Sunday, 31 July 2011

Parfit on Moral Concepts

Chapter 7 of On What Matters concludes the first part of the book and is a counterpart of the first chapter which concerned 'normative concepts', concerning as it does 'moral concepts'. This chapter is lengthy and dense and is one that requires summary primarily since it promises to be important for the discussions that will begin in Part 2 of the work.

Parfit opens the chapter by promising to disambiguate the varied senses in which something can be "morally wrong". The peculiarity of the way this discussion is framed is that it indicates an understanding of "facts" albeit ones that are "morally relevant". The 'ordinary' sense of "moral wrong" apparently refers to thinking about acts in the same sense as each other. However the first complication and reason for analysis concerns the apparent ignorance we often suffer when evaluating actions, ignorance with regard to some or all of the "morally relevant" facts. When faced with such ignorance (or with "false beliefs") we find the problem arises that we use "morally wrong" in different ways. Three such differences are distinguished by Parfit.

The first sense we use "morally wrong" in situations of partial ignorance or "false belief" or what we might term 'non-ideal' forms of moral evaluation is to term something 'wrong' in a "fact-relative" sense. Something is 'wrong' in this sense just when it would be wrong in the ordinary sense if all the facts were known. This contrasts with thinking of something as 'wrong' in a belief-relative sense which concerns something being 'wrong' if our beliefs about the acts in question are true. Finally, something is 'wrong' in an 'evidence-relative' sense when we would term it 'wrong' in the ordinary sense if we believe what the available evidence gives us 'decisive reasons' to believe and these beliefs are true.

Parfit's point in making these distinctions is to suggest that none of them overlap with the 'ordinary' use of 'wrong' and that no one of these senses of 'wrong' is sufficient to exhaust what is meant by it. In illustrating the differences between these senses of 'wrong' Parfit discusses the case of a doctor who has various unjustified beliefs about the effects of treatments and possesses the wish to kill someone. In the first example of this the doctor prescribes something in the belief it will save the life of the patient but which kills them as it was sure to do given its properties (hence in which the doctor acts with good intentions but imperfect knowledge).

By contrast, in the second example, the doctor prescribes a treatment in the belief it will kill a patient but which saves their life given its properties (in which the doctor thus acts with a bad intention but imperfect knowledge). Many would simply assimilate both these cases to wrong in the 'belief-relative' senses thus declaring the first conduct 'right' and the second 'wrong' despite the outcomes in question. By contrast Parfit claims the doctor in the first of these cases acted wrongly in two different senses, one in a fact-relative sense and one in an evidence-relative sense since it would have been possible to derive evidence here that would have informed me about what the consequences of the action I was performing would have been.

By contrast, in the second case, the doctor acted rightly in fact-relative and evidence-relative senses despite possessing a bad intention. Now, to complete the basis for his distinctions, Parfit next mentions some further variations on the case in question. In these variations the doctor firstly (Case 3) gives treatment that is almost certain to kill the patient but which saves their life as they unjustifiably believed it would. In Case 4, by contrast, they give treatment that is almost certain to save the patient's life but which kills them, as they hoped it would.

In cases 3 and 4 it would be common to assimilate them to the sense of 'right' and 'wrong' in an evidence-relative sense so that the doctor in case 4 can be declared to be acting rightly whilst that in case 3 is acting 'wrongly'. Again, the point is to get past such a view. In Case 4, on Parfit's view, we should not say that the doctor acted 'rightly' as, when we bring in belief-relative and fact-relative senses, we can say that the doctor in question had murder in mind and succeeded in murdering the patient in question. By contrast, in Case 3, it isn't enough to refer to the evidence-relative sense that would say the doctor in question acted 'wrongly'. We should also refer to fact and belief-relative senses of 'wrong' as these show that the doctor acted in these respects 'rightly'. 

The peculiar characteristic of these four examples concerns what is meant by declaring them wrong in 'fact-relative' sense since the usage Parfit gives does not seem to be based on 'morally relevant' facts but rather on actual facts and hence the 'fact-relative' sense does not appear to pick out a specific characteristic of the actions but to collapse into the 'evidence-relative' sense. In Cases 5 and 6 he aims to indicate why we should not say that things are only 'wrong' in a fact-relative sense. In Case 5 the example is that the doctor gives treatment that is almost certain to save life, as he justifiably believes but which, unfortunately in fact kills you. Here we shouldn't claim the doctor acted 'wrongly' since the sense of 'fact-relative' is here plainly distinguished from 'evidence-relative' but not in a moral manner so different (there is here a difference between 'fact' and 'evidence' but it is an epistemic difference not a moral one).

Finally, in Case 6, the doctor gives treatment that is almost certain to kill the patient as he justifiably believes but which, in fact, saves the patient's life. Again,the doctor did not act rightly since they saved the patient's life but, again, the difference between 'fact' and 'evidence' is only epistemic and not something morally salient. So none of the six examples show that Parfit does have a specific moral sense for his notion of 'fact-relative'. 

In listing the senses Parfit's intention was to identify different questions concerning wrongness. If we think of something as 'wrong' in terms of being blame-worthy, then we include questions about remorse and indignation and focus primarily on belief-relative senses of wrongness. Here we can isolate the non-moral beliefs of the agent and declare that if these beliefs had been accurate and they acted on them in the way indicated by them that this is sufficient for blame to be assigned. So in cases 2, 4 and 6 we can identify wrong action in the blame-worthy sense given the belief-relative notion of 'wrongness'. 

