One thing

City Slickers

In the 1991 film City Slickers, the wise old cowboy Curly explains something vital to Mitch, one of the three slickers who have come out to join the cattle drive and rediscover meaning in their lives. The dialogue goes as follows.

Curly: Do you know what the secret of life is?

Curly: This. (He holds up his right index finger)

Mitch: Your finger?

Curly: One thing. Just one thing. You stick to that and everything else don't mean shit.

Mitch: That's great, but what's the one thing? (Mitch smiles and holds up his right index finger)

Curly: That's what you've gotta figure out.

(Mitch looks uncertainly at his finger)

This is more intriguing, and more powerful, than the many self-help platitudes each of which claims to be the ultimate secret. It may not be in the league of Aristotle or Marcus Aurelius, but a one-liner does not have the same objective as a book. And figuring out what we might make of it from an impersonal standpoint is just as much of a challenge as an individual's figuring out what his or her personal one thing might be.

The phrase "one thing", while absolutely right in its original context in which Curly holds up one finger, can sound inelegant when used in other contexts. So we shall speak of an individual's project, with the implication that only one project will really matter to him or her at a given time.

In speaking of projects, we shall narrow the range of things that Curly invites us to identify. Projects have goals and results. Curly would allow an individual to select something that was not so teleological, for example "family" rather than "bringing up children". Our narrowing will allow us to be more specific in what we say than would otherwise be possible. But there would be other comments to make on a proposal to focus on things which were not defined in teleological terms.

This post will explore some complexities that come to light when we look at Curly's advice and its implications. Complexities are laid out, but not resolved. As may be apt for advice directed to individuals in relation to their own lives, resolution is left as an exercise for the reader.

What kind of project?

Options

There are plenty of projects on which someone might focus. Examples include bringing up a family (existing or planned), pursuing a career, or undertaking a business, academic or artistic project.

We can take it that Curly's recommendation would be limited to projects that really mattered to the individual, or that would at least have a good prospect of coming to be of great importance to the individual after he or she had got involved in them.

False steps

There would be scope for false steps, as there usually is in life. The individual might pick a project, even one that was already of great importance to him or her, and after a while find that it was not sufficiently important to justify making it the single focus of his or her current life. If the risk of that kind of false step was very high, one should perhaps not follow Curly's advice.

We here take it that the appropriate arbiter of importance is the individual, not some independent standard with which the individual might disagree. This reflects the fact that Curly offers advice to the individual. There is no proposal to manage society so that the various focuses of people would collectively produce the best overall result. Curly is a cowboy, and cowboys are not collectivists.

Breadth

An individual might select a project that was broad or one that was narrow. Both breadth and narrowness could have advantages. And a consequence of following Curly's advice and selecting only one project would be that there was a trade-off: an individual could not enjoy both the advantages of breadth and the advantages of narrowness at the same stage in life.

Robustness and adjustment

A broad project should be more robust than a narrow one. A narrow project, such as becoming an Olympic-level athlete, could easily be frustrated by some random accident, such as injury caused by slipping on an icy pavement. A broad project could be adjusted in its details to accommodate unexpected difficulties.

There is a risk in adjustment. Modest and rare adjustments would maintain the single focus that Curly recommends, even if the cumulative consequence was that the project after 20 years was not recognisably the same project as the one that was first adopted. But substantial or frequent adjustments would betoken a loss of focus, so that some of the benefit of following Curly's advice would be lost.

Guidance

The adoption of a project should yield guidance on what to do. There might be a lot of detail to fill in, not evident from the description of the project and perhaps not even implied by that description. But the general lines should be clear.

The statement of a broad project might not give much guidance. Statement of one of the broadest ones, such as "be happy" or "achieve worthwhile goals" would give hardly any guidance beyond indicating some types of activity to avoid (in our examples, activities that would induce misery or would waste time and energy). And adoption of a project as broad as that would not amount to following Curly's advice. But statement of a moderately broad project, such as "start a family" or "write a book", might give enough guidance. And statement of a narrow project should give quite a lot of guidance.

Focus

The nature of focus

The basic idea is that of focus on the individual's one project.

We take this focus to require primarily attaching value to progress in the project. How things go in relation to anything else is not to be important. There is indeed a suggestion in Curly's words that how other things go will automatically cease to matter, so long as the individual is focused on the one project.

Focus can seem like a good idea. More will be achieved in the area of focus, and the individual will be less bothered by things going on outside that area. At least, these results should follow so long as the project is not defined too broadly. But complexities crowd in quickly.

The sense in which a project matters

A project might matter to the individual, or it might be one that would be generally agreed mattered to society or reasonably could matter to an individual. Finally, if one were to allow talk of moral facts or other facts of a comparably unusual nature, one might see a project as mattering by reference to some factual standard.

Focus would only work as a motivator and as a way to stop other things mattering if a project really mattered to the individual, and mattering to the individual might be sufficient as well as necessary for focus to work. But we should not ignore the other ways in which a project might matter.

Social acceptance that a project was at least one which reasonably could matter to an individual might be needed in order to ensure that the individual was neither obstructed by social disapproval of the project nor deprived of friends.

The existence of a factual standard by which a project mattered would not contribute anything if the factual correctness of such standards was not manifest to most people, and it would seem that it would not be. People might claim that such-and-such standards existed, but there would not be any independent way to check their claims. And while ethical intuitionists might regard their conclusions as self-evident, general agreement with those conclusions would not always be found. So all we would have to go on would be claims that the standards existed, together with whatever support particular alleged standards might have garnered by virtue of having emerged from empirically informed debate over such matters as the social effects of respecting or violating certain standards. And there would be no good reason to heed any claims that were only espoused by a few people. Thus so far as power to motivate individuals went, reference to supposed factual standards would not in terms of content take us beyond reference to social acceptance.

Reference to supposed factual standards might however take us further in terms of motivation. If an individual thought that focus on the chosen project complied with standards that he or she took to be factual, or even better, if the individual thought that focus on the chosen project was (for him or her) positively recommended by such standards, that should encourage focus on the project.

Correspondingly, if relevant standards were supposed to be factual, an individual's fear that his or her choice of project might conflict with those standards would undermine motivation. It would not be possible for the individual to think that the conflict was merely with views of other people that could be disregarded because one was entitled to follow one's own star.

Neglecting other things

A focused life might have advantages, but one would need to ask whether it was acceptable for an individual to attach little or no importance to concerns outside the scope of his or her chosen project.

Whether it would be acceptable to the individual would depend on his or her psychology. Some people would not worry that other things which they might at other times have thought important were being neglected. Others would be seriously concerned. And while we might say that the former turn of mind would be more effective, we could hardly make an ethical judgement as to which turn of mind was psychologically preferable.

We could however discuss substantive ethical questions without judging the individual's psychology directly.

We may start with the positive benefits of focus. In favour of not worrying about the neglect of other things, one could say not only that more might be achieved but that a focused life would itself be a good life. It would not be the only kind of good life, but such a life would offer a better prospect of turning out to be good than several other kinds of life because it would be likely to be a life of high or at least respectable achievement. (We could only be sure of the quality of a life in retrospect. While it was being lived, the best one could do would be to act in such a way that there were good prospects.)

We now turn to the disadvantages of neglecting other things.

At the level of society as a whole, or by reference to long-term measures such as human progress, neglect would be very unlikely to matter. It would be exceptionally rare for what any one individual might have done or not done outside his or her main project to matter much. Someone else would have done something just as good. It is true that very widespread neglect of certain things, such as friendship, family, or civil society, would be a serious loss. But that is not a likely consequence of many people focusing on their single projects. Different people would focus on different things, and some people would focus on friends, family, or civil society.

There is a more troubling ethical question about the effects of neglect on people close to the individual, whether family or friends. (Work colleagues are not included here because someone who did not work hard enough would simply be replaced by someone who was willing to devote more effort to the job.)

We may distinguish two types of case, although the boundary between them would be decidedly hazy.

In a case of the first type, someone might gradually evolve into a highly focused person, with his or her personal relationships evolving in parallel to fit around that focus. Someone who needed a lot of time, or a nomadic life, to pursue his or her chosen project would form friendships of types that would fit with such demands. There would not need to be any specific person who could legitimately claim to have been unjustly left out of friendship, because those unable to be accommodated by virtue of the demands of the individual's focus would not have become friends, or at least not close friends, in the first place.

In a case of the second type, there would be existing personal relationships which would be damaged by the individual's coming to focus on a demanding project. One might expect this to apply to relationships with family members, who would automatically have a status of closeness to the individual and who might be counterparties to obligations on the individual. It could also apply to close or long-standing friends. Here there would be an ethical argument against adopting a project, focus on which would damage existing relationships. Having said that, one could also argue that each person was entitled to give priority to his or her own life and that it would be wrong for some friend or family member to expect that the individual should abandon his or her own aims. There is little hope of general rules to adjudicate individual cases. But we can say that there would be some dependence on whether the individual had voluntarily taken on the relationships in question, as when someone had chosen to start a family with a partner. There might also be some dependence on whether the individual's chosen project actually succeeded (something that could only be confirmed too late) or had a good prospect of succeeding (something that might be assessed in advance). We might here bring in what Bernard Williams said about Gauguin, who abandoned his family in pursuit of the fortunately fulfilled project of becoming a great artist ("Moral Luck", in Williams, Moral Luck: Philosophical Papers 1973-1980, Cambridge University Press, 1981).

Finally, we can ask whether an individual's life might fall short of being among the better sorts of life by virtue of constraints on relationships. Certain types of engagement with others which are widely considered to contribute to a good life might be ruled out by a focus on some demanding project. And it might be that while close relationships were formed, they would be at higher than usual risk of having to be broken off. That risk might taint relationships even before any break.

Change over life

Suppose that an individual selects a project and puts a great deal of energy into it for a few years. Various goals that fall within the scope of the overall project and that are worthwhile in themselves are achieved, so that the effort to date would not be wasted even if the project was no longer pursued. But there is a good deal more that could be done within the scope of the project.

