8.0 Non-analytic subjectivism
Christian next turns to non-analytic subjectivism. As someone who doesn’t think there are reasons, this is a worrisome beginning:
Here are the rules of the philosophy game we’re about to play. I’m going to propose a principle which tells us when we have a reason to act. I’m then going to raise various thought experiments that will prompt your intuitive responses. Some of these thought experiments will be outlandish and some will be tame. The point is to evoke your underlying beliefs about when we have a reason to act or not.
Given that I don’t think we have “reasons,” this is going to be a challenge for me, but let’s proceed:
Desire-based Theory: We have reason to do whatever would fulfill any of our present desires (OMW I, 58) [sic]
Since I don’t know what this question is asking, in that I don’t know what’s meant by “have reason,” I am unable to answer the question. I think talk of “reasons” only makes sense when its meaning is made clear in the relevant context. If someone asks about the reasons for why water turns to ice when it’s cold, there’ll be some explanation for this. If someone asks what reason I had to go to the store, I will report what my goals or motivations were. Reason-talk is, on my view, perfectly meaningful; it’s just that I don’t think we literally have reasons. So when someone asks me what my “reasons” are for something in some abstract, theoretical contexts, what are they asking me to do? I could convert their question into a question about my desires, in which case this question would just amount to:
I would desire to do whatever would fulfill any of my present desires.
Well, that’s true. So this statement is true, I suppose? But is that what I’m being asked? I don’t imagine Parfit intended to ask us to respond to a tautology. So what am I being asked?
I don’t know.
So there’s no way to respond. Note how Christian talks about this account, though:
On this theory there is no such thing as an irrational desire, all desires we have in the present moment provide us with a reason to act.
I agree that there are no irrational (non-instrumental) desires, though this remark doesn’t yet introduce a distinction between instrumental and non-instrumental desires. But what on earth would it mean to say that our desires provide us with a reason to act? If the claim is that our desires motivate us, that would be one thing. If a “desire” is operationalized by psychologists in terms of processes that motivate us to act, then this, too, would be rather empty. It would amount to saying:
All motivations we have in the present moment motivate us.
Christian seems to impute this ineffable notion of reasons, which desires allegedly can “provide” us, to the subjectivist. Next, we’re asked to consider this scenario:
Snake: While walking in some desert, you have disturbed and angered a poisonous snake. You believe that, to save your life, you must run away. In fact you must stand still, since this snake will attack only moving targets. (OMW I, 34). [sic]
Christian says:
In Snake you have a strong, present desire, predicated on false beliefs, to run away. Yet most of us would agree that this desire does not provide us with a reason to run. If we share this intuition, we ought reject the desire-based theory. Subjectivist [sic] could patch up their theory by incorporating one of two further principles to account for this tension.
This is a serious problem, not for me, but for how critics of subjectivism understand subjectivism. A sensible subjectivist doesn’t think that we “should” do whatever it is we desire to do, where “desire” is crudely understood to refer to whatever instrumental means we believe will achieve our desires. They think we should do what would in fact achieve our desires. What do I mean by this? I mean that, all things considered, the course of action I would like to act upon would be the one that most accomplishes my desires. Only I don’t know what, precisely, will achieve this outcome, since I don’t have perfect information or reasoning abilities. Given this, we can draw a distinction between:
Non-instrumental desires
The fundamental desires we want to satisfyInstrumental desires
Given our limited information and fallibility, that set of desires or motivations we act upon in order to achieve our non-instrumental desires.
Let me illustrate the distinction. Suppose I am thirsty and would like to quench my thirst. I see a glass of what appears to be water on the table. Only it is not water, it is poison. Do I “have a reason” to drink the water? On the crude and inaccurate characterization of subjectivism, the answer is yes. After all, we’re told:
We have reason to do whatever would fulfill any of our present desires
The reasoning here is supposed to be that:
I desire to drink the glass of water.
However, since doing so would result in me dying, this wouldn’t be in my interests. The critic of the desire-based theory thus reasons that:
The desire-based theory holds that in the glass of water scenario, the desire-based theorist is committed to the belief that people have reason to act on their present desires.
Since they presently desire to drink the glass of water, they will act in such a way that they will bring about their own deaths.
This is counterintuitive. We shouldn’t favor a theory that holds that we have reason to bring about our deaths.
