Storrs McCall

Introduction

Practical deliberation - bouleusis - is discussed by Aristotle in books 3 and 6 of the Nicomachean Ethics. Although Aristotle doesn't mention this, it is necessary to distinguish practical deliberation from what may be called "cognitive deliberation". Cognitive deliberation is deliberation over whether something is true or false, while practical deliberation is over what to do. In jury trials for example the jury is asked to decide the cognitive question, is the accused guilty as charged? Its deliberations concern this matter exclusively. If the judgment is "guilty", the judge must deliberate about something quite different, viz. what sentence to impose. The jury's deliberations are cognitive, the judge's practical. In this chapter, I shall be concerned with practical deliberation.

Aristotle makes two remarks about deliberation that are familiar to every philosopher: first that we deliberate about means not ends, and secondly that we deliberate about things that are in our power and can be done (EN 1112a30, 1112b11-16, 1139b6, 1140a32). I need a table; to make a table requires a hammer and saw; I have a hammer but no saw; so let me go out and buy a saw. Again a doctor, qua doctor, does not deliberate about whether a patient should be healed, but how to heal him. No one will disagree with this. To be sure, one can imagine circumstances in which a doctor might deliberate over whether a patient should be healed, say if there were 1000 patients and drugs for only 500. But in this case the doctor isn't really deliberating about ends, but about the practical problem of how to treat as many people as possible with the means available. If a doctor were truly to deliberate about ends and ask for example, "Is healing worthwhile?" then that would fall under cognitive rather than practical deliberation.

What exactly is deliberation? In what follows I try to give a philosophically adequate account of it, and to answer some difficult questions. I finish up by speculating about what sorts of structures and neurophysiological functioning in the brain would make possible in real life the philosophical description of the deliberative process that I have given.

Deliberation

Let's start off with a concrete example. Marsha has to decide whether to accept an offer of graduate study in philosophy at UBC, Western Ontario or McGill. Call these alternatives A, B and C. Each one has its advantages and its disadvantages, and it is important to make the right decision. Marsha deliberates.

The first step in the deliberative process is to be clear about the alternatives, to represent them accurately and keep them in focus. Is the list exhaustive? Should she make a last-minute application to McMaster, where her sister is studying? Should she simply do nothing, and not accept any of the offers? Call this last option D, the do-nothing alternative. Let's suppose that A, B, C and D exhaust the alternatives facing her, and that each one is a "live" alternative in the sense that (i) she can choose it, and (ii) if she chooses it her choice determines what happens subsequently. Thus if she chooses UBC, she goes to UBC. We may imagine that Marsha has lined up three envelopes on her desk, addressed to UBC, Western and McGill. All she has to do is to put a stamp on one of them and mail it, or alternatively forget about graduate study. The first requirement for deliberation, then, is the existence of a "choice set", a set of two or more alternative courses of action (A, B, C, D,...) each of which it is physically possible for the deliberator to perform or implement, and which together exhaust the available options.

Once the choice set has been established, the process of deliberation begins in earnest. Each option has its advantages, which constitute the reasons for choosing it, and its disadvantages, which constitute the reasons for not choosing it. I shall call these deliberation reasons. For example UBC has made a generous scholarship offer, but on the other hand has no member of staff who works directly in Marsha's area of interest. These facts constitute positive and negative deliberation reasons for A, and there will be other positive and negative deliberation reasons for B, C, and D. In deliberation, we weigh deliberation reasons. We compare their relative strength with the aim of arriving at an overall comparative evaluation. The process of evaluation is normally but not invariably the most time-consuming part of deliberation, and ideally should result in a list of the options ordered by preference.

Once evaluation is completed, it might seem that the deliberative process is at an end. But this is not so. There is one more step, frequently ignored in studies of rational choice and decision but still essential: the element of choice or decision itself It may seem difficult to imagine, once an ordered evaluation of the options has been made, what more a deliberator could need. But suppose the first two alternatives are very close in the ordering? Suppose they come out equal? What if the deliberator is faced with a difficult decision? Even in the case where an evaluation is unambiguous, with a clear-cut winner, something else besides evaluation is needed for action. The missing element is what Aristotle call prohairesis, deliberative choice.