Now suppose that the individual's priorities or attitudes change, as can easily happen as people get older. As a result the individual either deliberately abandons the project or, without a decision to abandon, devotes less and less energy to it.

There is no reason why such a change should devalue the achievements to date. The project was at the time worthwhile, it would still be directly perceived as worthwhile by anyone who had the priorities and attitudes the individual used to have, and its worth could still be appreciated indirectly by anyone who could imagine having those priorities and attitudes.

There is however a way in which the possibility of future change could legitimately concern an individual and might undermine his or her current motivation. Future change might arise not out of the kind of development in priorities and attitudes that is natural to human beings, but out of a realisation that a mistake had been made in selecting the project. The project might turn out to be too challenging, given the individual's abilities. Or it might have appeared to be one that was appropriate to the individual's priorities and attitudes, but only because there were implications of pursuit of the project that did not come to light early on. Careful thought in advance might reduce the risk of such a mistake, and once a mistake was appreciated there would be nothing for it but to start again. It is the possibility that a mistake would in due course come to light that might undermine current motivation. Not to lose motivation for that reason would be to exhibit the virtue of cheerfully living with uncertainty.

The plausibility test

In this post, we shall explore the test of whether responses to questions are plausible. We shall consider use of the test in philosophy and in history.

The plausibility test is in principle less demanding than the test of whether responses are correct. But it is still important, for two reasons. The first one is that it is not always possible to say whether a response is correct. The second one is that a focus on plausibility brings out certain important requirements for responses to be acceptable, such as that they should not outrage our background understanding and that (in some cases) they should confer Verstehen.

References are given at the end of the post.

The test

Suppose that one seeks a response to a philosophical question, or to the historical question of why events in the human world took the course they did. Such a response may be a one-line statement of some conclusion, or an elaborate account that implicitly answers the question.

We shall use the term "response" to cover the full range of such possibilities. And we shall sometimes be able to speak of responses being correct or incorrect, meaning either that they are answers to questions where those answers could be identified as correct or incorrect in the ordinary sense, or that they are more discursive responses which could nonetheless attract attributions of correctness or incorrectness which would be based on their head-on collisions with facts.

There will however be some responses which are too discursive to collide with facts in the required way, or which do not collide with facts in that way because of the nature of the subject matter of or the approach to it. Concepts of correctness and incorrectness will then be inapplicable. And there will be gradations, with some responses more or less open to classification as correct or incorrect.

When a response is not open to classification as correct or incorrect, an important control is to ask whether it strikes experts as plausible. In this context, plausibility will require making sense, not being outlandish or in serious conflict with how one thinks the world works, and so on. It will not just mean within the bounds of possibility. On the other hand, our intended sense of plausibility is not that a response is to be believed cautiously, or that it is to be assigned some respectable probability of being correct (such as 0.25). Rather, the sense is that adoption of the response would not be unreasonable.

This plausibility test is what will concern us here. When applied in an academic discipline, it will not stand alone. Even if a judgement as to correctness is not expected, there will still be scope for detailed argument as to the evidential and other reasons to accept or reject a response. And what is learnt in the course of such detailed argument should influence judgements as to whether the response is plausible. But the plausibility test will still take investigation a step forward from the stage of detailed argument. It is a safeguard against getting lost in the trees of analysis so that one fails to see the wood of the overall picture, an overall picture which will include the background of existing understanding.

The test also has a role when a judgement as to correctness is envisaged. Even in such cases, once all the detailed work on evidence and reasoning has been done, it is worth standing back and asking whether the response is plausible. This final stage may not be especially worthwhile in disciplines in which one can be confident both that all the relevant evidence has been collected and interpreted correctly, and that its analysis will drive one to an inevitable conclusion as to whether a response is correct. But such happy conditions are only met in physics, chemistry, and a few other parts of the natural sciences. Elsewhere, and certainly across the humanities, the plausibility test is a valuable additional check. This role of the test in writing history was for example highlighted by Geoffrey Elton (The Practice of History, chapter 2, section 5). Elton did not draw our distinction between contexts in which judgements as to correctness are envisaged and contexts in which they are not, but he did stress the need for a stage of detailed analysis of the available evidence.

There is no algorithm for the plausibility test. A judgement as to whether the test is passed will not rest directly on a detailed analysis of evidence or reasoning, even though it may be influenced by matters that have come to light in the course of such an analysis. Rather, whether a response is plausible will simply be manifest to an expert, who will not then seek further justification for their judgement.

One might speak of intuitive judgements. We shall however not do so, except when we refer to other authors' work on intuition. A reference to intuition may be helpful in grasping the notion of judgements that responses are plausible. But the concept of intuition is liable to bring some baggage with it. One can see some of this baggage in discussions of the role of intuitions in philosophy. Philosophers are widely thought to rely on intuitions, but there is also a case to be made that intuitions are not needed (Cappelen, Philosophy without Intuitions). One might argue that the proper role of intuitions in philosophy was not in reaching or directly supporting specific conclusions, but in making judgements of plausibility, albeit subject to the qualification that a response which failed the plausibility test might nonetheless be correct (since intuitions can mislead). We shall not pursue that line of argument here. But we do note that it would be challenged by the fact that a lot of philosophical discussion centres on the scope for intuitions to give direct support to specific conclusions (see the papers in Booth and Rowbottom (eds.), Intuitions).

We should add that expertise does matter. Someone must have had the right kind of education and experience for their judgements to count for much. To describe a response's plausibility or implausibility as manifest, or to refer to intuition, is not to throw the door open to the views of the ill-informed or the untrained.

Examples

We shall now set out some examples of use of the test. We shall include some comments on use of the test in different contexts. Later on we shall focus on the nature and the value of the test itself, rather than on examples of its use.

Our examples will be drawn from the disciplines of philosophy and history. The sense in which a response to a question can be plausible differs as between these disciplines. The normal sense in the areas of philosophy we shall highlight is that a response may be plausible given our own attitudes and habits of thought. In history, the point of reference is not our own attitudes and habits of thought but those of people of the past, along with what was feasible at the time studied. Our own attitudes and habits of thought, along with our current knowledge of people's physical and mental capacities and our grasp of the history of technology, will however be the starting point in compiling the point of reference we need.

Our treatment of the tests in the areas of philosophy we shall highlight and in history as a single test is justified by the fact that in both cases, the point of reference is primarily the nature of human beings. This also opens up scope for applying what would be broadly the same test across other humanities and to some extent across the social sciences. But while there can be good reason to ask whether an account in the natural sciences is plausible, the point of reference would be different enough that one could not say that the plausibility test would be the same kind of test.

Meta-ethics

Various responses to the question of the nature of ethical claims may be accepted by some ethicists and rejected by others on technical grounds. But the most powerful reason for rejecting some responses can be that the responses simply strike one as implausible against the background of a common human understanding of the nature of ethics.

Emotivism may be rejected because it seems plain that an ethical claim is more than an expression of preference, or even an expression of preference with which one would expect others to agree. And one can reject emotivism as implausible on such grounds even before noticing technical concerns such as those raised by the Frege-Geach problem (Geach, "Assertion", pages 463-464).

Some forms of moral realism may be rejected because it seems that there is no space in the world as we generally understand it for moral facts or moral properties on a par with non-moral facts or properties. (One source for such arguments is Mackie, Ethics: Inventing Right and Wrong, chapter 1, section 9.)

Substantive ethics

Suppose that an ethicist argues for some specific response to the question of how to act in a situation of a given type. Other people may agree, or they may disagree but think that the response is nonetheless plausible. As an example of such disagreement, one ethicist might endorse telling small lies to save people's feelings, and another one might say that while their own commitment to honesty was too strong for them to agree, they could still see that the use of small lies for such purposes could be a sensible policy. Whether there was agreement or such eirenic disagreement, the first ethicist's response would pass the plausibility test in the eyes of the second one.

On the other hand, an ethicist might offer a response from which many people would recoil. One such response would be that nobody should ever lie, even to protect someone from a potential murderer. Then the response would in the eyes of most people not pass the plausibility test, and we would start to look for defects in the reasoning or in the premises from which it started.

In order for us to fit the application of the plausibility test to substantive ethics into our discussion, we need to take it that there is scope for disagreement about what to do in situations of different types. An emotivist position would stand in the way of seeing scope for disagreement. But there are other meta-ethical positions, so we shall explore application of the plausibility test on the assumption that there is scope for disagreement about what to do.

Use of the plausibility test in substantive ethics draws on our personal inclinations. We may accept one response to a question as to what to do as reasonable, and reject another one as manifestly immoral, because of our own values. This need not be illegitimate. Ethics concerns how to conduct human lives. So it seems reasonable to give a conspicuous role to how human beings regard it as appropriate to live. But that does give rise to questions. How respectable are the origins of our views on how people should live? And are those views consistent enough between people and across cultures for judgements of plausibility to be thought of as having more value than idiosyncratic preferences?

On the respectability of origins, we may compare reaching a verdict on the plausibility of some response to a question of what to do with the route to conclusions that has been advocated by intuitionists. Intuitionists take the correctness of an ethical claim to be self-evident, without the need for direct support from argument (Stratton-Lake, "Intuitionism in Ethics"; for a full exploration of ways in which intuitionists may reach their ethical conclusions and advocacy of one particular way see Roeser, Moral Emotions and Intuitions). Correctness in the eyes of intuitionists does not however mean obviousness at first glance. It may become evident only after detailed thought which helps to bring out what is salient. Likewise, a judgement of the plausibility of a response will not depend on arguments that support the response directly, but it will be likely to follow detailed argument that relates to aspects of the response other than its plausibility. Having said that, there is a difference between the intuitionist approach and our approach. An intuitionist's claim that it is self-evident that a particular ethical claim is correct is a very strong claim. All conflicting claims are ruled out. To say that it is self-evident that a response which makes the claim is plausible can be to say something much weaker, because several conflicting responses might still be admitted to be plausible.