So the desire-based theory should (all else being equal) be rejected.
Yet there is a very clear and obvious mistake in this line of reasoning:
It’s simply false that the desire-based theorist’s only present desire is to drink the glass of water.
This line of reasoning fails to distinguish instrumental from non-instrumental desires. The desire-based theorist should (and probably does) think more like this:
I non-instrumentally desire to quench my thirst.
I don’t want to die.
This here is a glass of water which will quench my thirst.
I instrumentally desire to drink this glass of water in order to quench my thirst.
The desire-based theorist can thus identify the mistake being made as being a mistake with respect to one’s instrumental desires. But why is it a mistake? Why does the person in this scenario, who wants to quench their thirst and not die, desire to drink the water? Because they mistakenly believe it will achieve their fundamental desire. So the desire-based theorist can easily account for the “intuition” that the person does not “have a reason” to drink the water. They can easily explain the situation like this:
It would not satisfy their desires to drink the poison.
They mistakenly believe that it would.
Now, let’s go back for a moment to the original characterization of this account. We were told:
On this theory there is no such thing as an irrational desire, all desires we have in the present moment provide us with a reason to act.
In the present moment, my desire is to quench my thirst, and I believe that the means by which I can achieve this desire is to drink the glass of what appears to be water. While we can say I desire to drink the water, this is an instrumental desire. Instrumental desires can fail to achieve our non-instrumental desires. There just aren’t any issues here, unless, and only unless, one keeps trying to inject “reasons” into their account all willy-nilly. Subjectivism is entirely consistent with our instrumental desires being “irrational,” insofar as this consists in acting in ways inconsistent with our non-instrumental desires. However, this phrasing would be misleading. One might have no good evidence to believe what appears to be water is poison. A person mistakenly acting in a way inconsistent with their desires isn’t “irrational,” they’re just incorrect about what would achieve their non-instrumental desires. That’s all that’s going on in this scenario. There’s no need for fancy notions of rationality or normativity. Yet Christian’s claims would leave us with something like this line of reasoning:
Do we have a reason to drink the glass of apparent water? No! Because it would harm us! So the account that says we do “have a reason” must be wrong!
It’s easy to beat up on bad ideas. The idea here is that it’s counterintuitive to think you “have a reason” in this case. I don’t know what judgments or phenomenology are associated with people who’d report something like this, but my guess is that they think roughly that facts about what “count in favor” of an action shouldn’t be based on instrumental desires because instrumental desires can be mistaken and result in undesirable outcomes. Christian seems to treat the best solution for this to be to locate reasons elsewhere. But there’s a much simpler solution: abandon the notion of normative reasons entirely. If we construe this and any subsequent scenario entirely in terms of:
Psychological facts about what an agent believes
Facts about what is true or false
We can easily describe all of these scenarios without ever wondering whether any given set of facts, including facts about our desires, “give” us reasons. We can just say that there are facts about what our non-instrumental desires are, facts about what our instrumental desires are conditional on the information available to us in a given context, and facts about whether the instrumental desires would or wouldn’t achieve the non-instrumental desires. We don’t need to say anything else. Unless, of course, we have realist intuitions that aren’t satisfied by these accounts. But if that’s the case, of course these accounts aren’t going to appease such people. So what? It isn’t a requirement of a philosophical position that it satisfy the intuitions of people who have intuitions to the contrary.
When I, as an agent, am trying to decide what to do, my goal is always to try to achieve my non-instrumental desires by means of the instrumental desires I believe are most conducive to achieving those non-instrumental desires. We can then ask a host of epistemic questions about what course of action is most likely to yield success for me as an agent with bounded rationality operating under a given set of constraints. There are lots of questions to ask in such circumstances. But the question of whether any given set of instrumental desires, non-instrumental desires, or non-desires “give” me reasons never arises.
Parfit’s approach seems to kick up so much dust via reasons-talk that desire-based accounts look silly and indefensible. And they do look silly and indefensible if you twist them into pretzels by trying to force some intuitively satisfying notion of “reasons” into the mix. But such intuitive satisfaction is, I believe, a profound error; a result of buying into the reified and/or conceptually confused reasons-talk invented by philosophers, a form of thinking and discourse that clings like a parasite to some tentative conception of what we’d think and say in ordinary discourse, only extracted from those contexts and dissected outside of them in ways that appear to reveal anomalies: but those anomalies are only present due to the artificial armchair conditions in which the terms and concepts are being assessed. Philosophers are employing a method that can’t work, for much the same reason you cannot understand wild animal behavior by capturing animals and placing them into zoos. To understand how language works, you need to study it in the wild. Not study it in the armchair and insist that such study reveals its use in the wild.
