Introduction

The point of departure for this paper is the following quote from the summary of the judgement of the Court of Appeal in Re S (Adult Patient: Sterilisation):Footnote 1

Held, allowing the appeal, that while a number of different courses might be lawful in any particular case, there could only logically be one best option and it was for the court to decide which that was; that, once satisfied that the proposed treatment options were within the range of acceptable opinion among competent and responsible practitioners, the court should move on to the wider and paramount consideration of which of them was in the patient’s best interests;…Footnote 2

Here Dame Elisabeth Butler-Sloss puts forward what to us seem to be two extremely problematic views: (1) that there can only logically be one best option and (2) that the court should decide which of the available options is the best by considering which of them is in the patient’s best interest.

In this paper we will first consider what theories of “best interest” and rational decision-making that could support the first contention made by Butler-Sloss, and explore the implications of these theories.

We will then move on to consider what ethical theories that are compatible with such a view and explore their implications.

In the second part of the paper we will explore a range of possible criticisms of the view that “there could only logically be one best option”.

And in the third part we will criticise the view that the court is in a good position to decide what the patient’s best interest is.

In the fourth and final part we will first consider how intelligent persons could ever come to adopt the views put forward by Butler-Sloss and cited with approval in later judgements; and second put forward a reconstructive proposal aimed at saving whatever is sound in the “best interest” conception.

Rational Decision-Making and the Best Option

The idea that there can only logically be one best option in a given decision-making situation is closely aligned to the classic von Neumann–Morgenstern conception of rational choice [4]. According to the von Neumann–Morgenstern approach rational decision makers will make the decision that maximises their expected personal utility. A decision maker should therefore enumerate all the possible action available, identify all the possible outcomes of these actions, give a value to each outcome and estimate the likelihood/probabilities of a specific outcome eventuating given a specific action. Based on this it is then possible to calculate which action is best. We will here leave aside the question of whether it is logically necessary to be a utility maximiser and accept, for the sake of argument that it is. In that case it follows that a rational decision maker is logically compelled to choose the best option, on pain of irrationality.

However, the von Neumann–Morgenstern approach does not guarantee that there is one, unique best choice. There may be several actions that have the same expected utility, even under conditions of full information (Fig. 1). So it is strictly speaking not true that there “could only logically be one best option”. There could logically be any number of best options, from one upwards.

Fig. 1
figure 1

 More than one "best option"

Full size image

It is also worth noting that the outcome that is valued most highly may not be linked to the action that leads to maximising expected utility (Fig. 2), and that the choice that maximises expected utility may contain very negative outcomes in its outcome set (Fig. 3). The last problem may lead a risk-averse decision maker to adopt a maximin approach, choosing as the best option the one that guarantees the minimum loss. And although the standard von Neumann–Morgenstern approach holds that risk aversivity is irrational, most would accept that it is an open question whether some degree of risk aversivity cannot be seen as rational.Footnote 3

Fig. 2
figure 2

 Best outcome not in option that maximises expected utility

Full size image
Fig. 3
figure 3

 A problem for the risk-averse decision maker

Full size image

A further problem is that the real life decision-making situations we are discussing here, whether in the health care or the family context, are characterised by much more uncertainty than meets the ideal decision maker described above. The probabilities and utilities cannot be fixed with precision so a real life von Neumann–Morgenstern rational decision-maker would have to apply sensitivity analysis, running the calculation several times with, for instance the best and the worst parameter values for each option. This may show that the utility maximising, best option is significantly more risky than other options that are nearly as good, again casting doubt on whether there is any logical compulsion in calling any of these options “the best”.

With regard to ethical theories there are a number of theories that are compatible with a “logically only one best option” view. It is obvious that maximising consequentialism is compatible, since its decision making procedure is isomorphic with von Neuman–Morgenstern expected utility theory. The only difference is that “utility” and therefore “best” has changed meaning. It is no longer best for the person in question, but best in a universal sense, since the values given to outcomes in the consequentialist calculus is not the value to the decision maker but the value to all entities affected by the decision.Footnote 4

Gewirthian rights theory and a strict Kantian approach may also be compatible with the “logically only one best option” view, but neither guarantees that anyone will be in a position to decide what that option is. Versions of these theories that allow for the existence of true moral dilemmas (i.e. cases where I strictly ought to do A and strictly ought to do B, but can only do A or B) will deny that there is always a best option. If true moral dilemmas exist there are situation where we are left with a tragic choice with no good, or best options. And we submit that it is not obvious that the courts are better at making tragic choices than the people affected are.