On consistency of views, we may look to surveys that have been conducted under the banner of experimental philosophy. (Much has been and is being written in this area. Two starting points for what we say here, including our remark about Gettier cases under the heading of epistemology, are Knobe, "Philosophical Intuitions are Surprisingly Robust Across Demographic Differences"; Stich and Machery, "Demographic Differences in Philosophical Intuition: a Reply to Joshua Knobe". On personal identity under our heading of metaphysics see Tobia (ed.), Experimental Philosophy of Identity and the Self.)

There is some evidence from experimental philosophy that people's ethical views do vary across cultures and can be affected by framing. That would count against the worth of views on the plausibility of responses to ethical questions where those views were not based on detailed argument, perhaps reducing the views to expressions of idiosyncratic preferences. Having said that, there are reasons why we should perhaps not be too concerned.

One reason not to be too concerned is that interpretations of the data differ. Not everyone agrees that there are wide variations in views.

A second reason not to be too concerned is that the data often reflect the views of people in general, while we are interested in application of the plausibility test by experts (although sometimes experts are surveyed and turn out to have varying views). There may not be a clear distinction between people in general and experts when it comes to substantive ethics, but there is still a distinction between what people will say when answering a questionnaire quickly and what they would say if they were encouraged to reflect on their views before answering. And those who had already spent some time considering ethical questions could be expected to reflect more thoroughly and more effectively than those who had not.

A third reason not to be too concerned is that even if results varied between cultures, the results obtained in a particular culture could still be seen as valid for people within that culture. The significance of an ethical claim's being thought correct or of a response to an ethical question's being thought plausible would however then have to be limited to what could be concluded from its being thought correct or plausible relative to the relevant culture. No conclusion based on any supposed general correctness or plausibility could be drawn.

Finally, when reviewing data on what people think, we must be sensitive to whether the questions they were asked related to correctness or to plausibility. These can be hard to disentangle. A claim will typically be that in all situations, or in all situations of certain types, some specified conduct is required, acceptable, or forbidden. And the question put to people is quite likely to amount to "Is it required/acceptable/forbidden to perform such and such action in such and such circumstances?". Such a question would relate to the conduct in question rather than to the claim considered as a response to a question. One might conclude from what a subject said that agreement with a response encapsulated by a claim was being shown. For example, agreement that some specified conduct was required would imply a view that a response to the ethical question to the effect that the conduct was required was correct. It would be less straightforward to get at views on the plausibility of responses. Thinking a claim correct would imply thinking that a response which the claim encapsulated was plausible, but it would be left unclear how wide a range of people thought a claim incorrect but still regarded such a response as plausible, and therefore unclear how far views on plausibility of responses varied between different cultures or between segments of the population defined in other ways. One might however at least hope that if ratios of regarding claims as correct to regarding them as incorrect were much the same across cultures or across other segments, proportions regarding corresponding responses to ethical questions as plausible would be much the same too.

Metaphyiscs

In fundamental physics, any plausibility test that is of value will be couched in highly technical terms. It will only be usable by physicists who are deeply immersed in current research. It would be misleading to think of it in the way in which we have been thinking of the plausibility test more generally. And if the results obtained or their conceptualisation turn out to be utterly counter-intuitive to non-experts, so much the worse for those people. Any objection that a non-technical plausibility test was failed would rightly be ignored by physicists.

Turning back to philosophy, we do reasonably demand a commonsense metaphysics of space, time and matter. That is however allowed us, with no need to challenge physics. We may borrow an image supplied by Max Tegmark, without needing to accept his whole theory. The world as described by counter-intuitive physics can unproblematically be seen as giving rise to the consensus reality of space, time, objects, and the observable interactions of objects within which we conduct our lives (Tegmark, Our Mathematical Universe, chapter 9).

There are other areas of metaphysics in which there is not the same need to accept the supremacy of potentially non-intuitive physics and then recover our everyday world.

One example is given by personal identity. A response to the question of what constitutes personal identity over time may be reviewed to see whether it offers a secure identity through all sorts of changes. These will include both gradual changes such as those of maturing and ageing, and sudden changes such as those caused by brain injuries. And the identity must be substantial enough to fulfil its practical roles in settling to which people we relate in certain ways (as family, friends, colleagues, and so on), in settling attributions of responsibility and property, and so on.

How the plausibility test can affect debates is interesting. Philosophers will imagine strange cases, including brain swaps, brain divisions, and successful and interrupted teleportations. Then they will precisify or amend everyday notions of personal identity to find notions that will yield plausible verdicts on those cases. This will be an initial application of the plausibility test. But the test must be applied again in relation to cases that actually occur.

It will usually be trivial to show that a notion of personal identity yields plausible verdicts in the most straightforward everyday cases. But there may be difficulties in less straightforward cases, such as extreme memory loss, where we naturally want to say that identity is in fact preserved but have difficulty in doing so. There may also be concern that a notion provided by philosophers is not robust enough to meet our everyday requirements. For example, a criterion based on a sense of attachment to one's past or to one's forthcoming conscious life might be thought to be too focused on ephemeral mental phenomena. And philosophers like Derek Parfit (Reasons and Persons, chapter 12) who say that identity is not the important thing may be thought to have insufficient respect for a notion that is central to our personal and social lives. In all the cases in this paragraph we may see applications of the plausibility test, both by philosophers and by other people who take an interest. And even if one were to discount the views of non-philosophers, a notion that they would criticise as implausible might well also be criticised as implausible by many philosophers.

Another example is given by the question of human free will. From the inside, we have a clear sense of making our own decisions and acting on them. But from the outside, we may be told that all of our thoughts and actions as they may be characterised in human terms supervene on physical reality, and that this reality's evolution reflects a mixture of determinism and randomness.

(For an introduction to views we now mention see Fischer, Kane, Pereboom and Vargas, Four Views on Free Will.)

Some philosophers will tell us that free will is indeed illusory. Such responses to the question of free will would fail any plausibility test, even among philosophers, until good reason had been given to think that there was no alternative.

Other philosophers, the libertarian incompatibilists, will tell us that while the physical world cannot accommodate free will, we have it anyway. Such a response might fail the plausibility test not because it would be disappointing, but because it would be unclear how free will would arise.

Finally, there are the compatibilists who work on our initial conceptions of free will. Schopenhauer located freedom in our freedom to do what we will, and dismissed the idea of our being able to will what we will (Prize Essay on the Freedom of the Will). More recent philosophers have developed the notion of guidance control: our choices and actions reflect our own personalities because lines of causal influence flow through our brains and bodies, but this does not imply that things could have turned out differently by virtue of influences originating within ourselves and not directly or indirectly prompted by any prior events in the external world. Compatibilist responses to the question of free will would seem to have the best prospect of passing the plausibility test. This reflects the fact that a response can on reflection be found plausible by virtue of some adjustment to our initial demands, in this case an adjustment to everyday conceptions of free will.

Epistemology

Most people are confident that many facts are known by humanity, and that they themselves know a fair few. Even experts in various disciplines, well aware of the difficulty of making discoveries and the risk of error, take the same view of the contents of their disciplines. Responses to the question of how to define knowledge that would make knowledge very hard to obtain, for example definitions that would require no possibility of error, would therefore fail the plausibility test.

A more interesting case is that of definitions which discriminate between examples of knowledge and non-knowledge in ways that provoke debate, both among people generally and among experts in various disciplines. Typically the examples are justified true beliefs where there has been some element of luck, for example through harmless or even positively helpful reliance on false premises or on defective reasoning. A response that supplies a given definition may be thought to fail the plausibility test if the definition too often classifies such beliefs as knowledge when people are inclined not to do so, or if it too often classifies them as non-knowledge when people are inclined to think of them as knowledge.

As with substantive ethics, such verdicts as to plausibility can legitimately have force. The facts that we know may not be human constructs, save to the extent that we have invented specific concepts to express those facts, and even then the independent world may have forced us to use certain concepts and not others. But the notion of knowledge is our own construct. It has been created to capture important facts about our relationship to the world, such as the fact that people with knowledge tend to get on better than people without it. The notion has been tailored to capture the fact that there is something more advantageous than true belief, following the issue raised by Plato (Meno, 97-99). And the verdicts on individual beliefs that a definition gives had better not be too far out of line with what people would think prior to philosophical reflection.

But as with substantive ethics, we must ask about the consistency across cultures of pre-philosophical thought. Again, experimental philosophy has something to say. Views on Gettier cases and the like do vary. But as with substantive ethics, we can ask how great the variation really is.

One difference from substantive ethics is that it is harder with knowledge to minimise concerns about a lack of consistency by saying that views may be for particular cultures rather than for the whole of humanity. Cultural relativity need not be overly troubling in relation to substantive ethical views. Different people live in different circumstances, with different histories. So different judgements of good and bad, right and wrong, may be apt to different societies. But the notion of knowledge is closely tied to the notion of truth. And we tend to regard both the notion of truth and the set of propositions that are true as universal.

History

At first glance, the discipline of history might seem to amount to the narration of facts on the basis of evidence. That would leave no room for a plausibility test, save in making judgements when the available evidence was not decisive.

But history goes far beyond chronology. Historical accounts impute motives, they abstract from factual details to identify political, economic and social forces, and they give narratives that are powerfully explanatory and that confer understanding on those who read them.

Having said that, history is far from being a natural science. There is no scope for repeated experiments, nor for precise and decisive calculations of the extent to which evidence supports conclusions. And both the motives of human beings and the causal links between what they think and what they do are too ill-defined to admit of comprehensive calculation.

In such a context, there is work for the plausibility test to do. Are portrayals of people's thoughts, fears and desires, and identifications of reasons for significant actions, plausible?

The test will usually be passed in published work, because historians are also human beings and will have a good inner sense of what would be realistic portrayals of the people they study. But we may still see the test as having played a role in processes of thinking, writing and re-writing, perhaps playing that role without historians' being conscious of its having done so. 