Returning to Christian’s Snake scenario, it is easy to reanalyze this with these tools in hand:
I non-instrumentally want to avoid being bitten by a snake.
I mistakenly believe running away would achieve this desire.
Insofar as I could speak of “having a reason” to do things, I would cash it out in this way:
Given my mistaken beliefs, I “have a reason” to run away from the snake in the sense that I instrumentally desire to run away and am motivated to do this.
Given my non-instrumental desires, I “have a reason” to not run away from the snake in the sense that doing so would satisfy my desire to not be bitten.
The subjectivist faces absolutely no difficulties whatsoever in addressing this situation. The only apparent difficulties emerge from the confusions generated by talking about “having reasons,” and treating these as literal things one has. If one objects that I don’t actually “have a reason” to run away from the snake, because it won’t achieve my non-instrumental desires, then what sense are they using “have a reason” in? I can explain how I’d use the term; can they? Whereas they are stuck with an ineffable, apparently inarticulable sense that something about this or that scenario is “counterintuitive,” I can, in all such cases, discharge the meaning of the terms in familiar ways, then assess whether the statements are true or false, without need of intuitions. My account is better. It’s cleaner. The only thing it doesn’t do is appease realist intuitions. But again, I deny this is a requirement of a satisfactory account, for much the same reason a proponent of natural selection doesn’t have to satisfy the intuitions of people who insist nature appears to be “designed” and can’t shake the intuition. So much the worse for intuitive creationists. Just so, so much the worse for intuitive normative realists.
A desire-based theorist should adopt quietist subjectivism, and in doing so deny that any of their desires, instrumental or otherwise, “give” them or “provide” them with reasons. All the supposedly counterintuitive implications of subjectivism emerge from confusions generated by speaking in these terms. The way out for the subjectivist is to reduce and discharge all talk of having reasons by translating it into talk of instrumental and non-instrumental desires. Once they do, absolutely no problems for their position emerge at all.
I don’t need any intuitions to judge that, in a hypothetical scenario in which a person mistakenly thinks running away from a snake will serve their interests, and they are motivated to do so that they think doing so would serve their interests and are motivated to do so. This is simply what’s been stipulated. If that seems like a hollow analysis of the scenario, good. I wasn’t the one who adopted what I regard as a misguided method of analyzing such scenarios to demonstrate how one doesn’t have “reasons.” It isn’t a problem for me if the scenario doesn’t present any actual problems on my view; that’s the whole point. My approach to these issues doesn’t solve the problem by providing an answer; it solves the “problem” by showing it was never a problem in the first place. And this is exactly what I think about much of the way not only Parfit argues for his positions, but analytic philosophers in general: I am explicitly accusing them of using bad methods to generate pseudoproblems, then presenting their theories as solutions.
Christian (Parfit, too) is already aware of the instrumental/non-instrumental distinction, and moves on to allowing the desire-based theorist to introduce the distinction to “rescue” their position from absurdity:
Subjectivist [sic] could patch up their theory by incorporating one of two further principles to account for this tension.
(E) reasons are provided only by desires that depend on true beliefs
or
(F) reasons are provided only by terminal desires, or aims
This position is still inadequate because it continues to cling to ineffable notions of “reasons,” and so I still reject it in favor of a quietistic account that rejects what I’ve diagnosed as confused reason-talk.
I also want to draw attention to the unfortunate way Christian frames this: That the subjectivist can “patch up” their theory by making these moves, as if they hadn’t already done so in the first place. This is unfortunate because it is a common way of framing rival positions employed by philosophers. While I doubt Christian intends to employ rhetoric or strawman subjectivism, this way of framing things nevertheless serves this function even if it is entirely innocent and unintentional.
Let’s call it the Matryoshka Fallacy. My mom used to have a set of Matryoshka dolls: a set of hollow, painted wooden dolls, each designed to fit inside the other. When you opened the largest one, you’d find a smaller doll inside, and when you opened that, a smaller one still, until you reached the final doll.
