In the latest issue:

An Ordinary Woman

Alan Bennett

Anglo-America Loses its Grip

Pankaj Mishra

Short Cuts: John Bolton’s Unwitting Usefulness

Mattathias Schwartz

Smells of Hell

Keith Thomas

Mrs Oliphant

Tom Crewe

Tippett’s Knack

Philip Clark

At Tate Modern: Steve McQueen

Colin Grant

Catherine Lacey

Nicole Flattery

Churchill’s Cook

Rosemary Hill

The ‘Batrachomyomachia’

Ange Mlinko

On Dorothea Lange

Joanna Biggs

Paid to Race

Jon Day

Poem: ‘Traveller’s Tales: Chapter 90’

August Kleinzahler

The Soho Alphabet

Andrew O’Hagan

Old Tunes

Stephen Sedley

Victor Serge’s Defective Bolshevism

Tariq Ali

The Murdrous Machiavel

Erin Maglaque

Diary: Insane after coronavirus?

Patricia Lockwood

Why anything? Why this?Derek Parfit
Vol. 20 No. 3 · 5 February 1998

Why anything? Why this?

Derek Parfit concludes his essay on the Universe

4701 words

In the first half of this essay, I suggested how reality’s deepest features might be partly explained. Of the countless cosmic possibilities, or ways that reality might be, a few have very special features. If such a possibility obtained, that might be no coincidence. Reality might be this way because this way had this feature. Thus, if nothing had ever existed, that might have been true because it was the simplest way for reality to be. And if reality is maximal, because all possible local worlds exist, this may be true because it is the fullest way for reality to be. The highest law may be that being possible, and part of the fullest way reality might be, is sufficient for being actual.

If some cosmic possibility obtains because it has some special feature, we can call this feature the Selector. If there is more than one such feature, they are all partial Selectors. Just as there are various cosmic possibilities, there are various explanatory possibilities. For each of these special features, there is the explanatory possibility that this feature is the Selector, or is one of the Selectors. Reality would then be the way it is because, or partly because, this way had this feature.

There is one other explanatory possibility: that there is no Selector. If that is true, it is random that reality is as it is. Events may be in one sense random, even though they are causally inevitable. That is how it is random whether a meteorite strikes the land or the sea. Events are random in a stronger sense if they have no cause. That is what most physicists believe about some features of events involving sub-atomic particles. If it is random what reality is like, the universe not only has no cause. It has no explanation of any kind. This claim we can call the Brute Fact View.

Few features can be plausibly regarded as possible Selectors. Though plausibility is a matter of degree, there is a natural threshold to which we can appeal. If we suppose that reality has some special feature, we can ask which of two beliefs would be more credible: that reality merely happens to have this feature, or that reality is the way it is because this way has this feature. If the second would be more credible, this feature can be called a credible Selector. Return, for example, to the question of how many possible local worlds exist. Of the different answers to this question, all and none give us, I have claimed, credible Selectors. If either all or no worlds existed, that would be unlikely to be a coincidence. But suppose that 58 worlds existed. This number has some special features, such as being the smallest number that is the sum of seven different primes. It may be just conceivable that this would be why 58 worlds existed; but it would be more reasonable to believe that the number that existed merely happened to be 58.

There are, I have claimed, some credible Selectors. Reality might be some way because that way is the best, or the simplest, or the least arbitrary, or because its obtaining makes reality as full and varied as it could be, or because its fundamental laws are, in some way, as elegant as they could be. Presumably there are other such features, which I have overlooked.

In claiming that there are credible Selectors, I am assuming that some cosmic and explanatory possibilities are more probable than others. That assumption may be questioned. Judgments of probability, it may again be claimed, must be grounded on facts about our world, so such judgments cannot be applied either to how the whole of reality might be, or to how reality might be explained.

This objection is, I believe, unsound. When we choose between scientific theories, our judgments of their probability cannot rest only on predictions based on established facts and laws. We need such judgments in trying to decide what these facts and laws are. And we can justifiably make such judgments when considering different ways in which the whole of reality may be, or might have been. Compare two such cosmic possibilities. In the first, there is a lifeless universe consisting only of some spherical iron stars, whose relative motion is as it would be in our world. In the second, things are the same, except that the stars move together in the patterns of a minuet, and they are shaped like either Queen Victoria or Cary Grant. We would be right to claim that, of these two possibilities, the first is more likely to obtain.