The plausibility test acts as a filter to dispose of responses to questions of why events took the course they did which would not be much good, rather than as a way to show that a given response which passes the test is correct. In the context of history, the test may first be applied to see whether Vestehen (understanding) is conferred. Its conferral would be a good sign, given the need for consonance with our background understanding of people and the world if it is to be conferred. Then the test may be applied in two directions, running between Erklären (explanation in a broadly scientific sense) and Verstehen.

The first direction runs from Erklären to Verstehen. Suppose that some quasi-mechanical causal account has been given as a response to the question of why events took the course they did. If the account is good enough, it will amount to Erklären. But is it good enough? Given the lack of mechanical precision in the world as viewed by historians, it is useful to have a test that brings in a different set of requirements. This is what the plausibility test can do. It can be used to assess whether the quasi-mechanical detail in the account affords a route to Verstehen. The move is from testing the proposed mechanism by reference to technical and quasi-mechanical principles to testing whether the account would resonate with us on the basis of our common understanding of how human beings and the world work.

The second direction runs from Verstehen to Erklären. An account may seem to confer understanding, and that may be checked in an initial application of the plausibility test. But an impression of understanding may be too easily given. If understanding depends on accepting an account that assumes a pattern of causes and effects which is in quasi-mechanical terms implausible, then the account should be rejected as an implausible response to the question of why events took the course they did.

Historians must be cautious when checking for plausibility by reference to Verstehen, whether in an initial application of the test or to see whether the quasi-mechanical detail provided affords a route to Verstehen.

The need for caution arises from the fact that principles of human motivation and action, the satisfaction of which in an account is required for the account to confer Verstehen, vary as between societies. Given that the test should be of whether the people studied could plausibly have acted as described, the relevant principles will be the ones of that society. (In anthropological terms, an emic approach should be preferred to an etic one.) These relevant principles may not be the ones that first occur to historians, especially if the historians come from a society other than the one being studied or if they are looking back many centuries. Just how easy it can be for principles to differ can be seen from one anthropologist's attempt to explain the story of Hamlet to members of a west African community (Bohannan, "Shakespeare in the Bush").

A general consideration of the test

We now move on from specific applications of the plausibility test to a more general consideration of its worth.

Plausibility and correctness

A judgement that a response is plausible is not always a judgement that the response is uniquely correct or that it is among the correct responses. It is however at least a judgement that the response should not be discarded yet, but should continue to be kept in play and explored as a useful way to look at relevant features of the world. It is therefore at least a judgement that the response should not currently be regarded as incorrect, because responses regarded as incorrect are never worth keeping in play except perhaps as part of a lateral thinking exercise in which they may prompt new thoughts.

There are three possibilities to consider.

The first possibility is that it was expected that one response would deserve to be regarded as correct. Then the identification of only one response as plausible might amount to a judgement of its correctness.

The second possibility is that it was expected that several responses would deserve to be regarded as correct. If it was thought that all plausible responses deserved to be regarded as correct, a judgement of plausibility would amount to a judgement of correctness. (This case might not be reducible to the one-response case. Differences in the aspects of the topic on which responses focused might suffice to obstruct simply taking their conjunction and presenting that as a perhaps unwieldy single correct response.)

The third possibility is that the nature of the discipline or of the topic would make it inappropriate to assert that every response could be shown to be correct or shown to be incorrect, so that there would be space for an intermediate category of responses which, while plausible, could not have their status as correct or as incorrect determined. Then a judgement of plausibility would not in general amount to a judgement of correctness, although it might do so in relation to some responses. (There might or might not be responses which could never have their status determined. It might only be that at any one time, there would be responses which could not have their status determined in the near term. As the discipline progressed, they might have their status determined or they might cease to be of any interest, but it would be likely that new responses of indeterminate status would also come into play.)

When a judgement of correctness is made, a judgement of plausibility becomes redundant save as a path to a judgement of correctness. Correct responses have to be accepted whether one likes them or not. The favourable result of any test of plausibility would play no more than a supporting role, showing the absence of a certain kind of objection to a response. There is also pressure on experts to resolve any disagreement as to correctness.

It is when no judgement of correctness is made, or when conflicting judgements of correctness are made by different experts and there is no prospect of resolving their disagreement in the near term, that the role of plausibility becomes interesting. A judgement of plausibility is not rendered redundant, but may be valuable in its own right. So a test of plausibility may make a real contribution, moving us forward from a plethora of possible responses either to one plausible response, or to a modest range of plausible responses.

We should think in terms of a range of responses, however many responses are currently in play. Even if only one is in play, we should allow for a potential range that would encompass responses which could be introduced. That possibility would be interesting because while there might be grounds to think that only one response could be correct, and there would always be reason to think that when responses conflicted no more than one of them could be correct, there would not in general be reason to think even that only one out of a range of conflicting responses could be plausible, at least not when determinations of correctness were not expected to be available in the near term. 

It may be perfectly possible to consider each member of a set of responses plausible, even if various responses in the range would not sit comfortably together or would contradict one another. This would however need to be limited to saying that each one was plausible individually, not that the conjunction of all of them would be plausible.

If conflict fell short of contradiction, for example because different responses identified different ethical or experiential considerations as central while each allowed some role for considerations picked out as central by others, or because different responses identified different factors in the explanation of some historical event as the most significant factors, there would be a hope and maybe an expectation that either current approaches to the question or knowledge of the world would in due course advance so that some responses would drop out and the conflict would be resolved. It would however be possible to live with the thought that the conflict might never be resolved. If there was contradiction, there would be a more pressing need to resolve the conflict. Then a thought that the conflict might never be resolved would amount to a thought that our grasp of the world and of life might be irremediably inadequate.

One feature of the humanities would contribute to making it tolerable to regard several conflicting responses as acceptable. This is the fact that a given response is prone to carry with it a given way to weigh up competing considerations and sometimes a given way to interpret evidence. So someone who favours one response may see someone who favours another response not as debating with them against a background of a completely shared understanding of the significance of different considerations and of the meaning of the evidence, but as debating against a background that was in some respects different. The effects would be a bit like that of perspective-taking in the natural sciences, where what might be seen as tension between different accounts of the same phenomenon can be defused by a recognition that different scientists may approach a single topic from different perspectives.

The significance of tolerance

We now turn to the significance of the scope to tolerate several conflicting answers as plausible. We shall explore what it might mean for the worth of a response's passing the plausibility test, not merely in cases in which several conflicting responses are in fact considered plausible, but in general. We shall say something about the nature of the test which will be relevant whether or not, in a particular case, several conflicting responses all pass it.

A judgement of plausibility is at least a judgement that it makes sense to look at some feature of the world in a particular way. It is stronger than a judgement that it is pedagogically helpful to look at the feature of the world in that way. Pedagogical helpfulness will reflect the psychology of students. Responses that experts would regard as misleading may turn out to be helpful. We on the other hand are concerned with what fully trained experts, who would not need the same level of assistance as students, would say about keeping possible responses in play.

Having said that, a response may be plausible in our sense when the usefulness of keeping it in play reflects the powers of thought of experts, while some hypothetical more advanced being would regard the response as misleading. Such a being might view it in the ways that human experts would view ways to look at the relevant feature of the world which were needed only to help students. We cannot have more than a speculative grasp of what the powers of such a more advanced being might be, so we cannot tell which answers might be downgraded in the eyes of such a being. 

One might argue that such advanced beings were already among us, in the form of artificial intelligence systems. We might declare certain ways to look at features of the world of which the systems appeared to have no need to be of merely pedagogical helpfulness. We would however be reluctant to do so if, as would be entirely possible, we could not grasp how such systems thought. Dispensing with our own ways to look at features of the world would then leave us with artificial intelligence systems which could assure us that the evidence found should not surprise us, or which could make accurate predictions, but the systems would not give us any understanding. Concerns about degrees of sophistication and comprehensibility of different responses would however mainly arise in relation to the natural sciences, rather than in relation to the humanities which are our concern here.

While a judgement of plausibility is stronger than a judgement of pedagogical helpfulness, it is weaker than a judgement that the world actually is as described, at least when the judgement of plausibility falls short of a judgement of correctness. It is weaker than such a factual judgement even if the factual judgement is read in ways that either anti-realists or perspectivists in the philosophy of science might advocate. (We take perspectivists to read factual judgements as saying "From this perspective, the world is like this".)

The comparative weakness of a judgement of plausibility explains why it is possible to judge several responses plausible even when they contradict one another. If responses were judged to say how the world was, such tension between them would be intolerable. If one description of the world was thought correct, contradictory ones would have to be thought mistaken. This is not because the world is known not to be an awkward place which could encourage contradictory descriptions. The existence of sensible if inconclusive discussion about adjusting logic in the face of quantum mechanics shows that we cannot be confident of the world's not being so awkward. Rather, our aversion to regarding contradictory responses as correct springs from the fact that doing so would undermine our notion of judgements of correctness as saying how the world actually was and implying decisively how it was not. If our judgements of plausibility do not amount to judgments of correctness, we can tolerate contradiction more easily. Contradiction will however remain uncomfortable.

Likewise, the comparative weakness of a judgement of plausibility makes it easier to judge several responses plausible when they would conflict with one another in some way that fell short of contradiction than it would be to judge all of those responses correct. The prospect of the world's being awkward enough to encourage conflicting but non-contradictory descriptions is more tolerable than that of the world's encouraging contradictory descriptions. It would undermine not our notion of correctness, but our confidence that we were on the high road to a full understanding. And finding several conflicting but non-contradictory responses plausible would not be as depressing as thinking we had to regard them all as correct. It might indeed be encouraging, in that it showed we were capable of developing several lines of enquiry without knowing which one would turn out to be the most fruitful.

We can see one way in which it can be tolerable to keep conflicting responses in play by reflecting on how things are in certain types of philosophy and in history. An important feature of these areas of work, and of some other work in the humanities, is that decisive judgements as to the truth of assertions can often be expected to lie permanently out of reach. This is so because in order to say anything interesting, one has to go beyond what the evidence requires one to say or not to say, and to interpret it in ways that are optional. Given that the crunch point at which final verdicts of truth and falsity are announced is not expected to be reached, conflicting responses can be tolerated because their coexistence will not be forced to come to an end. This does not however mean that anything goes. There are still standards to be met. For standards in history see Baron, Epistemic Respectability in History.

The value of the test

As we have already said, the fact that a response passes the plausibility test does not show that it should be taken to be correct, at least not without some substantial additional conditions being met. And we can often tolerate having conflicting responses to the same question which all pass the test. The test is only a filter to reject some responses, while potentially leaving several others in play. Does all this mean that the test is too undemanding for its administration to have value? No, it does not.

The fact that the test is only a filter to rule out some responses does not undermine the whole process of review of responses, because it is not the only test. It will normally be administered in addition to a detailed analysis of evidence and reasoning.

But why should such a filter add much to detailed analysis?

One reason is that detailed analysis cannot be as conclusive in the humanities, or indeed in the social sciences and some parts of the natural sciences, as it can be in physics and chemistry. Forms of evidence are diverse enough for there to be scope to overlook relevant evidence, and the evidence that is found can be misinterpreted.

A second reason is that evidence may be open to alternative interpretations, all arguably legitimate, in the light of different background theories. The use of such theories may be required in order to make the evidence speak at all, so it is not to be avoided. From within a theory, its interpretation of the evidence cannot be seen as illegitimate, and there may be no neutral standpoint from which to weigh up the alternative theories and their approaches to the interpretation of evidence. Then the plausibility test, based on general principles which are not specific to particular theories, may help to screen out theories which fall short in some way. The test would typically do so by finding that responses in some way failed to make sense, implying that there might have been something wrong with the theories under which they were produced. The test would however still not offer a neutral standpoint from which theories could be examined, because it would not have the tools with which to examine theories directly. Indeed, its use might lead to the identification of a theory as unsatisfactory without specifying the defects in that theory, merely condemning it by reference to its results.

A third reason is that in the humanities in particular, there is a legitimate demand for Verstehen. Responses to questions must bear an appropriate relationship to the nature of human readers, one that really does confer understanding. It is hard to check for that through detailed analysis. A general test like the plausibility test is what is needed.

References

Baron, Richard. Epistemic Respectability in History. CreateSpace, 2019.

https://rbphilo.com/history.html

Bohannan, Laura. "Shakespeare in the Bush". Chapter 5 of James Spradley and David W McCurdy (eds.), Conformity and Conflict: Readings in Cultural Anthropology, fourteenth edition. London, Pearson, 2011.

Booth, Anthony Robert, and Darrell P. Rowbottom (eds.). Intuitions. Oxford, Oxford University Press, 2014.

https://doi.org/10.1093/acprof:oso/9780199609192.001.0001

Cappelen, Herman. Philosophy without Intuitions. Oxford, Oxford University Press, 2012.

https://doi.org/10.1093/acprof:oso/9780199644865.001.0001

Elton, Geoffrey R. The Practice of History, second edition with an afterword by Richard J. Evans. Oxford, Blackwell, 2002. (First edition, Sydney, NSW, Sydney University Press, 1967.)

Fischer, John Martin, Robert Kane, Derk Pereboom and Manuel Vargas. Four Views on Free Will. Oxford, Blackwell, 2007.

Geach, Peter T. "Assertion". Philosophical Review, volume 74, number 4, 1965, pages 449-465.

https://doi.org/10.2307/2183123

Knobe, Joshua. "Philosophical Intuitions are Surprisingly Robust Across Demographic Differences". Epistemology and the Philosophy of Science, volume 56, number 2, 2019, pages 29-36.

https://doi.org/10.5840/eps201956225

Mackie, John L. Ethics: Inventing Right and Wrong. Harmondsworth, Penguin, 1977.

Parfit, Derek. Reasons and Persons, corrected edition. Oxford, Clarendon Press, 1987.

Plato. Meno.

Roeser, Sabine. Moral Emotions and Intuitions. Basingstoke, Palgrave Macmillan, 2011.

Schopenhauer, Arthur. Prize Essay on the Freedom of the Will, edited by Günter Zöller, translated by Eric F. J. Payne. Cambridge, Cambridge University Press, 1999.

Stich, Stephen P., and Edouard Machery. "Demographic Differences in Philosophical Intuition: a Reply to Joshua Knobe". Review of Philosophy and Psychology, published online 2022, no volume or part number assigned at the time of writing.

https://doi.org/10.1007/s13164-021-00609-7

Stratton-Lake, Philip. "Intuitionism in Ethics". Stanford Encyclopedia of Philosophy, 2020.

https://plato.stanford.edu/entries/intuitionism-ethics/

Tegmark, Max. Our Mathematical Universe: My Quest for the Ultimate Nature of Reality. New York, NY, Alfred A. Knopf, 2014.

Tobia, Kevin (ed.). Experimental Philosophy of Identity and the Self. London, Bloomsbury, 2022.

Chatbots, AI, education, and research

Some sophisticated chatbots are now available. One is ChatGPT, which is being incorporated into Microsoft's search engine Bing. Another is Bard, which is being incorporated into Google's search engine. Connection with web searches allows bots to do more than write in a human style. They can also gather information on which to base what they write.

This post discusses some implications of chatbots, and of artificial intelligence (AI) more generally, for education at the university level and for academic research. We shall start with education, because that is the area in which the greater number of people will be affected directly. Some of what we shall say in relation to education will carry over to research. This is a natural consequence of the fact that at the university level, students should develop skills of enquiry, analysis and the reporting of conclusions that are also needed in research.

We shall assume that AI will get a good deal better than it is at the moment. In particular, we can expect systems that meet the needs of specific disciplines to develop. Such systems would draw on appropriate corpora of material both in their training and to respond to queries, and would reason their way to conclusions and present the results of their work in ways that were appropriate to their disciplines.

References to publications cited are given at the end of the post.

Education - developing students' minds

When a student finishes a course of education, they should not only have acquired information. They should also have developed ways to think that would be effective if they wanted to respond to novel situations or to advance their understanding under their own steam. Even if they did not continue to work in the disciplines in which they had been educated, such skills would be useful. Many careers demand an ability to think clearly and to respond intelligently to novel challenges. And there is also the less practical but still vital point that the ability to think for oneself is important to a flourishing life.

Ways to think are developed by making students go through the process of finding information, organizing it, drawing conclusions, and writing up the results of their work. They must start with problems to solve or essay questions. Then they must work in laboratories, or use libraries and online repositories to find relevant source material (whether experimental data, historical evidence, or academic papers). They must think through what they have found, draw conclusions, and produce solutions to the problems set or essays on the relevant topics.

If critical stages in this process were outsourced to computers, educational benefit would be lost. But which stages should not be outsourced, and which stages could be outsourced harmlessly?

The traditional function of search engines, to show what material is available on a given topic, seems harmless. It looks like merely a faster version of the old custom of trawling through bibliographies or the footnotes in books and articles which surveyed the relevant field. 

Search engines do however have an advantage besides speed over old methods. A search engine will typically put what it thinks will be the most useful results first. This is helpful, so long as the search engine's criteria of usefulness match the student's needs. But even then, it can detract from the training that the student should receive in judging usefulness for themselves.

The latest generation of search engines, with ChatGPT, Bard or the like built in, take this one step further. If students can express their requirements with sufficient clarity and precision to prompt the search engines appropriately, the results can be well targeted. Rather than giving several pages of links, the search engines can provide statements in response to questions. References to support those statements could be supplied along with the statements, but if not, they would be reasonably easy to find by making further searches. The assistance of search engines in going directly to answers to questions might be helpful, but it would also take away practice in reviewing available items of material, judging their relative reliability and importance, and choosing which items to use.

Moving on to putting material to work, developing arguments, and reaching conclusions, there is not much sign that the new generation of chatbots will in their current form be helpful.

This reflects the way in which they work. (A good explanation is given by Nate Chambers of the US Naval Academy in his video ChatGPT - Friendly Overview for Educators.) When working out what to say, they rely on knowledge of how words are associated in the many texts that can be found online. They get the associations of words right, but they do not first formulate sophisticated ideas and then work out the best ways to express them. Texts they produce tend to be recitals of relevant bits of information in some reasonably sensible order, rather than arguments that move from information to conclusions which are supported by the intelligent combination of disparate pieces of information or by other sophisticated reasoning.

So training in the important skill of constructing arguments would not seem to be put at risk by students' use of chatbots. But there is other software to consider, software which can engage in sophisticated reasoning. AI is for example used in mathematics (a web search on "theorem prover" will turn up examples), and in chemistry (Baum et al., "Artificial Intelligence in Chemistry: Current Trends and Future Directions").

If students relied on software like that to respond to assignments set by their professors, they would not acquire the reasoning skills they should acquire. And it would be no answer to say that if they went on to careers in research, they would always be able to rely on such tools. If someone had not in their training reasoned their own way through problems, they would not understand what the conclusions provided by AI systems really meant. Then they would not be able to appreciate the strengths and the weaknesses of those conclusions. They would also be unable to assess the level of confidence one should have in the conclusions, because they would not have a proper grasp of the processes by which the conclusions were reached and what false steps might have been made.

Software that does all or most of the reasoning required to reach interesting conclusions is not to be expected everywhere. We can expect it to be far more widespread in mathematics and the natural sciences than in other disciplines. In the social sciences and the humanities, there may be data which are open to being marshalled into a form that is suitable for mathematical analysis, and such analysis may yield robust conclusions. Some such analyses may be found under the rubric of digital humanities. But while such analyses might be devised and conducted entirely by AI systems, and their results might be eminently publishable, those results would be unlikely to amount to conclusions that conferred insights of real value, at least not in the humanities and to some extent not in the social sciences either. Humane interpretation of the sort that may confer Verstehen is still needed, and that does not yet appear to be a forte of AI.