More needs to be said about prohairesis, but we should first sum up what has been established so far. Our philosophical account of deliberation has distinguished three separate components of the deliberative process, ordered in strict temporal sequence:

(i) Representation of the alternatives,

(ii Evaluation,

(iii) Choice.

With decision, which is choice of one of the alternatives, the deliberative process ends. We turn now to decisions and the reasons for them, the latter being distinct from the deliberation reasons for the different options.

In deliberation we weigh and assess deliberation reasons, each being a reason for or against one of the alternatives. When eventually we decide, and choose one of the options, there is normally also a reason why that option was selected over the others. I shall call the latter an explanation reason, or, occasionally, a decision reason. In the deliberative process there are deliberation reasons, i.e. the reasons for or against the different options, but there are also explanation reasons, which are the reasons why, after examination of the deliberation reasons, one of the options is chosen.

A principal objective of this paper is to be clear about the exact difference between deliberation reasons and explanation reasons. Without a good understanding of the difference, I don't think we can know what deliberation is. Therefore I shall spend some time discussing the relationship between them.

Deliberation and Explanation Reasons

First, is it certain that a sharp line can be drawn between deliberation reasons and explanation reasons? On the face of it, yes. Deliberation reasons, in the deliberative process, have a relatively long life. Some may be present from the start, as when a diner at a restaurant perceives an excellent reason for choosing a chocolate mousse dessert at the very moment he becomes aware that it is on the menu. Others may emerge later, when the calorie count of the mousse is compared with that of the fruit salad. The explanation reason, on the other hand, comes into existence only at the end of the deliberation, when the choice has been made. As long as there is not yet a choice or decision, there cannot as a matter of logic be any explanation of it. Deliberation reasons are reasons for deciding this way or that; explanation reasons are reasons that or why a certain decision was made. A typical explanation reason might be of the form "X chose A rather than B because in the end X attached more weight to the deliberation reasons for A than to the deliberation reasons for B". If it exists at all, the explanation reason comes into existence only when the decision comes into existence. This isn't the case with deliberation reasons. Hence the two are quite different.

That being said, we still want to know more about explanation reasons. Could they perhaps be associated, or identified, with comparisons among deliberation reasons, or more strictly perhaps with comparative rankings of alternatives based on deliberation reasons? If Marsha for example were to conclude that the program and the second language opportunities at McGill were more important than the funding and the friends she has in Vancouver, would this not be a perfectly good explanation reason for the decision to go to McGill? We must proceed carefully here. A reason for going to McGill, even though it results from a comparative weighing of all the deliberation reasons, is still a deliberation reason. One can, if Aristotle is right in maintaining there is such a thing as akrasia, conclude that overall the reasons for doing X outweigh the reasons for doing Y, and still do Y. In Davidson's memorable example someone may conclude that there is nothing to gain and everything to lose from drinking a can of paint, and still one day drink it. No matter how strong it is, and how much it dominates all other deliberation reasons, a deliberation reason is not an explanation reason. Not until the decision has been taken, and the choice made, does the explanation reason come into existence.

Secondly, must every decision have an explanation reason? Or are there some totally irrational decisions, for which there is no explanation at all? Is the difference between a decision that is made lightly, casually or thoughtlessly, and a decision that is made carefully and deliberately, the difference between a decision which lacks, and a decision which has, an explanation reason? These questions are not easy to answer. Let us start with Buridan-type situations, in which a choice must or should be made, but in which there is absolutely no reason for choosing one thing rather than another.

Buridan's ass starved to death half-way between two piles of hay because there was nothing to incline his choice towards A rather than B. Here the deliberation reasons for A and B are equally balanced, and as a result no decision is taken. A similar problem, based on fear rather than desire, is the "railroad dilemma." What fascinates us in these examples is not just the spectre of decisional paralysis, but the feeling that the outcome - death by indecision - is in a genuine sense an affront to reason. If the ass had been rational, or more rational than he was, he would not have starved. He would have drawn straws and said "Long left, short right". Failing this, if he had been clever enough, he could simply have made an arbitrary or criterionless choice. If you believe this is impossible, reflect on how you manage to choose one of a hundred identical tins of tomato soup in a supermarket. The example of the soup tins was a favourite of Macnamara's, who used it with great effect in discussion. If you made the mistake of saying that an arbitrary choice in these circumstances was difficult or impossible, that would indicate that you, like Buridan's ass, were not very intelligent.

