I'm working with three criteria; a successful reasons-explanation of action must:
- Account for the connection between reasons and action in the agent
- Give reasons that the agent genuinely finds worthy
- Fit the action within the pattern implied by (a) and (b).
It's important to note, I think, that there's room in this T-model for the claim that, sometimes, I don't actually act for any reasons at all, even though I very much try to. This would happen if something I thought was an end I could achieve turned out to be unachievable. It's important to draw this out as many traditional models of reasons-explanations of action don't have this feature. That is, much as Descartes et al elevated "seeming" into an unimpeachable theoretical activity, these models elevate "trying" into an unimpeachable practical activity. (I'm about 75% sure this is Robert Brandom's idea. Been a while since I read Brandom.) Retreating to "seeming" allows me to never make theoretical (e.g., epistemic) errors. My theoretical reasoning can never go wrong if I can always claim that I wasn't reasoning about what actually was the case, only what seemed to be. Similarly, my practical reasoning can never go wrong if I'm not reasoning about what I can actually achieve, but only what I can try to achieve. Many traditional models of reasons-explanations of action, in part because they are exclusively psychological, fall into this trap. I can always find some erroneous belief, say, that legitimates my course of action; I was just trying to do what the belief, mistakenly, told me was possible. I want to resist this move because, as said, it looks just like the seeming move: it's a fairly cheap way to resist the conclusion that sometimes we just go badly wrong. That is, sometimes we set ourselves goals that, although we tried as hard as we could, and reasoned as best we were able, just can't be reached. We thought we were acting for reasons, but we actually weren't. (Note that I've now basically kicked purposes to the curb; ends have to be understood as goals, or else the "trying" problem re-emerges.)
Back to the main line of argument. The T-model can, potentially, satisfy the first of the three criteria given above. The particular T-model chosen will affect whether or not the criterion is satisfied. But, in principle, there's nothing more or less mysterious about the connection between reasons and action being the connection between (approximately) ends and means than being the connection between causes and effects. Both sorts of connection need to be analyzed in more detail, but the analysis is, prima facie, possible.
The T-model can account for the second criterion rather more easily than most other models, the causal model in particular. To be an end for some agent just is for the agent to find the end valuable. Agents don't have ends that they find totally worthless; any such "end" is either really someone else's end, or not best characterized as an "end" at all. In the latter case, I'm thinking of cases of compulsion and the like, where an agent pursues "ends" that the agent doesn't want at all; in this case, I'm inclined to call the "end" no end at all, but instead something like a "result". A smoker who doesn't want to smoke any more, but does anyway, has smoking as a result of some of his actions. But it seems bizarre to me to call smoking his "end" when, ex hypothesi, he doesn't want to smoke any more!
The connection between means and ends, and the value the agent sees in the end, sets up a kind of pattern, as required by (c). We might call this a pattern of efficacious increasing of value. Really, though, this pattern is something of a pseudopattern; there's lots of detail that's obliterated if we try to capture the pattern as a unit. I think it's better to consider the efficacious increasing of value as a category of patterns, including patterns where one increases moral value (e.g., leading a virtuous life), material value (e.g., seeking wealth), practical value (e.g., survival), aesthetic value (e.g., the life of Gauguin), and so on. My examples are generally global, but I don't see any grounds for saying the efficacious value-increasing patterns have to be this way; they could be entirely local. This sort of thing would occur when someone does something "out of character": when Ned Flanders snaps and curses out pretty much everyone in Springfield, for example. That's a pattern of action, but one that's generally foreign to (at least the adult) Ned.
Characterizing the T-model more precisely then requires, I think, two things:
- An account of the connection between means and ends
- An account of value, at least in sketch
(2) is a nasty problem, and I don't know, at this point, how to approach it. (1) is less nasty, although I don't have any firm opinions on it. So, I'll close with some sketches of two views on this subject, from GF Schueler and Scott Sehon. I'm more inclined towards Schueler's position, but not very strongly. My biggest problem with Sehon, I think, is that he doesn't really do (1). There's some gestures, but no significant characterization. Schueler is better in this regard but is, I think, too enamored of the psycholgistic accounts of reasons-explanations to go where he should -- i.e., goals -- instead of where he does -- i.e., purposes. It should be fairly obvious, though, that much of what I say above is drawing off Schueler's theory.