Sometimes, when someone is describing an opposing position, instead of simply misrepresenting the position and then attacking that misrepresentation—a classic strawman—they present various weaker forms of the position, note deficiencies with these, and then state that the proponent of the view can “fix” or “patch up” or otherwise address shortcomings with the position by modifying or adding onto it. This ends up crafting a narrative in which the actual position a person holds is depicted as if it were the result of a desperate retreat that has resulted from multiple concessions and ad hoc maneuvers. The narrative looks something like this:
The rival position endorses X. But here’s a problem with X. They could shore up this weakness by adopting X*. But there’s a problem with X*. One way out of this is to adopt X**. But, alas! That suffers from a problem of its own. Of course, they could bite the bullet and retreat to X***…
This is misleading when the person endorsed X*** all along. It makes the position they already held appear to be this battered, beaten thing, not because it’s been battered and beaten, but because you’ve depicted it that way.
I already construed desires in terms of instrumental and non-instrumental desires, and already would’ve readily been able to respond to Parfit’s critiques of desire-based accounts before ever encountering Parfit. I imagine the same would be true of any competent subjectivist.
The positions Parfit is targeting here are crude and unsophisticated and seem artificially engineered to be vulnerable to a series of critiques. It’s as though Parfit is just lining up a series of strawmen he can knock down so that he can depict the final opposition as a craven, retreating position fleeing in terror from Parfit’s philosophical blitz. In any case, Christian insists these moves aren’t going to work:
While (E) might seem more plausible at first, we can post-pone [sic] any updates to the Desire-Based Theory by appeals to truth. Instead, we can appeal to (F). This would result in the simplest plausible subjective theory. To motivate (F) consider Apple.
Now we’re given a rather extensive hypothetical:
Apple: Suppose that I want to eat the two remaining apples that are on some tree. I also want to climb a ladder so that I can reach the higher apple. Suppose next that this tree’s owner allows me to eat only one of these apples, and lets me choose which apple I shall eat. If instrumental desires gave us reasons, I would have more reason to choose the higher apple. If I chose the lower apple, I would then fulfill only my desire to eat this apple. If I chose the higher apple, I would fulfill not only my desire to eat this other apple, but also my instrumental desire to climb this ladder so that I can reach this apple. [So I therefore should choose the higher apple as two desires would then be satisfied]. But this reasoning is obviously mistaken (OMW I, 59). [sic]
According to Christian:
As Apple highlights, instrumental desires do not provide us with reasons. If they did, a life well lived would merely consist in causing ourselves to have many arbitrary desires, and then satisfying them. If you agree at [sic] the absurdity of the situation, we can incorporate (F) by updating the Desire-based theory to
Christian doesn’t quote the rest of what Parfit says in this passage. Parfit adds:
This reasoning is obviously mistaken. Since I want to climb this ladder, not for its own sake, but only as a means of reaching this apple, I have no further, independent reason to fulfil this desire. My reason to climb this ladder derives entirely from, and adds nothing to, my reason to fulfil my desire to eat this higher apple. (OWM, vol. 1)
What Parfit is trying to do is show that instrumental desires can’t “give us reasons.” Parfit is right, of course. Instrumental desires don’t give us reasons. But who thought they did? Perhaps one of the Matryoshka dolls. The problem is, as before, Parfit’s own confused talk of reasons generating the apparent problem. Parfit’s scenario also doesn’t make much sense, and may even be incoherent. Let’s unpack the scenario:
1. I non-instrumentally want to eat two apples.
2. One is higher in a tree, one is lower in the tree.
3. I can only eat one apple.
4. I instrumentally want to climb the ladder to eat the taller apple.
5. If instrumental desires gave us reasons, I would have more reason to choose the higher apple than the lower apple.
This whole scenario only generates problems if the desire-based theorist has weird views about reasons and thinks instrumental and non-instrumental desires alike both “give” us reasons, and that instrumental desires give us reasons independent of what would be needed to achieve our non-instrumental desires. This way of characterizing reason-giving also collapses the distinction between instrumental and non-instrumental desires into both giving us just one kind of “reason.” But why think my “reasons” for acting in accord with my non-instrumental values are on the same footing as my reasons for acting on my instrumental desires? There is no good reason for this: it only makes sense to think of us as having a reason to act on instrumental desires insofar as doing so would serve some non-instrumental desire. But Parfit has given us a scenario in which someone has a free-floating instrumental desire. This makes no sense. It’s just not how instrumental desires work.