In making that claim, we would not mean that it is more likely that the first possibility obtains. Since this possibility is the existence of a lifeless universe, we know that it does not obtain. We would be claiming that this possibility is intrinsically more likely, or that, to put it roughly, it had a greater chance of being how reality is. If some possibility is more likely to obtain, that will often make it more likely that it obtains; but though one kind of likelihood supports the other, they are quite different.

Another objection may again seem relevant here. Of the countless cosmic possibilities, a few have special features, which I have called credible Selectors. If such a possibility obtains, we have a choice of two conclusions. Either reality, by an extreme coincidence, merely happens to have this feature, or – more plausibly – this feature is one of the Selectors. It may be objected that, when I talk of an extreme coincidence, I must be assuming that these cosmic possibilities are all equally likely to obtain. But I have now rejected that assumption. And, if these possibilities are not equally likely, my reasoning may seem to be undermined.

As before, that is not so. Suppose that, of the cosmic possibilities, those that have these special features are much more likely to obtain. As this objection rightly claims, it would not then be amazing if such a possibility merely happened to obtain. But that does not undermine my reasoning, since it is another way of stating my conclusion. It is another way of saying that these features are Selectors.

These remarks do show, however, that we should distinguish two ways in which some feature may be a Selector. Probabilistic Selectors make some cosmic possibility more likely to obtain, but leave it open whether it does obtain. On any plausible view, there are some Selectors of this kind, since some ways for reality to be are intrinsically more likely than others. Thus, of our two imagined universes, the one consisting of spherical stars is intrinsically more likely than the one with stars that are shaped like Queen Victoria or Cary Grant. Besides Probabilistic Selectors, there may also be one or more Effective Selectors. If some possibility has a certain feature, this fact may make this possibility, not merely intrinsically more likely, but the one that obtains. Thus, if simplicity had been the Effective Selector, that would have made it true that nothing ever existed. And, if maximality is the Effective Selector, as it may be, that is what makes reality as full as it could be. When I talk of Selectors, these are the kind I mean.

There are, then, various cosmic and explanatory possibilities. In trying to decide which of these obtain, we can in part appeal to facts about our world. Thus, from the mere fact that our world exists, we can deduce that the Null Possibility does not obtain. And, since our world seems to contain pointless evils, we have reason to reject the Axiarchic View.

Consider next the Brute Fact View, on which reality merely happens to be as it is. No facts about our world could refute this view. But some facts would make it less likely that this view is true. If reality is randomly selected, what we should expect to exist are many varied worlds, none of which had features that, in the range of possibilities, were at one extreme. That is what we should expect because, in much the largest set of cosmic possibilities, that would be what exists. If our world has very special features, that would count against the Brute Fact View.

Return now to the question whether God exists. Compared with the uncaused existence of one or many complicated worlds, the hypothesis that God exists has been claimed to be simpler, and less arbitrary, and thus more likely to be true. But this hypothesis is not simpler than the Brute Fact View. And, if it is random which cosmic possibility obtains, we should not expect the one that obtains to be as simple, and unarbitrary, as God’s existence is claimed to be. Rather, as I have just said, we should expect there to be many worlds, none with very special features. Ours may be the kind of world that, on the Brute Fact View, we should expect to observe.

Similar remarks apply to the All Worlds Hypothesis. Few facts about our world could refute this view; but, if all possible local worlds exist, the likely character of our world is much the same as on the Brute Fact View. That claim may seem surprising, given the difference between these two views. One view is about which cosmic possibility obtains, the other is about why the one that obtains obtains. And these views conflict, since, if we knew either to be true, we would have strong reason not to believe the other. If all possible worlds exist, that is most unlikely to be a brute fact. But, in their different ways, these views are both non-selective. On neither view do certain worlds exist because they have certain special features. So, if either view is true, we should not expect our world to have such features.

To that last claim, there is one exception. This is the feature with which we began: that our world allows for life. Though this feature is, in some ways, special, it is one that we cannot help observing. That restricts what we can infer from the fact that our world has this feature. Rather than claiming that being life-allowing is one of the Selectors, we can appeal to some version of the Many Worlds Hypothesis. If there are many worlds, we would expect a few worlds to be life-allowing, and our world is bound to be one of these few.

Consider next other kinds of special feature, ones that we are not bound to observe. Suppose we discover that our world has such a feature, and we ask whether that is no coincidence. It may again be said that, if there are many worlds, we would expect a few worlds to have this special feature. But that would not explain why that is true of our world. We could not claim – as with the feature of being life-allowing – that our world is bound to have this feature. So the appeal to many worlds could not explain away the coincidence. Suppose, for example, that our world were very good, or were wholly law-governed, or had very simple natural laws. Those facts would count against both of the unselective views: both the All Worlds Hypothesis and the Brute Fact View. It is true that, if all worlds exist or there are very many randomly selected worlds, we should expect a few worlds to be very good, or wholly law-governed, or to have very simple laws. But that would not explain why our world had those features. So we would have some reason to believe that our world is the way it is because this way has those features.

Does our world have such features, ones that count against the unselective views? Our world’s moral character seems not to count against these views, since it seems the mixture of good and bad that, on the unselective views, we should expect. But our world may have the other two features: being wholly law-governed, and having very simple laws. Neither feature seems to be required in order for life to be possible. And, among possible life-containing worlds, a far greater range would not have these features. Thus, for each law-governed world, there are countless variants that would fail in different ways to be wholly law-governed. And, compared with simple laws, there is a far greater range of complicated laws. So, on both the unselective views, we should not expect our world to have these features. If it has them, as physicists might discover, that would give us reasons to reject both the All Worlds Hypothesis and the Brute Fact View. We would have some reason to believe that there are at least two partial Selectors: being law-governed and having simple laws.

There may be other features of our world from which we can try to infer what reality is like, and why. But observation can take us only part of the way. If we can get further, that will have to be by pure reasoning.

Of those who accept the Brute Fact View, many assume that it must be true. According to these people, though reality merely happens to be some way, that it merely happens to be some way does not merely happen to be true. There could not be an explanation of why reality is the way it is, since there could not be a causal explanation, and no other explanation would make sense.

This assumption, I have argued, is mistaken. Reality might be the way it is because this way is the fullest, or the most varied, or obeys the simplest or most elegant laws, or has some other special feature. Since the Brute Fact View is not the only explanatory possibility, we should not assume that it must be true.

When supporters of this view recognise these other possibilities, they may switch to the other extreme, claiming that their view’s truth is another brute fact. If that were so, not only would there be no explanation of reality’s being as it is, there would also be no explanation of there being no such explanation. As before, though this might be true, we should not assume that it must be true. If some explanatory possibility merely happens to obtain, the one that obtains may not be the Brute Fact View. If it is randomly selected whether reality is randomly selected, and there are other possibilities, random selection may not be selected.

There is, moreover, another way in which some explanatory possibility may obtain. Rather than merely happening to obtain, this possibility may have some feature, or set of features, which explains why it obtains. Such a feature would be a Selector at a higher level, since it would apply not to factual but to explanatory possibilities. It would determine, not that reality be a certain way, but that it be determined in a certain way how reality is to be.

If the Brute Fact View is true, it may have been selected in this way. Of the explanatory possibilities, this view seems to describe the simplest, since its claim is only that reality has no explanation. This possibility’s being the simplest might make it the one that obtains. Simplicity may be the higher Selector, determining that there is no Selector between the ways that reality might be.

Once again, however, though this may be true, we cannot assume its truth. There may be some other higher Selector. Some explanatory possibility may obtain, for example, because it is the least arbitrary, or is the one that explains most. The Brute Fact View has neither of those features. Or there may be no higher Selector, since some explanatory possibility may merely happen to obtain.

These alternatives are the different possibilities at yet another, higher explanatory level. So we have the same two questions: which obtains, and why?

We may now become discouraged. Every answer, it may seem, raises a further question. But that may not be so. There may be some answer that is a necessary truth. With that necessity, our search would end.

Some truth is logically necessary when its denial leads to a contradiction. It cannot be in this sense necessary either that reality is a brute fact, or that there is some Selector. Both these claims can be denied without contradiction.

There are also non-logical necessities. The most familiar, causal necessity, cannot give us the truth we need. It could not be causally necessary that reality is, or isn’t, a brute fact. Causal necessities come lower down. Similar remarks apply to the necessities involved in the essential properties of particular things, or natural kinds. Consider next the metaphysical necessity that some writers claim for God’s existence. That claim means, they say, that God’s existence does not depend on anything else, and that nothing else could cause God to cease to exist. But these claims do not imply that God must exist, and that makes such necessity too weak to end our questions.

There are, however, some kinds of necessity that would be strong enough. Consider the truths that undeserved suffering is bad, and that, if we believe the premises of a sound argument, we ought rationally to believe this argument’s conclusion. These truths are not logically necessary, since their denials would not lead to contradictions. But they could not have failed to be true. Undeserved suffering does not merely happen to be bad.

When John Leslie defends the Axiarchic View, he appeals to this kind of non-logical necessity. Not only does value rule reality, Leslie suggests, it could not have failed to rule. But this suggestion is hard to believe. While it is inconceivable that undeserved suffering might have failed to be in itself bad, it seems clearly conceivable that value might have failed to rule, if only because it seems so clear that value does not rule.

Return now to the Brute Fact View, which is more likely to be true. If this view is true, could its truth be non-logically necessary? Is it inconceivable that there might be some Selector, or highest law, making reality be some way? The answer, I have claimed, is No. Even if reality is a brute fact, it might not have been. Thus, if nothing had ever existed, that might have been no coincidence. Reality might have been that way because, of the cosmic possibilities, it is the simplest and least arbitrary. And, as I have also claimed, just as it is not necessary that the Brute Fact View is true, it is not necessary that this view’s truth be another brute fact. This view might be true because it is the simplest of the explanatory possibilities.

We have not yet found the necessity we need. Reality may happen to be as it is, or there may be some Selector. Whichever of these is true, it may happen to be true, or there may be some higher Selector. These are the different possibilities at the next explanatory level, so we are back with our two questions: which obtains, and why?

Could these questions continue for ever? Might there be, at every level, another higher Selector? Consider another version of the Axiarchic View. Reality might be as good as it could be, and that might be true because its being true is best, and that in turn might be true because its being true is best, and so on for ever. In this way, it may seem, everything might be explained. But that is not so. Like an infinite series of events, such a series of explanatory truths could not explain itself. Even if each truth were made true by the next, we could still ask why the whole series was true, rather than some other series, or no series.

The point can be made more simply. Though there might be some highest Selector, this might not be goodness but some other feature, such as non-arbitrariness. What could select between these possibilities? Might goodness be the highest Selector because that is best, or non-arbitrariness be this Selector because that is the least arbitrary possibility? Neither suggestion, I believe, makes sense. Just as God could not make himself exist, no Selector could make itself the one that, at the highest level, rules. No Selector could settle whether it rules, since it cannot settle anything unless it does rule.

If there is some highest Selector, this cannot, I have claimed, be a necessary truth. Nor could this Selector make itself the highest. And, since this Selector would be the highest, nothing else could make that true. So we may have found the necessity we need. If there is some highest Selector, that, I suggest, must merely happen to be true.

Supporters of the Brute Fact View may now feel vindicated. Have we not, in the end, accepted their view?

We have not. According to the Brute Fact View, reality merely happens to be as it is. That, I have argued, may not be true, since there may be some Selector which explains, or partly explains, reality’s being as it is. There may also be some higher Selector which explains there being this Selector. My suggestion is only that, at the end of any such explanatory chain, some highest Selector must merely happen to be the one that rules. That is a different view.

This difference may seem small. No Selector could explain reality, we may believe, if it merely happened to rule. But this thought, though natural, is a mistake. If some explanation appeals to a brute fact, it does not explain that fact; but it may explain others.

Suppose, for example, that reality is as full as it could be. On the Brute Fact View, this fact would have no explanation. On the Maximalist View, reality would be this way because the highest law is that what is possible is actual. If reality were as full as it could be, this Maximalist View would be better than the Brute Fact View, since it would explain reality’s being this way. And this view would provide that explanation even if it merely happened to be true. It makes a difference where the brute fact comes.

Part of the difference here is that, while there are countless cosmic possibilities, there are few plausible explanatory possibilities. If reality is as full as it could be, that’s being a brute fact would be very puzzling. Since there are countless cosmic possibilities, it would be amazing if the one that obtained merely happened to be at the maximal extreme. On the Maximalist View, this fact would be no coincidence. And, since there are few explanatory possibilities, it would not be amazing if the Maximalist highest law merely happened to be the one that rules.

We should not claim that, if some explanation rests on a brute fact, it is not an explanation. Most scientific explanations take this form. The most that might be true is that such an explanation is, in a way, merely a better description.

If that were true, there would be a different defence of the kind of reasoning that we have been considering. Even to discover how things are, we need explanations. And we may need explanations on the grandest scale. Our world may seem to have some feature that would be unlikely to be a coincidence. We may reasonably suspect that this feature is the Selector, or one of the Selectors. That hypothesis might lead us to confirm that, as it seemed, our world does have this feature. And that might give us reason to conclude either that ours is the only world, or that there are other worlds, with the same or related features. We might thus reach truths about the whole universe.

Even if all explanations must end with a brute fact, we should go on trying to explain why the universe exists, and is as it is. The brute fact may not enter at the lowest level. If reality is the way it is because this way has some feature, to know what reality it like we must ask why.

We may never be able to answer these questions, either because our world is only a small part of reality, or because, though our world is the whole of reality, we could never know that to be true, or because of our own limitations. But, as I have tried to show, we may come to see more clearly what the possible answers are. Some of the fog that shrouds these questions may then disappear.

It can seem astonishing, for example, how reality could be made to be as it is. If God made the rest of reality be as it is, what could have made God exist? And, if God does not exist, what else could have made reality be as it is? When we think about these questions, even the Brute Fact View may seem unintelligible. It may seem baffling how reality could be even randomly selected. What kind of process could select whether, for example, time had no beginning, or whether anything ever exists? When, and how, could any selection be made?

This is not a real problem. Of all the possible ways that reality might be, there must be one that is the way reality actually is. Since it is logically necessary that reality be some way or other, it is necessary that one way be picked to be the way reality is. Logic ensures that, without any kind of process, a selection is made. There is no need for hidden machinery.

Suppose next that, as many people assume, the Brute Fact View must be true. If our world has no very special features, there would then be nothing that was deeply puzzling. If it were necessary that some global possibility be randomly selected, while there would be no explanation of why the selection went as it did, there would be no mystery in reality’s being as it is. Reality’s features would be inexplicable, but only in the way in which it is inexplicable how some particle randomly moves. If a particle can merely happen to move as it does, reality could merely happen to be as it is. Randomness may even be less puzzling at the level of the whole universe, since we know that facts at this level could not have been caused.

The Brute Fact View, I have argued, is not necessary, and may not be true. There may be one or more Selectors between the ways that reality might be, and one or more Selectors between such Selectors. But, as I have also claimed, it may be a necessary truth that it be a brute fact whether there are such Selectors, and, if so, which the highest Selector is.

If that is a necessary truth, similar remarks apply. On these assumptions, there would again be nothing that was deeply puzzling. If it is necessary that, of these explanatory possibilities, one merely happens to obtain, there would be no explanation of why the one that obtains obtains. But, as before, that would be no more mysterious than the random movement of some particle.

The existence of the universe can seem, in another way, astonishing. Even if it is not baffling that reality was made to be some way, since there is no conceivable alternative, it can seem baffling that the selection went as it did. Why is there a universe at all? Why doesn’t reality take its simplest and least arbitrary form: that in which nothing ever exists?

If we find this astonishing, we are assuming that these features should be the Selectors: that reality should be as simple and unarbitrary as it could be. That assumption has, I believe, great plausibility. But, just as the simplest cosmic possibility is that nothing ever exists, the simplest explanatory possibility is that there is no Selector. So we should not expect simplicity at both the factual and explanatory levels. If there is no Selector, we should not expect that there would also be no universe. That would be an extreme coincidence.*

Send Letters To:

The Editor
London Review of Books,
28 Little Russell Street
London, WC1A 2HN

Please include name, address, and a telephone number.