[Précis of: Sehon, Scott R. 2005. Teleological Realism: Mind, Agency, and Explanation. Cambridge, MA: The MIT Press. (Amazon | MIT Press)]
Scott Sehon offers an explicitly teleological model of reasons-explanations of action. Says Sehon, "[t]eleological explanations explain the behavior by citing the state of affairs toward which the behavior was directed" (Sehon 2005, 135). We do this by assuming the agent is rational; for Sehon, teleologically explaining A's φing in order to ψ requires finding "a ψ such that φing is optimally appropriate for φing, given a viable theory of the agent's intentional states and circumstances." (ibid., 146).
The reliance on the rationality of agents is due to Sehon's claim that reasons-explanations of action are supposed to make the behaviour of agents make sense and thus, following Donald Davidson, show the agent to be maximally rational: "consistent, a believer of truths, and a lover of the good" (ibid., 139, quoting Davidson). Sehon identifies two axes or sorts of rationality: the appropriateness of behaviour for goals and the extent to which agents find their goals valuable (ibid.). So, in teleologically explaining action, we are showing that, given their circumstances, epistemic situation, and intentional states, agents' actions are both appropriate and serve some kind of practical value, such as desire-satisfaction or being pleasing (ibid., 139-40). States of affairs and other objects acquire value for Sehon by fitting into an overall pattern of life, i.e., a system of goals (ibid., 162-3). It should be noted that, as with Davidsonian radical interpretation, it is possible for Sehon's agents to be in error, if assuming the agents were correct in a given case would introduce other errors into the intentional system, or imply greater mystery in physical theory than is the case with other interpretations (ibid., 140).
Sehon is, oddly, against the idea that the reason for A's φing need be any sort of state. He agrees that it could be, but he claims that the natural way to read A φed in order to ψ is that A's reason for φing was to bring about ψ. "It is a mistake to look for a thing that is the agent's reason for φing, whether the thing be a state of affairs, fact, or psychological state." (ibid., 149). Furthermore, A's reason for φing may be whatever accounts for A's valuing of φing: A's desire for ψ or A's acceptance of ψing as a moral requirement are both equally good candidates, for Sehon (ibid., 149-51).
The ontological picture underneath this is a little complex. Reasons aren't real psychological states, nor real states of affairs. In fact, for Sehon, teleology itself is just irreducibly real (ibid., 172). We understand persons teleologically, he claims, and to try to do without teleology is to omit a large and important sort of knowledge and understanding (ibid., 224, 228). Teleology is thus not subject to elimination, for Sehon accepts a modest realist criterion of existence, such that whatever we can't do without in order to not leave inexplicable circumstances of mysteries must exist (ibid., 18-19). While teleology itself is, he acknowledges, a large mystery on this account, leaving it out seems to leave a greater gap (ibid., 171-2).
Once elimination is off the table, we must either reduce teleology, go to some sort of dualism, or adopt a nonreductive approach. The median is less simple than the latter, so it is immediately out of the running, except as a final default. Reduction, says Sehon, is a failure because of the impossibility of finding bridge laws that allow us to deduce the concepts of intentionality from the concepts of a (presumably more fundamental, on a physicalist picture) physical science (ibid., 29-30, 128). So, Sehon adopts a nonreductive account of teleology, founded on a supervenience relation between human mental facts and physical facts, such that the latter constitute the former (ibid., 114-6, 130). In short, then, human reasons for action -- purposes -- are constituted by physical facts and teleologically explain actions simply because we cannot understand persons as persons otherwise.
G. F. Schueler
[Précis of: Schueler, G. F. 2003. Reasons and Purposes: Human Rationality and the Teleological Explanation of Action. New York, NY: Clarendon Press. (Amazon | OUP)]
According to Schueler, reasons-explanations must refer to purposes, at least implicitly, or fail to explain (Schueler 2003, 1). Purposes, for Schueler, are assigned to objects or events, which requires that there be someone who assigned them (ibid., 4). In this way, purposes differ from functions, which, for a given object or event, are entailed either by its causal role or by its evolutionary history (ibid., 5). To say something has a purpose is to say it should or is supposed to do something, and not that it actually is doing this thing, or even that it could. Thus, nothing about what something is, or its history, or its activities, is relevant to determining its purpose (ibid., 6). Schueler does allow, though, that certain restrictions must apply: for one, "you can't assign a purpose to the nail on my wall." (ibid., 7, italics added).
The importance of purposes is that reasons-explanations succeed in explaining, when they do explain, partly by picking out the purposes. And this is not inconsistent, Schueler says, with the claim that actions are events and thus causally explicable (ibid., 8). Indeed, says Schueler, there's a sense of causal explanation such that to say x caused y is to say no more than y can be explained by x (ibid., 17). Regardless of how to understand causal explanations, of central importance for Schueler is that the purpose-based, teleological explanations of action cannot be further analyzed in other (non-teleological) terms (ibid., 20, 38, 56). Causal explanations are just one example of a failed to attempt to provide this analysis (ibid., 55).
When purposes explain an action, such as φing, says Schueler, the agent takes the importance of φing as a reason, while an observer would take the agent's pro-attitude towards φing as being his reason (ibid., 58-9). So, the agent's perspective is essentially normative in a way the observer's is not (ibid., 60). For Schueler, the agent's perspective is primary: "[i]t is only once we know how an agent understands the considerations she regards as relevant, which ones she holds more important, etc. … that we can figure out which, if any, pro attitudes to ascribe to her." (ibid., 67). Thus, we need not only understand the agent's purposes, but also what sort of character the agent has: for then we will understand why the agent takes certain features as important, and overlooks others (ibid., 69, 71). Character traits, for Schueler, are not an eliminable or reducible part of a reasons-explanation (ibid., 76).
A trait explains an agent's mental states, and allows an explanation to escape the chain of reasons. Any further explanation would not be a reasons-explanation, and would not need to be offered from the agent's perspective (ibid., 78-9). For Schueler, there is a difference between doing something generous and being generous: a complete set of extensionally equivalent mental states omits the role played by the whole set, which constitutes the trait (ibid., 77). To attribute a trait to an agent is just to say that "certain sorts of facts count for her as reasons (perhaps of a certain strength) to do certain sorts of things." (ibid., 81). These attributions are inherently evaluative, as they deploy what, Schueler says, Williams called "thick" moral concepts (ibid., 83).
Agents take things as valuable because of their characters; to take something, in this case, is to reason about it. An account of reasoning, for Schueler, must be an account of decisions between alternatives -- at least the alternatives of φing and not φing (ibid., 91). Practical reasoning is always about actions and has, as its conclusion, a claim about what one should do (ibid., 99). Theoretical reasoning follows the same model, where the action is the forming of a belief (ibid., 101). The difference between the two forms, says Schueler, is that beliefs have a test for success beyond justification, namely truth; there is no parallel for practical reasoning (ibid., 103). (Schueler just declares this, without considering some obvious examples, such as whether the action is sensible or appropriate.) For both theoretical and practical reasoning, reasons are reasons because they reflect "some value, a value derived from the purposive activity in which the agent is engaged." (ibid., 104). (On the face of it, this seems to be in tension with the idea that reasons derive their value from being held as valuable by agents with certain character traits. I suspect, though, that Schueler would want to say that the character traits make the agents sensitive to particular values associated with particular purposive activities. So, the values are in the activity, but the agent's character traits lead him to recognize the value and consider it important.) The epistemic values in theoretical reasoning derive from the activity -- belief-formation -- theoretical reasoning is used for (ibid., 104-5). So, practical reasoning generally, including theoretical reasoning, must show that "there is something to be said for doing whatever is being considered", some value that it serves (ibid., 105, italics removed). Correct practical reasoning requires appropriately weighing the facts being considered (ibid., 107). Facts or states of affairs are all that could give one reasons and thus are all that can be considered in practical reasoning (ibid.). Even factual errors or states that do not obtain can be used in practical reasoning (ibid., 114-5). Finally, practical reasoning must be evaluative, because it must yield a conclusion about what one ought to do. This is not possible if practical reasoning only takes into consideration descriptive claims of fact (ibid., 130).
In Schueler's view, explicit reasoning is the paradigm of acting for reasons. In practice, we may omit this explicit deliberation, but we cannot understand what we do without interpreting it as explicit deliberation (ibid., 136). Further, if one could have deliberated, and found better reasons and thus better courses of action, this is, in Schueler's view, grounds for criticism (ibid., 147). If the deliberative model is wrong, says Schueler, it must be impossible to act on deliberation. So, given that this is not impossible, the model must be correct. It understands agents as genuinely rational and thus is a wholly evaluative model (ibid., 149-50). By adopting the deliberative model, we see agents as rational in their selection of means and also of ends (ibid., 155). This two-pronged rationality is, for Schueler, the ultimate basis or ground for all other traits of character (ibid., 165).