Look at this feature of the scenario:
I also want to climb a ladder so that I can reach the higher apple.
Later on, Parfit says:
If I chose the higher apple, I would fulfill not only my desire to eat this other apple, but also my instrumental desire to climb this ladder so that I can reach this apple.
There is a serious problem here. The only purpose of an instrumental desire is to achieve some non-instrumental desire. What non-instrumental desire is achieved by climbing the ladder? None! Parfit has stipulated a scenario that simply makes no sense. Either there would be some non-instrumental desire which would be achieved by climbing the ladder rather than not doing so, or there isn’t. If there is, then it makes perfect sense for a person to climb the ladder. If someone really enjoyed climbing ladders, or wanted to exercise when choosing apples, or wanted to get a better view from being up higher, then there’s nothing absurd or unreasonable about them climbing the ladder. If they don’t have a non-instrumental desire that’d be achieved by climbing the ladder, then it makes no sense to talk of them having an instrumental desire.
Parfit has set up a scenario that employs a concept in a way that omits an integral component of what would render the concept meaningful in the first place. In doing so, he creates the false impression that the problem is with the agent in the scenario, and the resulting analysis desire-based accounts would give of the scenario. But the problem is with Parfit’s set up.
But let’s set that aside for a moment to see what the issue is supposed to be. According to Christian:
As Apple highlights, instrumental desires do not provide us with reasons. If they did, a life well lived would merely consist in causing ourselves to have many arbitrary desires, and then satisfying them. If you agree at [sic] the absurdity of the situation, we can incorporate (F) by updating the Desire-based theory to:
Terminal Desire Theory: We have most reason to do whatever would best fulfill or achieve our present terminal desires or aims.
I don’t think this scenario illustrates much. Subjectivists are not committed to thinking that it would be a good thing to satisfy instrumental desires for their own sake. Yet Christian’s remark suggests that the subjectivist has a goal like:
Maximize the number of instrumental or non-instrumental desires fulfilled by one’s actions.
This just isn’t something subjectivists are committed to. So it does not follow that a life well-lived would, for the subjectivist, consist in generating then satisfying desires. So Christian has not identified an absurdity with the scenario. I want to satisfy whatever my desires are. I have no desire to have desires in order to satisfy them. We don’t need to consider instrumental desires “giving us reasons” to move on to the terminal desire theory. This is already what a subjectivist is likely to favor.
I still think the subjectivist should reject this and any other account which employs an ineffable conception of reasons. I don’t endorse the view that what we have “most reason” to do is whatever would best fulfill or achieve our present terminal desires or aims. Christian also suggests subjectivists may want to incorporate a truth requirement into the position, but this would depend on an empirical question:
Can beliefs cause us to have a different set of terminal desires or aims?
I don’t know the answer to this question. But I also don’t care much about incorporating truth conditions into my own account, so I’m not interested in pursuing this line of inquiry. Christian seems to believe adding a truth condition to one’s subjectivist account would enhance its plausibility. I don’t. Roughly, the idea would be something like this. At a given time, I have one set of non-instrumental desires. However, with new information, my beliefs may change, and if these could cause my non-instrumental desires to change, then I’d have “most reason” to do something other than what I seem to have most reason to do in the absence of this information. We can imagine scenarios where, in other words, my non-instrumental desires are based on false beliefs. In these cases, I’d simply note that one would be motivated or desire to achieve outcomes based on false beliefs. There’d just be a descriptive question here. And I wouldn’t have any further position on whether one “has a reason” to believe or non-instrumentally desire something else instead. This modification is so rooted in considerations of what one “has a reason” to do that it’s irrelevant to my own position.
Part of the reason I’m not that interested in this distinction is that any interest in correcting for one’s false beliefs such that they’d subsequently prompt changes in one’s desires would, for me, be predicated on whether one desires such changes. Desires are still in the driver’s seat, and it’s simply a further descriptive question whether one has the desire for their non-instrumental desires to be receptive to such changes or not. There’s no non-desire-based fact about what I have “most reason” to do that is given by facts alone, rather than my desires, which—consistent with subjectivism—can be responsive to facts. I see no need to entertain further complications Christian introduces and will thus hop off the metaphorical train here.
nice workup!