottery-rigger might want. Roy’s success is a morally good outcome, so, if the lottery-rigger was benevolent or interested in furthering morally good ends, we might plausibly expect him to want to produce exactly the result that has actually occurred.

It will help to strengthen this principle if we introduce one more example:

The circumstances in Troy’s lottery are exactly the same as in Roy’s, apart from one difference. In Troy’s lottery, there is something that picks out Troy’s winning as the kind of outcome that the lottery-rigger might want, because mothers often want success for their sons. Assuming that Troy’s mother had the ability to rig the lottery, this is enough to enable us to conclude that Troy’s lottery was probably rigged. This option is preferable to the rival possibility, that Troy’s winning was due to chance.

If we incorporate this principle into our Life-Containing-Universe Argument, it reads as follows:

This Further Refined Argument is now immune to the objection that undermined the Refined Argument in section 2. We have defined a non-subjective sense of ‘significant’ which enables us to distinguish outcomes like Rory’s lottery from outcomes like Roy’s (and Troy’s). We may conclude,
therefore, that the Further Refined Argument is sound.

4. Two objections

I will now discuss two possible objections to the Further Refined Argument. Both these objections will fail, but the reasons for their failures will shed light on the limitations of the Argument.

This objection may be inspired by the difference between Roy’s lottery and Troy’s lottery. Our objector may point out that, in Roy’s lottery, we do not conclude that there must be a lottery-rigger who has a reason to allow Roy to win. Only in Troy’s lottery, where we already know of the existence of a potential lottery-rigger, are we licensed to infer that the lottery was rigged. The situation with the universe is therefore more like Roy’s lottery than like Troy’s, in that we do not already know that a purposive agent (the creator) exists. Only if we did would we be licensed to infer that the life-containing universe is evidence for the agent’s existence.

But this objection misrepresents its target. The Argument does not presuppose the existence of the agent; it merely says that the existence of any circumstance which might plausibly have been brought about by a particular agent is evidence for that agent’s existence. The relevant difference between Roy’s lottery and Troy’s lottery is not the difference between whether or not we already know of a potential lottery-rigger. It is the difference between whether or not the outcome is the kind of outcome which might plausibly be attributed to a lottery-rigger. In Roy’s lottery, there is no feature of Roy which makes him a plausible target for a lottery-rigger. In Troy’s lottery, there is a feature of Troy which makes him a plausible target for a lottery-rigger: his relationship to the lottery administrator. (And in the case of the universe, there is a feature of the universe which makes it a plausible thing for an agent like God to create: its life-containing status.)

Here is a parallel example. If an explorer found a set of large round footprints, he would be right to think that they were evidence for the existence of a creature for which creating such footprints was within its nature and power. This does not presuppose the existence of the creature; it moves quite properly from the evidence to the inference.

If features of Troy’s lottery are misleading, we should remember that, in Rory’s lottery, we do not know of a potential lottery-rigger, yet we are justified in concluding that the lottery has been rigged. It is the occurrence of an unlikely outcome which would plausibly be significant to a hitherto unknown agent that provides evidence for the existence of that agent. The presupposition that the agent exists is not necessary to any such argument, so the argument is not circular.

This is a much better objection, but it can be answered. Before we do so, we should draw a distinction between proof and evidence.

If we have proof for the existence of X, we know that X exists. If we have evidence for the existence of X, we may believe that X exists, but the strength of our belief would depend on other factors – the initial likelihood of X, the strength of the evidence against X, and the strength of the evidence in favour of alternative hypotheses.

Now, the Further Refined Argument claims that the life-containing universe gives us evidence for the existence of an agent who designed the universe; it does not give us proof. In the light of the evidence it gives us, we must still assess whether or not we believe in the existence of that agent. We do that by considering the initial likelihood of the existence of the agent, the strength of any evidence against the existence of the agent, and the strength of evidence in favour of alternative hypotheses to explain the nature of the universe. As a result, we may or may not finally believe that such an agent actually exists. But, even if we finally believe that no such agent exists, the fact remains that the life-containing universe still gives us evidence that he does exist. It is just that, in our judgement, the evidence it gives us is not strong enough to overcome other considerations, such as the initial unlikelihood of the agent’s existence.

This, then, is how we can answer the Second Objection. We admit that versions of the Argument would, strictly speaking, provide some evidence for the existence of a particular purposive agent no matter what the outcome was. In Roy’s lottery, the Argument would provide evidence for the existence of a purposive agent who is biased towards entrants with the name Roy, or entrants from Roy’s particular village, or whatever. And similar claims could be made no matter who had won Roy’s lottery. But we can point out that to provide evidence for a claim is not to prove that claim, nor even to make it (on balance) likely. Other considerations must be taken into account. In the case of Roy’s lottery, the initial likelihood of there being an agent with a bizarre bias towards giving money to entrants with the name Roy, or to entrants from a particular village, is extremely low. It is low because, to our knowledge, probably nobody has ever really had this bias. We should therefore reject this theory as very unlikely, even though Roy’s winning does (strictly speaking) give us a bit of evidence in its favour. By contrast, the initial likelihood of there being an agent with a bias towards doing good is high, since many people have this bias. And similarly, the initial likelihood of there being an agent with a bias towards benefiting her children is high, since many people have this bias. We should therefore look closely at the outcome of Rory’s and Troy’s lotteries; and, providing there are no other circumstances that make the existence of a lottery-rigger independently unlikely, we should accept this theory and believe that there is, in fact, a lottery-rigger.

This highlights an important feature of the Further Refined Argument: since it only provides evidence and not proof, the usefulness of its conclusions depends on the independent likelihood of its conclusions being true. But, as Bayesian theory reminds us, this is a limitation of every non-deductive argument, so it should not worry us too much.

5. Conclusions

If my arguments are right, the Further Refined Argument is sound. It takes as its premises the following claims:

A life-containing universe would be within the nature and power of a particular purposive agent to bring about.

It then proceeds, by reasoning that I have shown to be valid using examples from Roy’s, Rory’s and Troy’s lotteries, to the following conclusion:

This conclusion is interesting. If it is right, it provides evidence not only that a purposive agent exists, but that the agent has certain characteristics – namely, that it is such that the creation of a life-containing universe would be within its nature and power. This conclusion takes us some of the way, if not all the way, towards a definition of God which would satisfy many religious people.

However, the conclusion is also limited, as we saw when we considered the Second Objection. Its most significant limitation is that it only provides evidence and not proof. Since this is a limitation of every non-deductive argument, it is not an objection to the argument. But it is particularly worth bearing in mind in this case because it means that we must take into account all the other factors that affect the likelihood of the agent’s existence. If the agent’s existence is shown to be rather unlikely (or even impossible) by another argument, then the conclusion of the Further Refined Argument will not be enough to raise its plausibility very much. Since, as it happens, I don't believe that the universe was designed (and I am an atheist), I happen to believe that this is in fact the case.

Finally, with respect to the Design Argument in general. In this paper, I have tried to show one way in which the Design Argument does not fail. I have done this by illustrating that the notion of‘ significance’ can be given a useful interpretation in this case. But the fact that the argument does not fail in this way does not mean that it is either valid or applicable overall. There are many different objections to the argument, or to applying it with respect to God’s existence. Since this paper has not considered these objections, the success or failure of my arguments in this paper is independent of the ultimate success or failure of the Design Argument’s claims.

References

 

 

Portfolio | Academic | Misc | Web design