Having said that, where the thinking that AI might be expected to do would suffice to reach interesting conclusions, the combination of such a system with a language system to write up the results could be powerful. All the work could be done for the student. 

Such a combination will surely be feasible in the near future in mathematics and the natural sciences, where the results of reasoning are unambiguous and there are standard ways to express both the reasoning and the results. There are already systems, such as SciNote's Manuscript Writer, which will draft papers so long as their users do some initial work in organizing the available information. That package is, as its maker states, not yet up to writing the discussion sections of papers, but we should not suppose that accomplishment to be far away. To write discussion sections, a system would need to have a sense of what was really interesting and of how results might be developed further. But we should not suppose such a sense to remain beyond AI systems for long.

In other disciplines, and particularly in the humanities, it is much less clear that such a combination of AI reasoner and AI writer would be feasible in the near future. The results of reasoning are more prone to ambiguity, and ways to express those results are less standardized. It might also not be feasible to break down the task of building a complete system into two manageable parts. The distinction between reasoning and final expression for publication is never absolute, but it is decidedly hazy, involving substantial influences in both directions, outside the natural sciences.

Another issue is that in work of some types the expression of results, as well as the reasoning, need to be apt to confer Verstehen. As with reasoning, that does not yet appear to be a forte of AI. The main obstacle, in relation both to reasoning and to expression, may be that AI systems do not lead human lives. They do not have a grasp of humanity from the inside. Human experience might one day be fabricated for them, but we are some way off that at the moment. (There is more on the themes of Verstehen and the human point of view in Baron, Confidence in Claims, particularly section 5.6.) Having said that, AI may well improve in this respect. Systems may come to align their ways of thinking with human ways by having human beings rank their responses to questions, a method that is already in use to help them to get better at answering questions in ways that human beings find helpful.

We may conclude that AI could put the development of any of the skills that students should acquire, from gathering and organizing information to drawing conclusions and expressing them, at risk, but that the relevant software would vary from skill to skill and the threat would probably become serious in the natural sciences first, then in the social sciences, and finally in the humanities.

Education - grading

There has been concern among professors that students will use ChatGPT to write essays. So far, the concern seems to be exaggerated. If ChatGPT is offered a typical essay title, the results are often poor assemblies of facts without any threads of argument, and are sometimes laughably incorrect even at the merely factual level. But chatbots will get better, and may merge with the reasoning software we have already mentioned to yield complete systems which could produce work that would improperly earn decent grades for students who submitted it as their own.

Some remedies have been suggested. One remedy would be to make grades depend on old-fashioned supervised examinations, perhaps on computers provided by universities so as to compensate for the modern loss of the skill of handwriting while not allowing the use of students' own computers which could not reliably be checked for software that would provide improper assistance. Another remedy would be to make students write their assignments on documents that automatically tracked the history of changes so that professors could check that there had been a realistic amount of crossing out and re-writing, something which would not be seen if work produced elsewhere had simply been pasted in. A third remedy would be to quiz students on the work submitted, to see whether they had actually thought about the ideas in their work. A fourth remedy would be to set multi-part assignments, with responses to each part to be submitted before the next part was disclosed to students. This idea relies on the fact that current software finds it difficult to develop arguments over several stages while maintaining coherence. Finally, anti-plagiarism software is already being developed to spot work written by chatbots, although it is not clear whether it will be possible for detection software to keep up with ever more sophisticated reasoning and writing software.

Alternatively, AI might shock educators into the abolition of grading. It is not that AI would adequately motivate the abolition of grading. Rather, it could merely be the trigger for such a radical move.

There are things to be said against such a move. Grades may be motivators. Employers like to see what grades potential employees have achieved. And there are some professions, such as medicine, in which it would be dangerous to let people work if their knowledge and skills had not been measured to ensure that they were adequate.

There are however things to be said in favour of the abolition of grading. Alfie Kohn has argued that a system of grading has the disadvantage of controlling students and encouraging conformity (Kohn, Punished by Rewards). Robert Pirsig puts into the mouth of his character Phaedrus an inspiring account of how students at all levels can actually work harder and do better when grades are abolished. As he puts it when commenting on the old system:

"Schools teach you to imitate. If you don’t imitate what the teacher wants you get a bad grade. Here, in college, it was more sophisticated, of course; you were supposed to imitate the teacher in such a way as to convince the teacher you were not imitating, but taking the essence of the instruction and going ahead with it on your own. That got you A's. Originality on the other hand could get you anything - from A to F. The whole grading system cautioned against it." (Pirsig, Zen and the Art of Motorcycle Maintenance, chapter 16)

So the end of grading could have its advantages, particularly in the humanities where there is, even at the undergraduate level, no one right way to approach a topic. (This is not to say that all ways would be acceptable. Some would clearly be wrong. And at the detailed level of how to check things like the reliability of sources, there may be very little choice of acceptable ways to work.)

Research - the process

AI that collates information, reasons forward to conclusions, and expresses the results can be expected to play a considerable role in research in the reasonably near future. There is however some way to go. Current systems seem to be good at editing text but not so good at generating it in a way that ensures the text reflects the evidence. Their specialist knowledge is insufficient. And as noted above, they cannot yet write sensible discussion sections of scientific papers, let alone sensible papers in the humanities. (For an outline of current capabilities and shortcomings see Stokel-Walker and Van Noorden, "What ChatGPT and Generative AI Mean for Science".)

Influences of AI on research will be likely to parallel influences on education. The burdens of tracking down, weighing up and organizing existing information, analysing new data, reasoning forward to interesting conclusions, and expressing the results of all this work might be taken off the shoulders of researchers in the near future, leaving them only with the jobs of deciding what to investigate and then reviewing finished work to make sure that the AI had considered a suitably wide range of evidence, that it had reasoned sensibly, and that the conclusions made sense. Having said that, the issues would be different.

There could be very great benefit if more research got done, particularly when research addressed pressing needs such as the need for new medical treatments or for greater efficiency in engineering projects. There would be no such benefit in the context of education rather than research, although if the use of AI made education more efficient students might progress to research sooner.

There is also the point that if an AI system absorbed the content of all the research being done in a given area and interacted with human researchers, this could create a hive mind of pooled expertise and knowledge which would be more effective than the hive mind that is currently created by people reading one another's papers and meeting at conferences. (We here mean a hive mind in the positive sense of sharing expertise and knowledge, not in the negative sense of conformity to a majority view.)

The development of minds by thinking through problems would be less important at the level of research, because minds should already have been developed through education. The loss of opportunities for development on account of the use of AI would however still be a loss. Every mind has room for improvement. In addition to concern about the continued development of general skills, there is the point that not actively reasoning in the light of new research would reduce the extent to which someone came to grasp that research and its significance. Finally, only a researcher who had a firm grasp of the state of the discipline, including the most recent advances, would be able to judge properly whether the results of AI's work passed the important test of making sense.

A concern that is related to the development of minds is that there would be a risk that any novel methods of reasoning by AI would not be properly understood, leading to the misinterpretation of conclusions or a failure to estimate their robustness properly. Well-established techniques should have been covered in a researcher's student days, but new techniques would be developed. A researcher who had never gone through the process of applying them laboriously using pen and paper might very easily not have a proper grasp of what they achieved, how reliably they achieved it, and what they did not achieve.

Another concern is that the processes of reasoning by AI systems could be opaque to human researchers. This would be an instance of the AI black box problem. Reasoning might be spelt out in the presentation of work, but there would be a risk that the reasoning as represented was not the actual reasoning. If satisfactory reasoning were set out, that might appear to address the issue. But that reasoning might not in fact be related appropriately to the evidence (while the internal reasoning was so related), and this might not be noticed by human researchers reviewing the work.

One specific form of opacity that should concern us is the risk that when AI systems search for material, they may be influenced by inappropriate criteria. Search engines can already give high rankings to links that it is in their commercial interests to favour, or can push down the rankings material that is disfavoured by the political establishment (the practice of shadow-banning). If AI used by researchers did the same sort of thing, research could be skewed in wholly improper ways.

Research - credit

Researchers like to get credit for their work, and are annoyed when other people take credit for work that is not their own. Names on publications need to be the right ones, and if any material is taken from someone else's work it must be attributed in a footnote. One reason for this ethos is that non-compliance would be considered to amount to bad manners, or theft, or something in between these extremes. Another reason is that jobs, promotion and funding depend on the work one has done, so each person needs to be able to take credit for their own work, whether it is published by them or used by others.

Now suppose that a researcher had relied on an AI system to organize material and reason the way to conclusions, rather than merely to find material. And suppose compliance with the minimal requirement that the use of AI should be disclosed. How should its use affect the allocation of credit?

One might argue that the use of AI was not in principle different from the use of any other tool. In many disciplines, the use of sophisticated computer systems is routine and is not thought to give rise to special issues of attribution. Even in the pre-computer age, people relied on bibliographies and on other people's footnotes to track down material, and that was not thought to lessen the credit due to researchers who relied on such aids.

On the other hand, AI systems that were good enough to help with reasoning would learn as they went along, pooling knowledge gained in their use by all researchers in a given field. Then any particular user would rely indirectly on the work of other researchers, and might easily be unaware of which other researchers' work was involved. There could be no more than a general acknowledgement, enough to tell the world that not everything was the author's own work but not enough to give credit to specific previous researchers.

We should not however think that such a failure to credit previous researchers would be entirely new. At the moment, identifiable contributions by others are expected to be acknowledged. But there is also the accumulated wisdom of a discipline, which may be called common knowledge or prevalent ways to think. That wisdom depends on the contributions of many researchers who are not likely to be acknowledged. One may stand on the shoulders of people of middling stature without knowing who they were, and it is not thought improper to fail to acknowledge them by name. The new difficulty that would be created by the use of AI to produce reasoning would not lie there. It would instead be be that contributions which would be easy to acknowledge if one worked in traditional ways might accidentally go unacknowledged. On the other hand, one might get an AI system to track such contributions and generate appropriate footnotes.