Moving from Buridan-type examples to cases where there is a significant difference between the alternatives, but where the deliberation reasons are still equally balanced, the same question arises: can a choice be made for which there is no explanation or decision reason? If Marsha finds it difficult or impossible to decide between UBC and McGill because between them there is "nothing to choose", can she use some tie-breaking mechanism like a coin and in so doing make a choice which lacks an explanation reason? Not really. Although in this case the explanation is different from what it would have been if the evaluation process had produced a winner, an explanation reason still exists. If Marsha is asked why she chose UBC rather than McGill, she may answer that she flipped a coin. The reason why she chose UBC is that the coin fell heads. There need be nothing irrational about this. Compared to the paralysis of the Buridan example, and bearing in mind the approaching deadline for mailing the letter of acceptance, it is a supremely rational procedure. Sometimes we just have to stop deliberating and decide. It's a sign of rationality that we're able to do this.

Although the conclusion towards which we're moving seems to be that decisions always have explanation reasons, even if the reason in question may be something like "Because the coin fell heads", there still remain other cases to be examined. All our examples have been decisions that were made carefully, thoughtfully, deliberately. Even if a coin had to be used to break ties, its use was deliberate and rational. But what of decisions that are taken carelessly, without adequate thought? What if we are inebriated, and our deliberative powers impaired? Such cases may provide examples of decisions without explanation reasons.

The making of a decision, like the activity of evaluation, is a rational process. In fact deliberation as a whole is a rational process, and each of the components which make it up is subject to norms by which it can be judged.

(1) In the formation of the choice set the alternatives should be realizable, and each option should be such that the decision of the deliberator alone is sufficient to implement it, or to initiate its implementation. The alternatives should also be exhaustive: no important or significant alternative should be excluded from the choice set.

(2) For the process of evaluation, different norms apply. Have all the deliberation reasons for all the alternatives been recognized and collected? Has each deliberation reason been appropriately weighted? If there are several deliberation reasons for the same alternative, has the total weighting of all of them, including possible negative reasons, been reasonably assessed? Finally, given appropriate total weighting for each alternative, has the comparison between them been carried out honestly and without bias? Or if there is bias, is the deliberator conscious of it, rather than having it work behind the scenes and affect the evaluation without the deliberator being aware of it? These are the questions to be considered if an evaluation is to be judged rational. It is true that frequently we lack sufficient insight and self-knowledge to be sure whether in a particular case we are giving due and appropriate weight to a set of deliberation reasons, particularly if the decision is one that touches us emotionally. In many cases the degree of objectivity and detachment required may be beyond us. Nevertheless, if the deliberative process is to be a rational one, a certain level of objectivity and selfknowledge are needed.

Finally (3) there are, in addition to the above requirements, norms for rational choice. The first requirement is that to be considered rational, a choice or decision must have an explanation reason. If for example a choice is made for no reason at all, not even with the aid of a randomizing device like a coin as part of the decision making process, then the decision cannot be considered rational. Furthermore, even in cases where an explanation reason is provided by the deliberator norms of rationality still apply to it, and the rational assessment of a decision must take them into account. For example, if Marsha were to explain her decision to go to UBC by saying "Because I'm a Libra", the relevance and consequently the rationality of this supposed explanation could be questioned. Decision therefore, like choice-set formation and evaluation, is a procedure which it is appropriate to subject to rational norms, and to judge rational only if it satisfies these norms. The same applies, mutatis mutandis, to the whole deliberative process.

An important question which should be addressed is whether an explanation reason for a decision is a cause of that decision. The immediate answer would seem to be no. An explanation reason for a decision comes into existence only once the decision is made, while a cause should antedate the thing it causes. The whole question of causality and its role in the deliberative process is nevertheless of great interest, and must be looked at carefully.

