Phil 33401, A02—Lecture 6 notes
I. Causation and explanation
We’re now going to move away from the specific details of Putnam’s and Burge’s arguments for externalism, and also at the specific externalist theses at which they arrive (on the one hand the claim that certain beliefs and propositional attitudes are partly constituted by one’s interactions with natural kinds, and on the other with the claim that a wide array of one’s propositional attitudes have their contents partly constituted by how others in one’s linguistic community use words.
We’re going to just assume for sake of argument that the general externalist claim is true—namely, that in at least some cases that a person has a propositional attitude with a particular content doesn’t supervene on her internal physical constitution but rather is constitutively dependent on her relationships to her environment. The question of what aspects of her environment—be it the nature of the natural things she encounters, or be it facts about her socio-linguistic world—help constitute her propositional attitudes’ contents we shall for this unit and the next abstract from.
For we’re going to look at two worries one might have about externalism that apply regardless of the specific details of one’s externalist position.
The first worry is this: content externalism has the effect of making what one believes, desires and intends completely irrelevant to the explanation of anything one does. If we accept content externalism, so goes this thought, we have to give up on the idea that the contents of one’s beliefs and thoughts (insofar as they’re externally constituted) have no explanatory bearing on anything one does. Mental content is a pointless residue.
Obviously this is a view one is loathe to accept. Hence it can seem strong reason to give up content externalism.
As we’ll eventually see, however, some argue that what content externalism forces up to give up is not the explanatory role of content-involving mentality as such, but rather a particular way, a broadly speaking mechanistic way, in which we are inclined to conceive that role.
More on that later. Before getting anywhere we need to look a bit at how the related issues of causation and explanation have played out in contemporary philosophy of mind. In fact, we need a fair bit of groundwork laid out before we can turn to the Dretske reading.
II. Rational-psychological explanation
Let’s begin by introducing a term of art.
Let us call rational-psychological explanation the ordinary, familiar practice of attempting to explain people’s actions through the ascription to them of beliefs, desires, intentions, moods, emotions and so on. I might say, F left the room because he believed this would be an especially boring lecture and wanted to avoid being bored. That’s an example of rational psychology. It’s a very simple one. We come up with more subtle explanations of people’s behavior in rational-psychological terms all the time, and you can find endlessly complex and nuanced such explanations in say, the novels of Dostoevsky or James (so-called ‘psychological novels’).
Now one crucial feature of rational-psychological explanation, one that distinguishes it from other forms of explanation, is the crucial involvement of the idea of rationality, or less heavy-handedly, of making sense. Of course, all explanations attempt to make sense of what they’re trying to explain. Explanation just is making sense of things, be it the workings of the physical world, the behavior of an animal, or the actions of a friend. But rational-psychological explanation incorporates that notion in a special way, which we might put like this: the project of explaining someone’s actions in terms of her beliefs and desires is a matter of showing, as far as possible, how the person’s action could have made sense to her. Such an explanation works by making a person’s action intelligible in a very particular way—namely, by showing how the action might have seemed the thing to do to to the person in question, given her beliefs, her goals and so forth. We ought to be very hesitant, for example, to claim that F left the room because he thought the lecture was going to be exciting and he wanted to hear it. Such an explanation would present his action as senseless and so in that respect unintelligible.
Contrast explanations in, say, cellular biology, to take one example out of many possible. Cellular-biological explanation doesn’t explain the behavior of a cell by saying doing such-and-such at this point—say dividing—made sense to the cell in light of its goals or what have you. How such explanations do explain is a question for the philosophy of science—it appears to involve both appeals to general physical laws and to hypotheses about natural selection. Regardless, the point that matters for us is just that such explanations don’t explain by saying how dividing seemed the thing to do to the cell. Such explanations, being explanations, attempt to make sense of cellular behavior, but not by showing such behavior to have made sense to the cell.
We could dwell on these points for a while, but now’s not the time. Call the claim I just made about rational-psychological explanation the claim that such explanations are rationality-involving.
Now, in addition to being rationality-involving, rational-psychological explanations also appear to be causal explanations.
The idea of a causal explanation is so fundamental it’s hard to develop at any length in other terms. A causal explanation is an explanation that explains a phenomenon by saying what brought it about, what produced, what, one can only say, caused it.
So take an explanation like this
F left the room because he believed the lecture would be boring and desired that he avoid boredom.
If what I was just saying is right, this explanation works, if it does, by showing how leaving the room could have made sense to F, could have seemed the reasonable thing to do. At the same time, it seems also to make a claim about what caused F’s leaving the room, an explanation of what brought about or led to his leaving the room.
For reasons that are very interesting, but that we can’t get into at, these two purported features of the explanation can seem to get into conflict.
There are many perceived problems about mental causation, but we’re only going to focus on one.
Now, the second crucial preliminary is a distinction between two kinds of causal talk, of talk about causes. It was a distinction to which Donald Davidson drew attention, offered a partial analysis of, and then applied to the case of rational-psychological explanation.
III. Singular causal statements vs. causal explanations
The distinction is between two kind of causal statements, namely singular causal statements and causal explanations.