We now turn to the significance of credit when allocating jobs, promotion and funding. The basic idea is a sensible one. The aim should be to employ, promote and fund people who produced the best work, and ensure that it was their own work without undisclosed reliance on other people's work. How might the use of AI to reason the way to conclusions matter? We shall again assume that its use would be disclosed.

Given that the difference made by the use of AI would vary from one piece of work to another, and that a researcher who routinely relied on AI might in fact have the talent to do just as well without its help, it might become harder to decide reliably between candidates. On the other hand, such decisions are unlikely to be particularly reliable as it is, at least not when choices are made between several candidates all of whom are of high quality. So any loss might not be great.

Finally, there is the question of credit for the clear and elegant description of work and expression of conclusions. AI that was rather more sophisticated than currently available could do this work, and a human author might take credit. Fortunately it is not normal to credit other researchers for one's style of writing anyway, so there would be no appropriate footnotes giving credit to others to be omitted even if the AI had developed its style by reviewing the work of many researchers. And when it comes to the allocation of jobs, promotion and funding, reasoning should in any case be a good deal more important than style.

References

Baron, Richard. Confidence in Claims. CreateSpace, 2015.

https://rbphilo.com/confidence.html

Baum, Zachary J., Xiang Yu, Philippe Y. Ayala, Yanan Zhao, Steven P. Watkins, and Qiongqiong Zhou. "Artificial Intelligence in Chemistry: Current Trends and Future Directions". Journal of Chemical Information and Modelling, volume 61, number 7, 2021, pages 3197-3212. 

https://doi.org/10.1021/acs.jcim.1c00619

Chambers, Nate. ChatGPT - Friendly Overview for Educators.

https://www.youtube.com/watch?v=fMiYNrjDPyI

Kohn, Alfie. Punished by Rewards: The Trouble with Gold Stars, Incentive Plans, A's, Praise, and Other Bribes, with a new afterword by the author. Boston, MA, Houghton Mifflin, 1999.

Pirsig, Robert M. Zen and the Art of Motorcycle Maintenance: An Inquiry into Values, 25th anniversary edition. London, Vintage, 1999.

SciNote Manuscript Writer.

https://www.scinote.net/manuscript-writer/

Stokel-Walker, Chris, and Richard Van Noorden. "What ChatGPT and Generative AI Mean for Science" (corrected version of 8 February 2023). Nature, volume 614, number 7947, 2023, pages 214-216.

https://doi.org/10.1038/d41586-023-00340-6

Why is there something rather than nothing?

 

This was the question at our Cambridge philosophy café on 22 January 2023. The first impression is that it is both a question that demands a decent answer, and a question that cannot have one. This post does not provide an answer. Instead, it sketches some of the territory.

Why is an answer demanded?

There are items of many types in the world (meaning not just the world as it is today, but the world with all its history). There are physical objects, events, relationships of space and time (or of spacetime when we focus on physics), laws of nature, mathematical results,  thoughts, feelings, and so on. We may be more or less generous in what we regard as an item in the world. But whenever we admit something as an item, we can ask why it is in the world. And we tend to think either that an answer will already be known, or that the discovery of an answer would be perfectly conceivable. Even if we do not have much hope that an answer will in fact emerge, for example where historical records have been lost, we still think that an answer could have been found if things had been different in perfectly identifiable ways. At the extreme, we might forgo even that hope and say that no answer could be found, but even then, we would think that there was some unknowable reason for the presence of the item in the world. 

To put all this in traditional philosophical terms, we have a strong inclination to subscribe to the principle of sufficient reason. And we are disturbed when quantum mechanics suggests that there may be no reason why some observations rather than others come to be made. We are left hoping that physics will advance to restore compliance with the principle. The hope may be forlorn, but it is there.

If we routinely find reasons for the presence of items in the world, and if we can put quantum mechanical worries to one side by noting that they do not extend to everyday items when characterized in the macroscopic terms we in fact use to individuate and describe those items, it would seem reasonable to ask why the whole ensemble is present (not present in the world, for the ensemble is the world, but simply present). And the question of why is there something rather than nothing is a less demanding question than that of why the particular ensemble we find is present, in that an answer to the latter question would automatically be an answer to the former one but not vice versa. Moreover, not only would the question seem reasonable. A response that it should not be asked, or could not be answered, would seem to be unreasonable.

Types of explanans

Our explanandum is the existence of some non-empty world or other (we do not need to explain its actual make-up). Our explanans could be causal and within the world, causal and external to the world, or non-causal.

By way of background, we shall make some remarks on ways to explain the existence of items within the world. Then we shall consider causal options, followed by non-causal options. Finally, we shall look at the option of dismissing the question.

Ways to explain the existence of items within the world

We can find reasons why individual items are in the world.

Sometimes reasons are directly causal, as when stresses between continental plates cause earthquakes, or the values of certain physical constants together with some conditions in the early universe determine which forms of matter have come into being.

Sometimes reasons are indirectly causal. We may for example say that a particular taxonomy of animals has been created because of the causal influences that have produced the actual variety of animals. Or we may say that a thought exists because of causal influences on neurons. Or we may say that an emotion in general (as distinct from individual instances) exists because causal patterns in the world are such as to generate particular types of human reaction in particular circumstances, where those reactions are reasonably systematic.

Sometimes reasons are not causal at all, as when we identify the reason why some mathematical statement is a theorem. 

There are also items that would be amenable both to non-causal explanation and to indirectly causal explanation. For example, the existence of Mannerism as an item in the history of European art could be explained non-causally by the presence of distinctive formal features in the relevant paintings and sculptures. And it could also be explained causally, albeit in an indirect way, by an analysis of the thoughts in the minds of artists and their patrons, in turn explained by a causal history of their neurons and of preceding developments which created the artistic context.

Internal causes of the whole

Now let us consider the whole universe, in all its history. The prospects for a causal explanation that relies on causes within the world are not bright.

Causes generally precede their effects. A cause might arise at the same time as its effect, but even then we require a direction of causation. If one thing caused another, it was in fact the case that the second thing did not cause the first one, even if in other circumstances the second thing could have caused the first one.

Moreover, we do not expect A to cause B to cause C to cause A. In mathematical terms, we want the causal connections between things to form a directed acyclic graph. In particular, cycles directly from something to itself are ruled out. No self-caused things are permitted.  We can however allow that A might cause B on some small scale, which then caused A to grow, which caused B to grow, and so on. Causal feedback is allowed.

(As usual, quantum mechanics complicates things. See for example the analysis and the references to earlier work in Barrett, Lorenz and Oreshkov, "Cyclic Quantum Causal Models", at https://doi.org/10.1038/s41467-020-20456-x.)

As soon as we have a direction of causation, we get causal chains to trace backward. We would expect to find something at the head of a chain into which all chains merged as we worked backward, or maybe several things at the heads of various chains which did not merge. A head item would explain everything that followed in its chain, so if it went unexplained, we would still not have explained why there was something rather than nothing because explanations for the existence of other things would be contingent on its existence. And everything that was not at the head of a chain would have to stand in some chain or other, so its existence would only have such a contingent explanation.

At this point we may remark that telling us that empty space has non-zero energy does not answer the question, at least not in the obsessive form in which philosophers are apt to pose it. Empty space with its scientifically determinable properties is not nothing, but a something that could have led in a causal way to what we see today. It has been said that when Lawrence Krauss published A Universe from Nothing, it should have been entitled A Universe from Not Very Much. And to a philosopher, that criticism has bite. (Krauss is however fully aware of the issue. He discusses it in chapter 9. On page 149, he acknowledges that he takes empty space and the laws of physics to exist within his "nothing".)

We must tread carefully here. It would be possible for everything to be caused without there being anything uncaused, in the way that every positive real number has a predecessor positive real number (in fact, infinitely many of them) without there being any first one.

More interestingly, we may need to be careful because temporal precedence relies on there being time, and strange things may have happened with spacetime at the very beginning. Temporal precedence may not be the only type of precedence to give a direction of causation, if we allow for causes to be simultaneous with their effects. But it is the prevalent type. If the notion of temporal precedence were to get into difficulties in the context of the early universe, cosmologists of that early stage might be able to offer us a loophole that would allow a cause of things which was within the universe to be fully explanatory.

There would also be an argument to be had over whether one should see the quest for a causal reason for there being things in general in the same terms as the quest for a causal reason for there being some particular thing or other, even an unspecified thing that would stand as a representative member of things in general.

While one might raise such doubts about the easy argument that there cannot be, within the universe, a cause of the collectivity of things, there is enough room for worry that we should not expect to find such a cause.

External causes of the whole

Perhaps there could be a causal explanation which would be saved by not being within the world, so that it did not itself stand in need of a worldly causal explanation.

Sadly, this idea looks like a non-starter. Something that was causally effective but outside the world would look suspiciously like a god. And what could reasonably be substantiated about a supposed god would be so little as to make the supposition of one nothing more than a place-holder for an answer. Theists may rely on holy texts to describe their gods, but those texts are claims, not evidence. And in the absence of substantiated properties of a god or gods, the assertion of their existence does no more to answer our question as to why there is something rather than nothing than to say "If we had an answer, it would go here". Leibniz, who made an explicit connection with the principle of sufficient reason, may have thought he had done more (Principles of Nature and of Grace, 7-9; Monadology, 36-39). But he had not.

Non-causal options

It is time to look at explanations which would avoid the requirement to respect a unidirectional relation from each specific explanans to its specific explanandum. They would be non-causal explanations.

Such explanations would comprise facts about the world, or facts about the abstract realm independently of any connection it might have to the world, rather than comprising things within or outside the world. The facts might mention things, but they would not be those things. Given the difficulty of conceiving of facts about things outside the world, aside from unsubstantiated and possibly incoherent claims about supposed gods, we need not distinguish between internal and external facts in the way that we distinguished between internal and external causes. But we must recognize that facts may be about abstract entities, any relationship or lack of relationship of which to the world is not given and is not to be presumed (beyond the observation that logic and mathematics are readily applicable to the world). Such abstract facts may include structures of relationships in which the specific relata are unimportant.

We seek facts that would make the existence of something rather than nothing unsurprising. There are options.

Many possible universes

One option is the claim that there are a great many possible universes. An easy argument would be that since there would be only one possible empty universe and a vast number of possible non-empty universes, it is no surprise that a non-empty universe should turn up. All that would then be lacking would be an explanation of why any possible universe at all was actual. If some random one were actual, it would probably be a non-empty one.

This argument would rely on the claim that there was only one empty possible universe (or at least, not many of them). Fortunately, within the realm of the merely possible, the claim of a single empty universe could be made just as, in mathematics, there is only one empty set. It is only in the concrete world that there are many empty boxes.

Many actual universes

Another option is the multiverse claim that there are in fact a great many universes, all but one of them beyond our ken. It would not be necessary to argue that most of them would be non-empty. Any one non-empty universe would suffice. But this argument would only give a route from there being many universes to there being something rather than nothing. It would leave unanswered the question of why there actually were many universes.

Either this option or the option of many possible universes could be backed up by a weak anthropic principle to the effect that whatever we observe will be a universe sufficiently complex to support conscious life, and therefore not empty. An empty universe might be a possibility, or there might be one or more empty universes among a number of actual universes, but we could not be in an empty universe.

The universe as an abstract structure

A third option would be to say that the universe was itself some abstract structure, not dependent on anything for its existence.

The leading example is the mathematical universe hypothesis that Max Tegmark has set out in Our Mathematical Universe (2014), but that is better and more briefly explained in his paper "The Mathematical Universe" from 2007, available at https://arxiv.org/abs/0704.0646 (take care to obtain version 2, dated 8 October 2007).

In Tegmark's view, the universe actually is a mathematical structure. This claim goes beyond the observation that the universe is amenable to mathematical characterization. Moreover, we are substructures within that structure who are conscious and self-conscious by virtue of the complexity of those substructures.

Tegmark's approach has the advantage that it is plausible, although not uncontentious, to think that mathematical structures simply exist, independently of anything else's existence. In particular, and importantly for him, they can be seen as existing independently of any human predisposition to think in particular ways (for example to think in terms of objects and causation). They are all form, definable in strictly mathematical terms, and no non-formal content. And this is argued to be enough to characterize any universe. As physics digs deeper and deeper into the nature of reality, it identifies symmetries and conservation laws which say everything that physicists feel the need to say. The need for a separate stage of explaining how the mathematical structures identified give rise to what we perceive does not detract from the sufficiency of physics without that extra stage.

Tegmark's approach can however be challenged.

An apparent difficulty is that we need to explain how it is that the particular mathematical structures we find here are instantiated, rather than all possible structures being instantiated to form each and every possible universe. (He does posit a multiverse.) The problem is that once mathematics gets going, even with an empty set and minimal set theory, there is nothing within itself to stop its expansion. We would get all of mathematics everywhere, subject to a few large-scale decision points such as whether to adopt classical or constructivist mathematics or whether to adopt the axiom of choice.

This apparent difficulty is easily addressed. Tegmark thinks that what comprises any given universe is not a theory but a model of a theory. That allows for variation between universes, especially if we allow universes to comprise models of parts of theories.

This does not however leave an entirely satisfactory picture. A clue to the difficulty lies in Tegmark's invocation of the mathematical principle that "same up to isomorphism" amounts to "same" (the 2007 paper, section II.D). If our universe is isomorphic to a mathematical structure, then in his view it is that structure.

The problem is that "same up to isomorphism" does not amount to "same" in the everyday sense of "same". When we think in physical terms, we are happy with the idea of several distinct but isomorphic things. We could even imagine several isomorphic universes within a multiverse, so long as the multiverse was realised in a physical or quasi-physical way and not in a purely mathematical way. We would have to presuppose Tegmark's conclusion that we should think of the universe as a mathematical object in order to require ourselves to think of sameness in the mathematical way. 

Dismissing the question

The option of dismissing the question of why there is something rather than nothing remains.

One could simply take the existence of something rather than nothing as a brute fact.

One could add that the principle of sufficient reason did not have to be accepted. There are after all things in the quantum world which, so far as we can tell, simply happen. More precisely, there are observations which are such that we cannot currently give any sufficient reason why a specific observation was made rather than an alternative.

One could dismiss all questions as to why things were as they were and and simply ask how things came to be as they were. An argument for doing so, given by Lawrence Krauss, is that why-questions assume that there is some purpose to be found, and that there is no sign of any such purpose at the level of the universe (A Universe from Nothing, page 143). But it is not clear that this is so. A why-question need not assume purpose. It may be a rephrased how-question which, in the context of asking why there is something rather than nothing, carries the implication that one is going to go on asking how-questions until one reaches something that plainly has to be the case so that the stream of how-questions can come to a stop.

One could challenge the formulation of the question, on the ground that it was not possible to talk about nothing. But that argument might easily not succeed. While there might be nothing to which to refer, one could talk about the falsity of every proposition of first-order predicate logic which started with an existential quantifier, the variable of which actually bound an occurrence later in the proposition. (This is only a sketch of a solution. It would need to be worked out in detail.) Or one could point out that we have no difficulty in thinking mathematically of the contents of the empty set, even if philosophers have found that natural languages run into difficulties over such notions.

Finally, one could treat the existence of something rather than nothing as a mystery. In the words of Wittgenstein (Tractatus 6.44), "Nicht wie die Welt ist, ist das Mystische, sondern daß sie ist". But that would seem to amount to no more than accepting the existence of something rather than nothing as a brute fact and taking up a particular emotional attitude to the fact, unless one had good grounds to think that the mystical was itself a repository of information that was so far inaccessible and might eternally remain so. That would be a hope without justification.

The humerus side of life

In November, when with a friend on our way back from seeing someone depart for Paradise by way of Kensal Green, I tripped and fell on a railway platform and sustained a fracture to my left arm, at the upper end of the humerus. This must rank very low on the scale of injuries requiring hospital treatment, and indeed the orthopaedic surgeon who first looked at it chose to put the arm in a sling and let it heal, rather than resorting to surgery. But the injury was still enough to provoke a few thoughts, which I record here.

Before doing so, however, I would like to record my very great thanks to everyone who helped, both friends and transport and medical staff. Among the latter I number (in order of encounter):

1. the platform staff at Oxford Circus station. They were with me within 30 seconds of my falling, and made sure I was alright and able to continue safely (in the company of the friend who was with me) to where I could get medical help;

2. the pharmacist at Boots on Kingsway whom I first approached in search of a sling and painkillers, who told me that there might be a fracture and sent me to hospital;

3. the Accident and Emergency and Urgent Treatment Centre staff at University College Hospital. I was passed smoothly round the necessary doctors, orthopaedic surgeon, nurses and radiographers, and while work in such departments must be stressful, they were at all times calm and friendly;

4. the staff of the Fracture Clinic at the same hospital, which I visited three times from December up to early February. Again, orthopaedic surgeons, nurses, radiographers and administrative staff ran the operation very smoothly;

5. the radiographer and the orthopaedic surgeon who took a look to make sure there was no displacement when I was in Hong Kong for a while in December;

6. the physiotherapist whose clinic at University College Hospital I now attend, and probably will attend for a little while to come. Each time I go I get a friendly and efficient review of progress, and am sent away with clear instructions as to which exercises to practise.

The National Health Service sometimes gets a bad press. This has been by far my most significant encounter with the Service to date, and I want to say out loud that they have been absolutely superb. The next time you read a story saying that they have got something wrong, please bear in mind that they get things right thousands of times a day, and the newspapers hardly ever notice.

Now to the thoughts provoked by the fracture. None of them is original, but some of them only come to mind in abnormal circumstances.

1. What happens is so contingent on trivial details. If I had stepped off the train a little bit differently, my course might have been to one side of where it was, or my feet might have fallen at each pace a few centimetres behind where they in fact fell, and the trip might have been avoided. As a corollary to this, there was no practical way to see the risk coming. Even approximations to the predictive power of Laplace's demon are unavailable.

2. Following on from this first thought, trivial differences can make a big difference to the next few months. My reading and writing were interrupted and then slowed down, and a couple of talks I was due to give had to be cancelled. (A trip to Hong Kong and Singapore, on the other hand, starting ten days after the accident, went ahead, although I don't think the orthopaedic surgeon in the London fracture clinic was very happy about that.) But the effects need not be life-changing. My life is now back to what it probably would have been if the fracture had not occurred. Of course something life-changing might have happened at the talks, had I given them, but that is mere speculation, and the possibility is too ill-specified for there to be any fact of the matter about whether something life-changing would have happened. (For comparison, I buy a lucky dip lottery ticket for each Saturday's draw, but given that the numbers are generated randomly and depend on the shop one uses and the precise time of purchase, there is no fact of the matter as to whether, when I happen to miss a Saturday, I would have won had I bought a ticket.)

3. It is sometimes said that one should get on and do the important things in life, because one might for all one knows die quite soon. This argument is less persuasive than it used to be, at least when addressed to people who are young or in middle age, because while there are early deaths, the probability of dying when young or in middle age is much lower than it used to be. We should perhaps replace the argument with one which turns on the fact that one might lose some important abilities relatively early in life. My fracture is healing nicely, but I could easily have suffered a permanent injury. And in a world that is structured around people with four fully functioning limbs, good eyesight, and so on, moderate injuries can impose significant limitations, despite provision being made for the disabled. To take a trivial example, when one arm is out of action or cannot be used to apply any significant force, it takes much longer than normal to get dressed, and you have to find someone else to do up your shoelaces.

4. It is remarkable how little one notices about one's body until something goes wrong. The big lesson for me was that no body part is an island. Everything is connected under the skin. Thus for the first week, any movement of the upper left arm was painful. So I took care to do everything with the right arm. But certain movements of the right arm led to pain in the left arm. And if I had to pick something up from the floor, crouching down (moving only the legs, not the arms) also led to pain in the left arm.