After making this point Parfit further undermines the moral standing of the fact-relative notion of 'wrongness' by showing that it produces the conclusion that whether an act was 'wrong' is purely a matter of luck (how things happen, contingently, to turn out). There are alternatives to this view which effectively confirm the undercutting of the notion of 'fact-relative' wrong. The alternatives are two forms of Kantian view. The basic Kantian view mentioned by Parfit is that an act's blameworthiness doesn't depend on luck. A semi-Kantian view, by contrast, would say it doesn't depend entirely on luck but an act that doesn't depend on it would be more blameworthy than one that did.

As Parfit indicates the so-called 'semi-Kantian' view is much less plausible than the Kantian one since it concedes something to the 'fact-relative' view. Going back to the cases already mentioned, cases 2 and 4 are equally blameworthy for the Kantian but not for the 'semi-Kantian' who takes the success of the attempt to kill the patient in Case 4 as indicative of a greater wrongness in this case than in Case 2. In rejecting this view Parfit underscores the point that epistemic justification of beliefs is not the cardinal point in the blameworthiness of actions which seems thus to indicate that the fact-relative sense of 'wrongness' does not involve moral characteristics at all.

Further, Parfit endorses the Kantian view and this seems sufficient to me to undermine any plausible independent moral status for the 'fact-relative' notion of 'wrongness'. This does not mean that the 'outcome' is not bad when something is, as Parfit put it, fact-relatively wrong. But its badness does not reside in an independent moral quality which is why it can be assimilated to 'luck'. 

Parfit next introduces a different sense of 'wrong' to indicate that something is 'wrong' in a 'moral-belief relative' sense when an agent acts in a way that 'they believe' to be wrong, which, as Parfit puts it, is equivalent to saying that if someone does what they think wrong then this is wrong even if no other consideration would lead one to take it to be wrong. However, this is also an asymmetric case since there is not 'rightness' in a correlative sense as simply believing something to be right is not sufficient for us to reach the conclusion that acting in accordance with this belief makes it right.

This 'moral-belief-relative' sense is, again, insufficient to describe an independent moral characteristic since it assumes that the belief of the person in question can itself make something wrong that no other characteristic of the action would be sufficient to produce the view that it was wrong and this seems false. However, it is often taken, all the same, to be the case that something is 'blameworthy' by being wrong in this 'moral-belief-relative' sense. This may only rest on a confusion since in taking something to be wrong in this way it is possible that the agent is being blamed or blaming themselves only by standards of a belief that itself has no salient force. (It does, however, appear that often when 'conscience' is at issue it is this kind of belief that is being invoked though 'conscience', it seems to be, has quite a different force to this.)

Now, returning to the 'fact-relative' notion it appears that much of its force concerns what will make things turn out right and hence on consequentialist reasoning. However if this notion is interpreted to mean acting in the way that can produce the best outcomes overall then, as Parfit indicates with his consideration of a case involving trapped miners, this may not be an action we can perform under some cases. Hence Parfit follows the intuition I have been indicating throughout this posting of effectively rejecting the notion of 'fact-relative' wrongness and replacing it by what he calls 'expectabilism' which defines the rightness of an act in terms of outcomes that are 'expectably-best' (and which may thus not be ones that have the best outcomes overall). Thus we should replace the notion of 'fact-relative' wrongness with that of expectable-wrongness where an action is understood to be one that cannot (or is very unlikely) to have the best results that the situation supports (and hence is not being assessed in terms of best overall outcomes possible). Parfit completes the rout of the notion of 'fact-relative' wrongness by pointing out that there is, in any case, no way of acting that would be fact-relatively rightly since we can only act rightly in a belief-relative sense.

So fact-relative effectively gets assimilated to belief-relative in terms of decisions about what we ought to do. Decisive reasons, to use the vocabulary of normative concepts, provide us with the sense of what we would do under ideal conditions. Under conditions that are non-ideal then we should adjust our beliefs to what the evidence points to. 

Parfit next turns to further refinements in the discussion of the 'ordinary' sense of 'wrong' pointing out, for example, that we take something to be 'right' if it is permitted (hence not 'wrong") but to be our duty if it would be 'wrong' not to do it. There are, in the context that these points are made, also some additional points raised about what really does and does not belong to the 'moral' point of view. Thus, unlike Sidgwick, Parfit does not take rational egoism to be a 'method of ethics' but to be an external rival to ethics. The reason for this is that rational egoists operate with a normative conception of 'decisive reasons' that does not take account of moral considerations since the only decisive reasons are ones that refer to personal effects (advantages and disadvantages to the self) without broader consideration of the rightness or wrongness of the acts they point to.

Similarly, consequentialism in the general sense that was rejected when the 'fact-relative' notion of 'wrongness' was undermined is also taken to be an external rival to the moral point of view by Parfit (again, in apparent contrast to the view of Sidgwick). Although adopting wholly impartial reasons for actions involves acceptance of sacrifice in relation to ourselves and thus appears to indicate that we have a moral consideration here in issue Parfit points out that adopting this standard as such (which is equivalent to crude act-consequentialism) involves subordination of all normal moral concerns. Hence someone who adopts this view is led to scepticism concerning duties, blame, remorse and indignation and thus is "an external critic" concerning morality. In making this point Parfit appears to have conceded much of the case against act-consequentialism made by such writers as Bernard Williams.

No comments: