Attempts to avoid Rosati’s objection

If inclusion is problematic, there is the obvious choice of trying to do without it. For such an attempt to be successful, the naturalist must argue that there is no need for an ideal of the person.

Here are three such attempts: the first two concerns the ideal advisor theory only, the third is compatible with both actual and hypothetical desire theories. The naturalist can hold that the process of idealization is such that it will wash away the differences that arise from the different psychological make-up of agents. Railton suggests this when he says that since belief and desire co-vary systemically, i.e. those who have similar beliefs tend to have similar desires, fully informing agents will bring both their beliefs and their desires into line. (Railton 1997, 161 19n;

2003a, 57-8) And, as we saw, Smith holds that since idealization involves the systematic justification of desires, agents will converge on the same desires concerning a particular situation.

(Smith 1994, 165; also Firth 1952, 322) If this is so, no matter how we picture ideal advisors they will always be the same, hence they will always give the same advice to their actual self.

Attila Tanyi, THESIS, Page 93

Consequently, it is pointless for Sally to look for an advisor other than the one idealization produces: they will all tell her the same thing.

There are several problems with this suggestion. First, it relies on an unwarranted empirical claim. Nothing assures that beliefs and desires covary or that agents converge on their desires after idealization. To this, however, Smith says that the above point is conceptual not substantial, that is, it is a claim about our concept of reason and not about the world. The world might indeed disappoint us producing different ideal advisors; yet, this does not change the fact that this is not what we mean when we talk of ‘fully informed and cognitively perfect’ advisors.

(Smith 1994, 166) But even if we accept Smith’s controversial claim, we can still say two things.

First, we should not forget another point of his: that if all ideal advisors are the same, this means that our reasons are no longer anchored in our actual desires, which is exactly what the Model needs. Furthermore, we can point out that even if he is right and even if the world produces convergence, this will still not satisfy our needs. (Rosati 1995b, 57) For our question is which traits are important in an ideal advisor. Hence, even if advisors are the same, their advice may still not be authoritative for us because, say, they are all cold hearted, calculating or nihilistic. Of course, advocates of the view can defend the traits idealization produces. But, at this point of the argument, it is hard to see how they could do this without resort to normative judgments.

Certain of Railton’s remarks suggest another putative reason not to include an ideal of the person in our account of the ideal advisor. (Railton 2003a, 58, 60) He says that we should be careful not to distance the advisor too far from the actual agent because we need her to be someone who is like the agent. The idea seems to be this. If we keep the ‘non-belief properties’ of the advisor as close as possible to the actual agent, we thereby allow the agent to serve as her own ideal advisor. And since we ordinarily regard our own judgments as authoritative, at least insofar as the influence of our psychology is concerned, we would be more inclined to follow the

Attila Tanyi, THESIS, Page 94

judgments of our fully informed self were that self to be construed in a way that does not disrupt its continuity with the agent. What drives Railton here is the intuition that generally stands behind the ideal advisor theory (it leads to motivational existence internalism, as I will explain in Chapter IV): that the agent’s good or reasons should not be alien to the agent. For if they are not alien to the agent, they are also authoritative: or so the thought goes. And if we keep continuity with the agent, we will achieve just this.

I don’t want to question Railton’s internalist insight (I won’t do that even in Chapter IV).

Even so, Railton’s argument does not produce the required result. Note first that there are two ways the ideal advisor account can keep continuity with the actual agent. (Rosati 1995b, 59-60) One is that whatever suggestions the advisor makes should be constrained by facts about what the actual agent is like: her physiological and psychological constitution, capacities, circumstances, history and so on. (Railton 1997, 142) These facts determine the agent’s capacity to change, the costs of these changes and so on. (Railton 2003a, 54) And, no doubt, the advisor must take these facts into account when making a decision about what the actual self should desire. In this sense, then, any version of the ideal advisor theory should keep continuity with the actual agent. But there is another sense, and this is obviously what Railton is referring to above, in which continuity is a more controversial matter: it concerns how the facts should be viewed (and judged upon). What Railton suggests is that they should be viewed from an angle, which preserves as much as possible from the agent’s actual self; whereas what Rosati suggests is that they should be viewed from an angle that contains an ideal of the person (which, importantly, is endorsed by the agent herself).