Best Option and Values

What account of values do we need to adopt in order to make “logically only one best option” plausible? It is perhaps easier to see what kind of value system that would be incompatible with this view. If a value system contains incommensurable values, i.e. values that cannot be scaled or ranked against each other there cannot logically be a best option in cases where both values are engaged, if there is no option which allows us to instantiate both values. Whether incommensurability is a possible relation between values is a hotly contested question in philosophy, but it is definitely not warranted at the current stage of that debate to discount the possibility that incommensurability may pertain.

A further question is whose values that should be determinative in the decision. There may be several value systems in play, each held by some of the actors and the “best option” can only be fixed if we choose one of them as dominant. But which one? From rational decision making theory it seems to follow that we should choose the value system of the subject, in so far as that is known because the best option is the one that maximises utility for that person. But this approach has been rejected by the English courts, even in cases where we have reasonably reliable knowledge of the person’s value system. In the interesting case of An NHS Trust v. A [2005] EWCA Civ 1145 concerning withdrawal of treatment from a patient who had strong Muslim beliefs that were acknowledged by all parties, Lord Justice Waller put forward the following remarkable argument (at 84):

The final criticism made by Mr Glancy is that the judge failed to take into account the family’s feelings and their religious beliefs. The fact that the judge put these at the end of his judgment does not in my view show that he did not have them properly in mind. It should be remembered in particular that the treating doctors had themselves had very much in mind the religious concerns of the family, and indeed all the concerns of the family. He was clearly right to consider what was certainly the key question first, as to whether there was in his view any chance of recovery of any quality of life so as to make the discomfort to which Mr A was being put justified. Once he had formed that conclusion—that it was not justified—it was obviously going to be difficult for the religious views and the views of the family to overcome the obvious point that, since any decision to put Mr A through further suffering would produce no benefit to Mr A, it would be difficult to see that it could be in Mr A’s best interests. But on any view he did consider the family’s position and their religious beliefs.

What is essentially claimed here is that the court should first apply the court’s value system, although hiding this under the rhetorical device of presumptive definition in the statement of “the key question”, and only later consider the patient’s views. But what is in the best interest of the patient must to a large extent be determined by the values of the patient.

A possible reply could be that when the case comes to court the value system embedded in the law automatically becomes dominant, but that reply is problematic for a number of reasons. First “the value system embedded in the law” is to a large extent a fiction, there is no hope of reconstructing a coherent value system from any developed body of law. Second, it is unclear what the justification could be for imposing this particular value system in all those cases where the decision will mainly affect one person, or where all the persons significantly affected share another value system. Should we say “You think that X is in your best interest, but because we have a different value system we think that Y is in your best interest and because we are powerful we will impose Y on you”? This issue can be side-stepped rhetorically by claiming that the court is justified in applying a legal conception of “best interest” because it is making a legal decision, but in that case the court should probably make this clear by only talking about “legal best interest”, not “best interest” simpliciter.Footnote 5

Third, imposing a default value system significantly advantages parties (such as NHS trusts) whose values are more in line with the dominant value system. Even if they are not nearly as affected by any decision, they can rely on a good chance of getting their view of the right decision endorsed against the protests of those who are really affected.

Some Further Epistemic Issues

Let us, for the sake of argument assume that Butler-Sloss is right in maintaining that there is one and only one best option. This would still (1) not entail that option can actually be found in real life and (2) not settle the question of who should try to divine which among the many options is the best, or by what procedure they should try to do so.

From the premise “theory affirms that there is an action in this context which is in the person’s best interest” there is no entailment to the conclusion that we will be able to get to the judgement “this specific action is the one that is in the person’s best interest” in real life cases. This can be seen if we consider the analogy to scientific questions. There are many scientific questions where we know that they have one and only one specific answer, but where we currently have either no methods to ascertain what that answer is, or where application of our best methods still leave room for doubt.

When we described the von Neumann–Morgenstern approach above we glossed over one of the more complicated epistemic issues, the issue of how we can ever get to a final, fixed assessment of the consequences of a particular action. Many actions have chains of consequences that stream out infinitely into the future and there may by no non-arbitrary way of deciding when it is acceptable no longer to count these consequences as consequences of the act.

In some circumstances it also seems to make sense to ask whether the decision we are making is a decision concerning what is in this person’s best interest, or whether it is more correctly described as a decision concerning what person should exist in the future. In decisions concerning young children in divorce cases the choice may lead to radically different childhood experiences (Trotskyite or Tory?), and similarly in some decisions concerning permanently incapacitated adults. In certain circumstances one of these life courses will be unequivocally better, in the sense that it dominates alternative life courses in all time slices with regard to welfare in relation to any conceivably acceptable value system, but this will very rarely be the case.

An Alternative Interpretation

Returning to Dame Butler-Slosss’s statement, it may be noted that she argues that the role of the court is to make a decision in the context of ‘the range of acceptable opinion among competent and responsible practitioners’. This suggests a recognition of the fact that medical practitioners need not be expected to agree. Such is the complexity of medical science, that a certain amount of disagreement between practitioners is to be expected, and indeed is a sign of vital and critical inquiry within the discipline. Such disagreement might be characterised through the Rawlsian notion of ‘reasonable disagreement’. The problem that such disagreement poses is that, at least in the short term, a single course of action has to be chosen. It may then be suggested that the court’s task is the choice of such a course of action. While we will still argue that the characterisation of that course of action as the only logical best option is deeply problematic, this reading nonetheless opens up an alternative approach to the understanding of legal decision-making, which is to say, an approach grounded in legal pragmatism.

Pragmatism emerged as a distinctive philosophy in the late nineteen century, with the work of Peirce, James and Dewey. As a young lawyer, Oliver Wendell Holmes Jr. was part of the intellectual circle within which pragmatist ideas were first articulated, and essays such as ‘The Path of the Law’ [1] and The Common Law [2] are influenced by pragmatism. At the core of philosophical pragmatism lies the argument that the truth of a proposition is determined by the practical usefulness of believing in that proposition (and crucially, as Peirce was to stress, in the long run). Pragmatism may therefore be seen to pit itself against rationalist, and in particular Cartesian approaches to philosophy, that would seek to deduce the truth of propositions, with a mathematically certainty, from indubitable foundations. Pragmatism is, rather, committed to fallibilism. The truth (and even utility) of a proposition cannot be known prior to its practical application.

A legal pragmatism may be characterised initially through its opposition to conventional legal thinking. It may be suggested that the first interpretation that we have given of Butler-Sloss’s comments represent precisely that conventional approach. The remarks are couched within a broadly rationalist or Cartesian framework, or within ‘legal formalism’. The assumption here would be that the law constitutes a more or less coherent system. The task of the judge is then to deduce an appropriate decision, much in the way that a mathematician deduces the solution to an equation. As Posner presents this, ‘legal questions presented to judges for decision have but a single right answer, which judges can find by the use of legal reasoning’ ([3], p. 149). While, as Posner admits, a pure formalist approach to legal decision-making, and as such an approach that grants the judges little or no power of discretion, is something of a caricature of actual legal practice, ‘a looser formalism remains significant’, and it is this that we suggest characterises the first interpretation of Butler-Sloss’s remarks.

A pragmatic interpretation of Butler-Sloss’s comments would differ from the conventional formalist interpretation by emphasising the fallibility of the decision that is to be made. This fallibility is reflected in the reasonable disagreement that characterises the medical community. The ‘best option’ thereby ceases to be the single objective solution that can be deduced from the legal system. Rather it becomes a practical experiment, arrived at as much through the discretionary powers of the judge as by his or her deductive powers. Its truth and effectiveness cannot be known, with any degree of certainty, before the judgement is put into practice.

The distinction between a pragmatist and formalist approach may be explored further by considering their respective approaches to decision-making. As Posner is at pains to emphasise, while pragmatism is concerned with consequences, it is not a form of consequentialism ([3], p. 151). It is here that we can see the difference between a pragmatist approach and the von Neumann–Morgenstern rational choice theory that we have attributed to Butler-Sloss. The von Neumann–Morgenstern approach attempts to identify the (single) best possible option through a deductive procedure that evaluates and compares the utilities of all possible options. The pragmatist will reject such an approach, firstly for its assumption that all possible options can be deduces prior to practical application of the decision, and secondly for its assumption that the value of each option can be known in advance of practice. These two points of difference highlight a tension within the conventional formalist approach. In striving to assess all possibly consequences of a judgement, it is not clear that the court recognises that it is still making a legal decision. Posner notes that the pragmatist does not require that the court take into all the potential consequences of its decision ([3], p. 151). The court necessarily has limited information at its disposal, and a limited jurisdiction. Within the pragmatist interpretation, as we suggested above, the court is primarily ensuring that reasonable disagreement is translated into action—that something is done. The court is merely ensuring that this is, within its limited jurisdiction and incomplete and fallible information, the most reasonable thing to do, with the all important caveat that it may turn out to be wrong (for example, by being scientifically unworkable, morally offensive, or unworkable as future social policy).

Further, and perhaps more critically still, the von Neumann–Morgenstern approach presupposes that the result of deductive analysis is normatively binding. As we have suggested above, this is a highly problematic assumption. In pragmatism, as Posner expresses it, at best ‘analyses identifies the consequences of legal decisions but leaves it up to the judge… to decide how much weight to give to those consequences’ ([3], p. 152). That is to say that the judge must draw, not simply upon his or her personal values, but rather upon relevant values that are current within society, in order to estimate the moral significance and acceptability of the judgement. Again, this highlights a crucial difference between pragmatism and formalism. The formalist seemingly makes a decision solipsistically. At best he or she draws on the fiction of the court’s values to frame the decision. In contrast, the pragmatist makes a decision in the light of social values, and perhaps more significantly in the light of a plurality of such values. This is to suggest that the pragmatist does not merely acknowledge reasonable disagreement (be it within a specialist scientific community, or within a wider public), but further incorporates that disagreement, as an explicit recognition of the fallibility of the judgement, into the decision-making process. Put otherwise, the court’s judgement is itself reasonable, drawing on the forms of justification that will be recognised within relevant specialist scientific communities and publics, and not rational ([3], p. 151).

Why so Attractive?

Why are intelligent persons attracted to the conception of best interest and the role of the court espoused by the first interpretation Butler-Sloss when it is so philosophically problematic, and as we have argued rather obviously unsustainable?

One reason might be that in a situation where an actor (e.g. a judge, a doctor or a family member) has to make a contentious decision, believing that the decision is not only right, but logically the only possible decision must provide some measure of comfort.

A second reason may be the need to maintain two features of the legal system, one of them probably best described as a legal fiction. The first feature is the reluctance to declare a case moot, based on the argument that the parties must have, need or should get a decision. The second more fictional feature is the idea that the decision taken by the court is the uniquely justifiable decision which in some sense flows from the laws of the legal system (i.e. the decision is the legally right decision). But we would submit that in cases where the parties have very different stakes in the outcome, for instance where one of the parties would be relatively unaffected by the court decision, but where the other will be significantly affected, it is not obvious that the court should make a decision. In many cases the decision should be left to those who are most affected. We would also submit that the rhetoric of correct decision making is only rhetoric, and in this case strongly misleading. It is misleading because while it may recognise that a decision is difficult it must deny the tragic dimensions of moral dilemmas which are incompatible with the view that there is one right decision. This is for instance exemplified in the leading judgement of Ward LJ in the conjoined twins case:

In the past decade an increasing number of cases have come before the courts where the decision whether or not to permit or to refuse medical treatment can be a matter of life and death for the patient. I have been involved in a number of them. They are always anxious decisions to make but they are invariably eventually made with the conviction that there is only one right answer and that the court has given it.Footnote 6 (our emphasis)

A third reason might be that people underestimate the epistemic problems outlined above and/or the open ended nature of the consequences flowing from many of the choices discussed here.

But being able to understand why people hold fallacious view provides absolutely no justification of such views.

Can Best Interest be Salvaged?

It clearly makes sense when making decisions for somebody else to ask what the character of the decisions is, what the consequences of each of the possible decisions will be, how these will affect the future wellbeing and life of the person in question etc. But how should we describe that task so that we don’t lose any of its complexity, whether epistemically or procedurally, and are not led down the path of believing that there is always, or perhaps even often one discernable best interest that a court should base its decision on?

We should probably first, in the context of decision making for others, get rid of all labels containing the term “best”, since such labels are potentially misleading. If we were to accept that ‘best interests’ are a legal fiction, necessary in order to resolve reasonable disagreement and thus to get something done, then this best interest can explicitly be separated from any epistemological or moral reality. However, as a pragmatic response the judgement as to ‘best interests’ must be aware in explicit awareness that it necessarily falsifies (or fictionalises) the tragic moral reality that lies at the centre of the case. As such the judgement must be open to correction in the light of the consequences of actions based upon the court’s decision. This would allow the tragic and moral dimension to reassert itself against the legal fiction. The problem at present is that, within the conventional formalist framework, ‘best interests’ acquire an unassailable objectivity, that blinds the judgement to the consequences of its own implementation.