Here’s an example of a singular causal statement:
The elephant stampede caused the abrupt ending of our lawn party.
Davidson had the following simple thought about what such statements do: they assert a relation between two specific events, more specifically, they assert that the one event caused the other.
Thus we are led to the following simple proposal about the form of such statements:
A caused B,
Where “A” and “B” are replaced by either descriptions or names of two events.
(Davidson also had ingenious suggestions about the forms of other kinds of causal statements, which I’ll bypass. Involves quantificational logic and takes us too far afield.)
Now given Davidson’s analysis, a singular causal statement is true just so long as what replaces “A” refers to or picks out an event that caused that which what replaces “B” picks out or refers to. That means that we can replace what replaces “A” with any description that picks out the same event or with any name that refers to the same event and preserve the truth-value of the original statement.
So if the elephant stampede was the only thing that happened while I was in the kitchen yesterday, then:
The only thing that happened while I was in the kitchen caused the ending of our lawn party.
You may recall that we can put this kind of phenomenon this way: Singular causal statements allow for the intersubstitutability of co-referring names or co-designating descriptions salva veritate.
Now consider the following:
That there occurred an elephant stampede explains why our lawn party ended abruptly.
Could one equally say:
That something happened while I was in the kitchen explains why our lawn party ended abruptly.
Clearly not. Even though the elephant stampede was the only thing that happened while I was in the kitchen, even thought there occurring an elephant stampeded explains why our lawn party ended abruptly, it clearly does not follow that there occurring something while I was in the kitchen explains why our lawn party ended abruptly.
Or again: The soprano’s singing of high E may explain why the glass shattered. Even if her singing of high E was in point of fact her singing of the most clichéd word in the whole aria, that she was singing the most clichéd note in the whole aria hardly explains why the glass shattered. Glass is sensitive to high frequencies, not to the presence of clichés. Similarly, lawn parties are sensitive to elephant stampedes, not to correlations between occurrences and my being in the kitchen.
Davidson himself was mostly mute on how to interpret sentences of this latter sort. But many people have taken the following view to be suggested by his treatment of singular causal statements.
Take a simple sentence of the form: O is p.
This sentence is about an object, O, and says of it that it is p. It predicates something of O. To put it another way, it ascribes a certain property to O. So properties correspond to the predicate place of sentences (not that we can’t nominalize talk of them).
Now events are different from objects. The soldiers who fought World War II are different from World War II itself.
But events like objects correspond to the subject place of a sentence:
E has p.
So like objects, they can be ascribed properties. “World war II was brutal”—this ascribes the property of being brutal to that event.
The thought is this. While events are what cause things, what explains why a given event causes what it does is some particular properties that it has.
So what caused the ending of the dinner property is an event. This event has various properties, including that it was an elephant stampede and that it was something that happened while I was in the kitchen. Now, since it really had both of these properties, either one can be used to pick out, to describe, the event in question. But only one of these properties explains, is responsible for, is relevant to the question of, its causing the end of the dinner party. It caused the end of the dinner party because it was an elephant stampede, not because it happened while I was in the kitchen.
Compare the case of the soprano….
So events are what cause, but it is events’ possession of particular properties, features, that causally explains.
Now, philosophers after Davidson have endorsed this picture, but have tended to loosen Davidson’s stricture that only events are causes. Ordinary speech allows objects to themselves be causes: e.g., I caused you to cry. The pothole has caused numerous crashes.
One might try to take those sentences as implicitly pointing to events within which those objects participate, but even so, it’s still a question why we can’t allow the objects themselves to be causes.
And in philosophy of mind, philosophers have gone further still. They have taken things that they call ‘states’ to be causes as well. A person’s propositional attitude is the quintessential example, on this view, of this category of mental state. It may well be that the way these philosophers employ the notion of state is confused, that it tries illicitly to straddle the boundary between object/event and property. I won’t get into these issues at all now. Though when we read McDowell we shall be in their vicinity.
But given this move, we seem compelled to draw the Davidsonian distinction in talk about propositional attitudes as well.
So consider the following:
F’s belief that the lecture would be boring caused him to leave.
(Here I’m setting aside the desire to simplify things.)
Here one might say this. Suppose beliefs and other ‘mental states’ are identical to neurological states of the brain, of certain configurations of neurons in the brain. Suppose in particular F’s belief that the lecture would be boring is identical to neurological state n of his brain. Then given Davidson’s point about the intersubstitutability of co-referring expressions in singular causal statements, we get:
F’s neurophysiological brain state n caused him to leave.
And now the question arises: which of these properties of F’s state—its being a belief that the lecture would be boring and its being an instance of neurophysiological state n causally explains why F left? [Write these on board]:
That F had a belief with the content that the lecture would be boring explains why he left.
That F was in brain state n explains why he left.
Given this distinction between singular causal statements and causal explanations, we see that we can’t simply infer from the truth of the latter the truth of the former.
But clearly for rational-psychological explanations to be true, it must indeed be the case not only that belief and other propositional attitudes cause behavior, but that their being beliefs and propositional attitudes with particular contents explain behavior.