First, our discussion from the start has taken us outside the framework of causal determinism, since the very existence of a choice set with two or more physically realizable alternatives is not consistent with determinism. One might attempt to remove the inconsistency by requiring that in any choice set one and only one alternative was causally possible, the deliberator necessarily being ignorant of which one that was. As long as the deliberator believed (falsely) that each alternative in the choice set was realizable, deliberation could proceed as usual, the eventual decision being taken in favour of the one option that was in fact open. This is deliberation according to the script that would be written for it by a determinist. The re-writing does indeed remove the inconsistency, but at the price of (i) abandoning choice sets with two or more real alternatives, and (ii) drawing a veil of ignorance over the eyes of deliberators as to which is the sole choosable option. Behind the veil, "deliberation' can proceed. But re-writing the script in this way is unnecessary, if it can be shown that it is simpler and more elegant to abandon determinism.

Since the 1920's, quantum mechanics has provided the example of a science which is probabilistic rather than deterministic. As a result, replacing deterministic models of deliberation by indeterministic ones is not such a daunting or unthinkable project today as it would have been for Hume, or Mill or Brentano, or any philosopher working within the paradigm of Newtonian science. This is not the place to speculate exactly how or in what respect the overall neuronal functioning of the brain could be regarded as probabilistic rather than strictly deterministic, but at least the hypothesis that the brain is a complex indeterministic mechanism makes sense, is testable, and may one day be confirmed or falsified.

What is important for present purposes is the fact that if the central nervous system functions probabilistically rather than deterministically, then this functioning permits the formation of choice sets in deliberation. At the neurophysiological level, a choice set requires different physically possible neural states n(A), n(B), n(C),... corresponding to the different actions A, B, C, ..., the states being such that if n(A) obtains then A is performed, if n(B) obtains B is performed, etc. Just as each of A, B, C..excludes the others, so each of n(A), n(B), n(C),…excludes the others. These states are "missing" states of the brain which neuroscientists should one day be able to specify and describe, but which during deliberation are not actual existents. What is essential however for the existence of a choice set is that throughout the deliberative process each one of the neural states n(A), n(B), n(C),...should be physically possible, i.e. capable of becoming actual at the end of the process. Since each state is incompatible with the others, the only way this could be the case would be if the brain functioned indeterministically. Indeterministic neural functioning consequently underpins choice set formation in the sense that it makes it possible, that it is a necessary condition for its existence. If we believe in choice sets, with each option separately realizable, I don't see how we can avoid regarding the operations of the brain as in some yet-to-be-discovered way indeterministic.

The next step is to consider probabilities. If each of the outcome neural states n(A), n(B), is physically possible, it will have a probability value, and the probability values of all the members of a given choice set will sum to one. Needless to say, in keeping with the idea of the brain as a complex indeterministic mechanism, we are talking of objective probability values, not epistemic ones. With the introduction of probability comes the possibility of saying something about causality and perhaps ultimately being able to conclude whether deliberation can be considered a causal process.

At the beginning of the deliberative process, once the choice set is formed, the existence of the underpinning neurological mechanism, with the different target neural states, gives each alternative of the choice set a precise probability value. An interesting and important question is whether these probabilities change during deliberation.

If the probabilities change during the deliberative process, considerations of probabilistic causality may apply. For example if J is some factor that enters consciousness during deliberation a thought, a desire, a fact, a memory, a goal - and if A is one of the options of the deliberator's choice set, then if

p(A\J) > p(A\not-J)

we may say that J is a (prima facie) probabilistic cause of A. This is not equivalent to saying that J causes the alternative A to be chosen, since together with the above inequality we may also have

p(B\J) >p(B\not –J),

and could not be the cause of A being chosen and also B being chosen. Hence probabilistic causality is of limited interest in connection with deliberation.

Again, the factor J may increase the probability of option A's being realized, but can it increase it to unity? Can any deliberation reason, or deliberative factor, be of such strength that it makes it physically or causally impossible for any alternative other than A to be selected? Might it, before the choice is made, reduce the choice set to a single choosable option? This takes us back once more to determinism. One cannot rule out the possibility that during deliberation something should occur, e.g. paralysis or a violent fit, which restricts the alternatives to one option only. But such cases are so far removed from the norm as not to qualify as instances of deliberation at all. At the end of a deliberation we are left either with the choice set with which we began, or with a reduced choice set, or an enlarged one, but in any case a set consisting of at least two members. The probabilities of these options can change, but if any of them reaches the value one deliberation ceases.