But it is hard to see why we should privilege Railton’s viewpoint over the one proposed by Rosati. First of all, if we do so, we will loose the ability to allow the agent to pose certain questions about the reasons she has: this is just the problem Rosati’s objection uncovers. Second,

Attila Tanyi, THESIS, Page 95

the viewpoint Rosati proposes, i.e. the viewpoint that incorporates an ideal of the person is also meant to keep the continuity between the agent and the advisor: it keeps continuity with those

‘non-belief properties’ that the agent identifies with and throws out those that she does not endorse. (Rosati 1995b, 60) Third, there is good reason to think that the ideal advisor as Railton reads him would not be continuous with the agent after all. Recall Rosati’s previously mentioned, though in detail not presented, argument. It is supposed to show that the idealization not merely changes the agent but perturbs his psychology drastically. Hence, instead of getting someone who is continuous with the agent, we get someone who is dramatically different from him, who is a stranger to him. And to establish an appropriate connection between advisor and agent, the best we can do is to appeal to an ideal of the person, which the agent shares and our account of the advisor incorporates.

This takes us to the third attempt. It makes use of a simply intuition that might be taken to back Railton’s preference. For a proper presentation, let me switch from reasons to the agent’s good since it is in this respect that the idea typically appears. It goes like this. We can make a distinction between the agent’s good and personal ideals such as moral or aesthetic ideals. We can further add that while the former is attached to the agent’s present standpoint as defined by his present motivational capacities, the latter is about which standpoint to prefer. Rosati’s question clearly belongs to the sphere of personal ideals since it prefers a standpoint that incorporates an ideal of the person. Hence Rosati’s objection does not point to concerns about the naturalist construal of the agent’s good as it appears in the ideal advisor view. As a result, though the naturalist account does indeed leave questions open, these questions are not about the agent’s good; insofar as this notion is concerned the ideal advisor view is tolerably revisionist.

This is, I believe, the crux of the matter in the present debate. Although advocates of the ideal advisor theory do not in press mention this objection (Rosati reports that the objection was

Attila Tanyi, THESIS, Page 96

made by Railton in personal communication), both Hubin and Noggle explicitly defend their view in this way. (Rosati 1995b, 59 note 45; Noggle 1999, 323-6; Hubin 1999, 41-2) I, however, don’t think that this response settles the controversy in the naturalist’s favour. As a way of ending discussion let me offer two reasons. First, as noted, the response is typically made in discussions about the agent’s good. But it is not clear whether the same kind of distinction can be made in the field of reasons. It seems that insofar the justification of conduct is concerned questions about standpoints, i.e. about what sort of person to be may also be relevant and legitimate.⁴⁸ In fact, it is not at all clear why the claim would be true even about the agent’s good. This, it seems to me, is a rather elusive matter in which decision does not come easily: it needs an argument, not just a declaration of position. Second, rejecting the idea that determining the agent’s good (or reasons) involves questions about what sort of person to be is also in the business of choosing standpoints.

For this is exactly to say that the preferred standpoint is provided by the agent’s current motivational system. And this puts us back to the situation we have left Railton’s suggestion with: that a reason needs to be given that justifies preference for the agent’s current standpoint.

But, as we saw, no such reason is presented.

2.3.6. Summary

Let me sum up. The previous discussion has shown that (1) naturalist accounts of the Model leave important questions about reasons open because of their lack of an ideal of the person, that (2) though including such an ideal is not impossible, no acceptable account has yet been proposed, and that (3) trying to do without such an ideal is not a plausible solution to the problem. As a result, we can conclude that the Model is intolerably revisionist and that therefore the triviality charge follows. This last point shows, however, that an important shift has taken

48 Except Hubin (1999), pp. 42 who deals with reason and thinks that such questions are preconditions to the agent’s reasoning, not elements of it.

Attila Tanyi, THESIS, Page 97

place since our first encounter with Parfit’s argument. While Parfit’s charge is intended it to be a knock-down argument against naturalism per se, the condition of tolerable revisionism leaves open the possibility that one day a proper naturalist account of reasons, one that is compatible with the Model, will be proposed.⁴⁹