“Because” is an ambiguous word on Davidson’s reading; sometimes it means, ignoring grammatical niceties, ‘is caused by’ and sometimes ‘is explained by’. But it’s clear that in the sentence above and in rational-psychological explanations in general, the “because” must be the “because” of causal explanation. We’ve just said that these are explanations, after all.
Now consider again my claim that such explanations are rationality-involving. The aim of rational-psychological explanation is to make sense of a person’s actions by showing them to make sense to her, to be in that sense rational. We have a rational-psychological explanation of an action to the extent that we see that the agent’s beliefs and desires rationalize the action. The property of a belief, desire or other propositional attitude that matters for its capacity to rationalize an action is, of course, not any neurological properties it might have, but its content: we understand an action as rational to the extent that we can see how, given what the agent believed or desired or feared or intended, the action might have seemed appropriate. Talk of what is believed is talk of content. Now, if we want to hold onto the idea that rationalizing explanations are also causal explanations, as nearly all philosophers nowadays want to do, we must suppose that beliefs and desires cause actions because of the very qualities that enable the beliefs and desires to rationalize the actions. Only so can there be a way of specifying a belief or desire that serves to pick out both what rationalizes and causes an action.
Clearly, then we are committed not only to beliefs being causes, but also to the explanatory relevance of their being the beliefs that they are. A hard-headed scientistic sort of philosopher might say: yes, beliefs do cause behavior. That’s because beliefs are, in point of fact, certain neurological (or cognitive-scientific) structures in the brain. And it’s their being the neurological structures they are that explain behavior. Their being the beliefs that they are is just irrelevant. But if we accept this view then, we’ve just seen, we have to given up rational-psychological explanation, we have to give up explaining what people do by seeing what would seem like a justifiable thing to them to do given their beliefs and aims.
Call the hardheaded view:
Content epiphenomenalism: No propositional-attitude state causes what it does because of content.
Rather, it causes what it does because of its neurological or computational or whatever features.
There are many reasons why philosophers have worried that this view might be true. Some don’t worry about it; they’re delighted that (it seems to them) true. These philosophers have psychological problems.
We won’t discuss any of those reasons except the one that Dretske does. What Dretske does is illustrate in a very clear and compelling way a special problem that externalists have in particular for avoiding content epiphenomenalism.
IV. Dretske on causation and externalism
.We’re not interested in Dretske’s own solution to the problem, outlined in the final section of his paper. We’re only interested in the problem he lays out, and then only as an expression of a problem that many philosophers perceive.
The problem is this. Causation, as the thought is sometimes put, is a local affair; an object’s causal powers must be determined by the way things stand with the object here and now. External properties, properties involving an item’s standing in a relationship to some other, possibly quite distant, object, are thus ‘screened off’ by the item’s intrinsic properties from doing any work in the item’s causal transactions. Sometimes we can predict an item’s behavior on the basis of its possession of certain extrinsic properties. But such a prediction will only be possible because possession of those extrinsic properties correlates with possession of certain intrinsic properties; it will be these intrinsic properties that are responsible for, and so explain, the relevant behavior.
Dretske develops the problem with his vending-machine analogy. If I know a coin has the value of $1, which is of course a relational and historical, hence external, property of that coin, I can predict that putting it into this vending machine will produce a soda. But since the mechanism of a vending machine is responsive to the shape of a coin, and not to its historical pedigree, it is the coin’s shape, not its value, that explains the result of putting it into the machine. Knowing the value of a coin is predictively useful in this case only because a coin’s value reliably correlates with the intrinsic properties that really matter for the workings of a vending machine. (As Dretske puts it, the value weakly supervenes on certain intrinsic properties of the coin. Read this as: correlates fairly well as things stand.) Similarly, to suppose with the externalists that content properties are extrinsic properties is to suppose that, while knowledge of content properties may enable us to predict behavior, content properties play no genuine role in the production, hence explanation, of behavior.
My putting a coin of size s and shape h in the vending machine caused its emitting of a coke.
We can equally say:
My putting a dollar coin in the vending machine caused its emitting a coke.
We can equally say this because, again, these are descriptions of the same event.
But now consider this pair:
That I put in a coin of size s and shape h explains why the vending machine emitted a coke.
That I put in a coin worth a dollar explains why the vending machine emitted a coke.
Dretske thinks that, although we might ordinarily feel comfortable saying something like the latter, we can be brought to see on reflection that its status is quite problematic.
What makes a coin worth a dollar is facts about where it was made plus certain broadly sociological facts about how it is used. Vending machines know nothing of that. Their mechanism is not sensitive historical or sociological properties, properties the coin’s possession of which are constitutively dependent on far-flung relations it stands in to items in the larger world. They’re sensitive only to what’s the deal here and now with the coin.
Dretske’s worry is that if we grant that a belief’s having the content that it does is itself an external property of the belief, something constitutively dependent on far-flung relationships to the physical and social world, we run into precisely the same difficulty. This property will be ‘screened off’ from doing any causal explanatory work by the intrinsic properties of the belief, presumably, by its being in a certain neurological or computational state.