A final question concerning probabilities, which reintroduces the subject of explanation or decision reasons, is this. At the end of the deliberative process, before a decision is made, the deliberator is faced with a choice set of different alternatives A, B, C,..., each with a corresponding neural state n(A), n(B), n(C),... The objective probabilities of realizing these different options, and hence of the different target neural states becoming actualized, will not in normal circumstances be known to the deliberator. Each probability will in fact be non-zero and some may be exceedingly small. Is it possible for a deliberator to choose, deliberately and intentionally, an alternative of very low objective probability?

From the point of view of probabilistic science, the occurrence of an event of low probability is not impossible, but would not be expected. If it happened, the only explanation that could be given would be something like "By the laws of chance, improbable events do occasionally occur. By pure chance, a bridge player may be dealt a hand of 13 spades." There exists no better causal/probabilistic explanation than this of why an improbable event occurs. But this isn't true of a choice made by a deliberator. No matter how objectively improbable the chosen option may be, a perfectly good explanation reason for the choice may be forthcoming. The deliberator who has for years drunk nothing but whiskey may say that this time he chose a Tia Maria, no matter how objectively unlikely his choice may have been, because he felt like it.

The lesson here, I think, is that in the realm of deliberation and choice there may occur events for which there is no causal/probabalistic explanation, but for which there is an intentional explanation. Some intentional explanations may be of the form "Ivan chose the Tia Maria because he felt like it;" others may be of the form "Unbelievably, Sally isn't here today because she deliberately and intentionally chose to go to jail." Perhaps the clearest examples of intentional explanations are those given in circumstances where non-intentional explanations, i.e. causal/probabilistic ones, don't exist. But detailed examination of the difference between intentional and causal/probabilistic explanation is a lengthy matter, and must await another occasion.

Conclusion

Summing up, I began by listing the three essential components of a deliberation. These are, in strict temporal order, (i) choice set formation, (ii) evaluation, (iii) decision. In evaluation the reasons for and against the different options (deliberation reasons) are both appropriately weighted and weighed by the deliberator. Eventually one option is chosen. Each element of the deliberative process is subject to rational norms by which it may be judged, a common but non-mandatory norm of element (iii), choice or decision, being the existence of an explanation reason for the choice once it is made. If choice set formation is underpinned by an intedeterministic neural mechanism, with a specific neural state n(Ai) and a specific probability corresponding to each alternative Ai of the set, then the explanation of why the deliberator chooses one of the alternatives will not in general be a causal/probabilistic explanation based on those probabilities. Instead it will be an intentional explanation. The precise relationship between intentional and causal/probabilistic explanations in deliberation remains to be investigated.

References

Bratman, Michael. 1987. Intention, Plans, and Practical Reason. Cambridge, Mass: Harvard University Press.

Cartwright, Nancy. 1989. Nature's Capacities and their Measurement. Oxford: Clarendon Press.

Davidso, Donald. 1980[1963]. Actions, reasons and causes, in Donald Davidson, Essays on Actions and Events. Oxford: Clarendon Press.

DeVito, Scott. 1997. Probabilistic Causality Without Propensities, typescript, University of Pittsburgh.

Eells, Ellery. 1991. Probabilistic Causality. Cambridge: Cambridge University Press.

Fetzer, James H. 1988. Probabilistic metaphysics, in James Fetzer (ed.), Probability and Causality. Dordrecht: D. Reidel Publishing Company.

McAdam, James. 1965. Choosing flippantly or non-rational choice, Analysis 25:132-6.

McCall, Storrs. 1987. Decision, Canadian Journal of Philosophy 17:261-87.

McCall Storrs. 1994. A Model of the Universe. Oxford: Oxford University Press.

Ofstad, Harald. 1961. An Inquiry into the Freedom of Decision. Oslo: Norwegian University Press.

Rescher, Nicholas. 1959. Choice without preference, Kant-Studien 51:142-75.

Suppes, Patrick. 1970. A Probabilistic Theory of Causality. Amsterdam: North-Holland Publishing Company.