Vol. 20 No. 4 · 19 February 1998

Why does a literary magazine exist at all? It might have been the case that no magazine existed: no cover, no list of contributors, no contents. We have to go on and ask why we have the magazine we do have. Consider the Null Possibility. There could have been a journal with nothing in it. Consider next the All Worlds Hypothesis, in which our periodical would contain every possible kind of article. Somewhere in between is the publication we buy. Perhaps the Brute Fact View applies and we have to put up with what we get between the covers and not ask questions. On the other hand, there may be a Selector or a set of partial Selectors which determines what kind of magazine we experience. I am trying to get round, of course, to asking the ‘Selectors’ what was going on when they decided to publish Derek Parfit’s two-part article on the meaning of the universe (LRB, 22 January)? The world we live in is unfair enough, with the LRB appearing only fortnightly, and that terrible gap after Christmas, the deepest abyss in the year. To surrender two and a half pages in each of two issues to this meticulous but rather pontifical philosophical analysis is enough to make us cry out ‘Why?’ to the heavens.

Leonard Pepper

Vol. 20 No. 8 · 16 April 1998

Derek Parfit’s tour de force (LRB, 5 February) raises basic philosophical problems about the role that explanations can legitimately play. Parfit assumes that the notion of an explanation is sufficiently clear and uniform, that philosophical analysis will show when an explanation is needed and how to evaluate different candidates, and that the criteria apply across the board: from everyday facts to scientific conjectures of varying generality. This however is doubtful. ‘Explanation’ is an umbrella term covering an assortment of loosely related notions (explanations in physics, mathematical explanations, elucidations, clarifications, what have you). There is a danger of philosophical illusion in treating them as species of one generic concept.

Let me substantiate this by a particular objection to Parfit’s arguments. Parfit distinguishes those extremely unlikely events that do not require explanation from those that do. His example: a thousand people, I among them, face death. Only one person can be rescued and that person will be chosen by lottery. If I win (with odds 1 in 1000) and my life is saved, the fact requires no explanation. ‘Someone had to win, and why not me?’ In the second scenario I am sentenced to death, but the sentence will be remitted if the gaoler draws the longest straw out of a thousand. The longest straw is drawn and my life is saved. This, Parfit argues, is something special that requires an explanation. Appealing to coincidence won’t do. We will be right to conclude that it is much more likely that the draw was rigged. Sounds convincing. But is it? And how far can this line of thought be pushed?

Imagine a third scenario. As in the first, there are a thousand doomed people, one of whom, chosen by lottery, will survive. If I happen to win, no explanation is required. But let the lottery be implemented as follows. Going over the alphabetical list of the thousand people, an official draws, against each name, a straw from an initial bundle of a thousand straws. The winner is the person against whose name the long straw (1 in 1000) is drawn. Once this happens, the draws are discontinued. I am the first on the list and the first try succeeds. Exactly as in the second scenario, I am saved by a single draw of the longest straw out of a thousand, with no other draws taking place. The intention, to be sure, was to carry on in the event of failure. Why should this intention matter? Say I had lost, and the lottery continued. Again, why should this counterfactual reasoning make a difference? Counterfactuals are usually invoked in causal explanations: ‘Had John not slammed on the brakes, the car would not have swerved,’ showing that the slamming was the cause of the swerving. But in our case causality works in the other direction: from my winning to the termination of the lottery. It does not explain my winning. And it is not clear why it should obviate the need for explanation. And if there is no need here, why should there be one in Parfit’s second scenario?

Haim Gaifman
Columbia University, New York

send letters to

The Editor
London Review of Books
28 Little Russell Street
London, WC1A 2HN

Please include name, address and a telephone number

Read anywhere with the London Review of Books app, available now from the App Store for Apple devices, Google Play for Android devices and Amazon for your Kindle Fire.

Read More

Sign up to our newsletter

For highlights from the latest issue, our archive and the blog, as well as news, events and exclusive promotions.

Newsletter Preferences