Saturday, January 15, 2011

On Climatic Explanations of Ideas and Actions

Benjamin Nelson has an interesting post criticizing my post On Sobriety in Times of Stress. It's worth looking at a bit more closely because it raises important issues of the way in which appeal to a 'climate' or an 'environment' or a 'culture' can function in historical explanations.

Before I get into the meat of the issue, though, I do want to address one issue. One of the pieces of evidence that plays an important secondary role in the discussion is the question of Loughner's connection to American Renaissance; and at one point, after quoting me, he says, "this was written before we found out that Loughner is associated with American Renaissance, so it’s not fair to criticize Brandon for not making that connection." In actual fact, though, the post was written after the rumors went around that the DHS had linked Loughner to American Renaissance (a rumor that went around on January 9) and a few hours after the DHS the next day denied the rumor, thus denying the only original reason for holding that there was any connection at all. It's still conceivable, of course, that evidence might turn up vindicating the original rumor, but as far as I am aware none has.

But Benjamin is quite right that the actual status of the connection is not the really important matter here, so let's suppose that such a connection did exist, and were clearly known to do so. Given this, would it then be a reasonable foundation to appeal to a climate of violence as an explanation in the case? I think the answer is that it could only do so if there were additional information; on its own it is not the right kind of fact to support such an appeal.

To see this we need to step back and think about what it means to appeal at all to a climate or environment as a causal factor. A climate or environment is obviously only causal at all if we think of it as a name describing an entire cloud of causal factors. For instance, if I am trying to explain why, in a war situation, nation A overreacted to intelligence about nation B, I might appeal to a climate of paranoia about B. What I clearly mean here is that, among the many very different causal factors affecting the reactions of A, there were lots of causal factors that increased the likelihood of people in nation A having motivations with regard to B that can be classified as paranoid, and that people with these motivations were actually involved in the overreaction.

Now, how does one provide evidence for such a claim. It seems clear enough, if the description just given is at least in the ballpark that you would need evidence for all of the following:

(1) that there are causal factors increasing the likelihood of of people in A being paranoid about B
(2) that there are lots of such factors
(3) that the circumstances in which such factors operated was such that the actual reaction could be due to them

If new evidence turned up, for instance, that showed that all the supposed causal factors tending to paranoia about B were actually just made up by clever intelligence officers from B, there really is no climate to appeal to. Likewise, if it turned out that the so-called 'climate of paranoia' was really just one or two paranoid people, and that there wasn't much paranoia elsewhere, then, while paranoia about B might still be part of the causal explanation for the overreaction, it's simply a mistake to characterize such explanation as involving a 'climate of paranoia'.

Both (1) and (2) are conditions for there being a climate to appeal to at all. But in appealing to anything for an explanation, one must identify it as relevant, and this is where (3) comes in, and (3) is far and away the most important one for evaluation of explanations, because it's what actually makes something as vaguely identified as a 'climate' still useful for explanation.

Climatic explanations are causal explanations in which the cause is recognized to be manifold (many causal factors) but in which the precise causal actions are not being identified (perhaps because we don't know them with sufficient precision). This vague plurality of causes does make it differ to some extent from other causal explanations, but one thing is clear enough: there is the same need for there to be reason to link this complex group of causal factors with the phenomenon to be explained. We have no handy word for this in the context of climatic explanations, so I'll call it canalization of climate: there needs to be some channel connecting the climate to the actual thing to be explained. I think a good way of seeing how this works is to look at an argument and example Benjamin gives:

A culture is a feature of populations, not just particular interacting persons. You don’t need to know the details about how a society connects specific people with other specific people in order to understand how the culture has had a predictable influence. You just need to establish that the person plays some role in the culture, and that the culture has certain features. By analogy, we will sometimes explain a case of the flu by saying, “there’s a flu going around” — we don’t bother going through the effort of naming the exact person who gave you the virus.

So let's take the case of "there's a flu going around" as an explanation for why someone, let's say Leroy, gets the flu. It's true that this can be explanatory without our having to identify the exact source of the flu. But we still need evidence for canalization. Let's start with the most basic evidence required: as Benjamin says, we need to establish that the person plays some role. If, for instance, there's a flu going around New York and Leroy is in Fishtail, Montana, obviously we have no particular reason to think that the one is an explanation for the other. If, on the other hand, there's a flu going around Fishtail, that is much better. But it's important to see that this is still not enough. In explaining Leroy's flu by a climate of flu, we're not just assuming he's in an area where there is flu; we're assuming that Leroy has been in this area in such a way as to be able to catch it. If Leroy is a hermit who doesn't see people for weeks at a time, during which he catches the flu, saying "there's a flu going around" does not alleviate one whit the mystery of how Leroy got the flu.

All this is with something like the flu, which is, for the purposes of the explanation, a passive state. It is something that happens to you. What is more, it is something contagious: we have excellent reason to think that exposure to the flu, on its own, increases your chances of actually catching the flu. There are aspects of human life that are usefully explained as contagious passive states: biases, for instance. Very simple ideas can also often be explained in such a way. In such a circumstance, when you've established an actual climate favorable to a bias or idea, not 'catching' the bias or idea is what usually needs explaining. If you find that someone lived in an anti-Catholic climate but that they themselves were very pro-Catholic, this is a fact in need of explanation; whereas the mere fact of being anti-Catholic while constantly exposed to an anti-Catholic environment does not -- the environment itself, or rather the exposure to it, explains the fact.

When we trying to explain actions, things are somewhat more difficult, and more precise evidence is required to establish canalization. The reason for this is that action as an explanandum is a much more particular and precise explanandum than something like a bias is. Biases are consistent with a wide range of actual actions: someone who is prejudiced against blacks could be violently prejudiced, or they could avoid them, or they could try to overcome the prejudice, etc., etc. If you're merely trying to explain why someone is prejudiced against blacks, there's very little need for precision. Things change if you are trying to explain why someone goes out of his way to go into primarily black neighborhoods to rock windows; that's much more precise, and the explanation that was adequate for explaining the mere fact of prejudice against blacks is not adequate for explaining such active racist behavior.

Now let's return to the original case and see how climatic explanation works in such a case. We are explaining something that is not a contagious passive state but an action, and because of the sketchiness of the evidence, there are actually quite a few possible explanations still on the table. We could have a climatic explanation; we could explain it as a personal vendetta; we could explain it as having a very particular political cause; and so forth. Such explanations need not always be mutually exclusive, but even if they aren't, if they aren't ruled out, then either the climatic explanation is a rival explanation or the climatic explanation assimilates them; and if the latter, we need to have actual evidence linking them to the canalization of the climate, i.e., actual evidence that they are part of the channel whereby the climate affected the shooter in particular so as to at least explain some features of his actions. In the Loughner case, the weakness of the argument for the superiority of the climatic explanation was part of the target in my post; the evidence put forward to establish canalization was weak at best, and the evidence at the time was mounting that it was something else, namely, an act by someone having gone through a long period of stress while in a deteriorating mental state, and who also held a longstanding personal, albeit irrational, grudge against the victim; and the work of assimilating this evidence to the climatic explanation was simply not being done by those who still preferred the climatic explanation.

But suppose we did have a clear connection to right-wing groups? Would this be adequate to establish canalization? In combination with other things, yes, but on its own, no. We would have to know details about the right-wing group and Loughner's relation to it. We'd obviously need to ask whether there was actually a 'culture of violence' in this right-wing group. Suppose there was a lot of violent rhetoric and preparation for the revolution to come. We'd still need to know that Loughner's participation in this culture was of the right kind to serve an explanatory role. If, for instance, we had evidence that Loughner was drawn to the group because he already had violent tendencies and already fully agreed with their message, appeal to the culture of violence in that group would contribute nothing of significance to the explanation. We wouldn't need to know every detail, but in order to assess the quality of the climatic explanation at all we'd still need to know enough details to establish that both the group and Loughner's participation in it were of the right sort to make canalization possible -- i.e., to make the climate an actually discernible contributing cause to Loughner's behavior. The participation in the right-wing group would be a solid foundation for inquiry into whether it was the cause, because we know that extremist groups can be causal factors in behavior. It would be reasonable to ask the question. But what makes for a good foundation for a question can be very different from what makes for a good foundation for an explanation.

There are other things we'd have to consider, though. Since extremist right-wing groups, considered generally, tend to exist regardless of the general climate, even if we identified the culture of violence in an extremist right-wing group as a real contributing cause, we wouldn't be able to draw the conclusion that a more general culture of violence was responsible -- we'd have established canalization, but only for the local climate of the right-wing group. In order to go farther we'd need to do one of two things: (1) find evidence for another, distinct canalization that showed that the general climate played a separate, direct contributing role; or (2) find evidence supporting the claim that the local climate of violence was itself canalized from the general climate of violence, thus establishing that the general climate played an indirect contributing role. Nobody attributes a cross-burning by the Klan to a general culture of violence, even if they recognize that there is such a general culture of violence, for the reason that everyone recognizes that the local culture of local chapters of the Klan is entirely capable of being violent independently of the general culture; if somebody wanted to hold that in a particular case the general culture played a role, they'd have to argue either that in this particular case there were additional factors clearly at work, or else that this particular local culture of the Klan originated, or was revived by, the general culture. To establish that the culture of the "Palin vanguard" had a role, whether direct or indirect, we need to establish that there actually was a channel of the right kind from the Palin vanguard. If we don't have that, we haven't established that the right culture is involved.

Thus climatic explanations end up being much more complicated than I think Benjamin suggests. It's not just a matter of establishing that there was a culture and that the person played a role in it; one must also establish that the circumstances of the culture, and the role played, are of the right sort to make them explanatory rather than incidental. And since everyone is surrounded by lots of cultures and climates and environments all the time, attributing a causal role to any one of them requires some very specific evidence; we need to show that it's at least plausible to say that the climate channeled into the behavior somehow.

In the end, I tend to approach these questions as someone who does history of philosophy; one must occasionally resort to climatic explanations in history of philosophy as in any other history. Climatic explanations can always be given greater specificity, and one hopes to, but sometimes we're stuck with "it seems to have been in the air" or "the idea was going around" or "it was a key part of the culture of the time", and sometimes we make use of such explanations for legitimate practical purposes even when we can specify further. But it's also a very dangerous sort of explanation. Because it leaves the key explanatory factors as vague plurality, we have to be especially careful about what I've called canalization here: if you play it too loosely then it's very easy to pass off non-explanations as if they were explanatory, or even to think that something explains something when there is a much better explanation available that requires no appeal to climate or culture or environment at large. And this is true with regard to all sorts of areas outside of any historical discipline. If, for instance, I wanted to attribute some phenomenon to the slothfulness of American culture, I need to establish that there is actually a 'channel' of the right sort between the slothfulness and the phenomenon, that some of the things making American culture slothful are actually operative in bringing about the phenomenon. And, likewise, if we want to argue that a particular violent act is due to a larger culture of violence, we must have definite evidence establishing a channel there, too. Otherwise we aren't drawing reasonable conclusions but speculating wildly.