If you’re following along with the discussion of causality via the commentary on Judea Pearl’s text that we’ve begun here at TMD, the good news is that we’re still at it…the bad news is that we’re still debating (check out the comments section on Part 1) the metaphysical implications of this view I’m attributing to Pearl in order to make my own (seemingly crazy) view seem more mainstream. So much for progress…
The sticking points thus far seems to be two: whether and in what ways this minimalist view of mine is anything different from Hume and just how serious is this minimalist view anti-metaphysical. I’ll let Aristotle demonstrate my position:
For my money, philosophical questions are best when they’re epistemological (when they’re about our beliefs of the world and the logical and evidential relations therein etc.) and worst when they’re metaphysical (when they’re about the “fabric of the cosmos” and the way things “are” etc.). This puts me in an interesting position since I am fascinated by causality, a topic that has been the purview of metaphysicians since there were metaphysicians and philosophers falling down wells and what-not. I fully accept that this puts the burden fully on me and that’s fine. Here’s what Pearl says:
But causality is not mystical or metaphysical. It can be understood in terms of simple processes, and it can be expressed in a friendly mathematical language, ready for computer analysis(p.427).
Now, to be fair, Pearl is a brilliant computer scientist and mathematician, but not a brilliant philosopher by trade, and sometimes philosophers tend to hold others to a very strict manner of speaking even when it is ill-advised to do so. Maybe he is making a grand claim about the metaphysical status of causality, maybe he’s just being a flamboyant writer. Who knows. The rest of the book (that we will start focusing upon this week…) is devoted to the mathematical analysis of causal claims and providing a set of tools for professional researchers to formulate and test causal hypotheses; it is not a book about the metaphysics of causation. But, and here’s the sticking point, it needs to be a book about the metaphysics of causation even if it’s not. Simple as that. So here’s what I propose.
I think that we should start with the simple and provocative position I described in Part 1 and see whether Pearl’s mathematics can do everything we want out of a formal model of causation. Then we should see whether it can stand up to philosophical attention…serious philosophical attention concerning those tricky metaphysical issues I’m hoping we feel safe to abandon once we’re finished with the book.
So here’s what I propose: we start out with the assumption that causation is an essentially pragmatic notion, that causal hypotheses are the product of isolating a system, describing that system as a composite of discrete entities and relations, and examining the inter-relations between members of that system in an artificial way (which is not a dig against the scientific method by any means; it’s just the way research has to be done). The challenge is to see how far we get using that minimalist concept.
I’ll leave you with a paraphrase of van Fraassen’s concerning causal explanations. Imagine some committee is charged with adjudicating responsibility of some particularly bad car accident. On the committee is an engineer, a physicist, a civil planner, an auto mechanic, a psychologist, and a cognitive scientist. Now asking for who or what is responsible for a car accident is clearly a request for a causal explanation (“what or who caused this accident?” or “Who or what is at fault here?” etc.). van Fraassen asks the reader to imagine what kinds of explanations we’re going to receive from the respective committee members. The civil planner will focus attention on the placement of traffic lights, their timing periods, the placement and state-of-maintanance of the bushes and trees, etc. of the location where the accident occurred. A physicist will give an explanation involving static and kinetic friction, applied force, braking distance (a function of acceleration, initial speed, and friction) etc. A mechanic will give an explanation involving the operation of the braking system, the state of the tires, the coolant system (I guess, I’m a bad mechanic…). The cognitive scientist a will discuss reaction times of the human brain in relation to the speed of travel, and attention. The psychologist may talk about the mental state of the driver at the time of the accident (maybe he was worried about work, or was cheating on his wife and was thinking about how to resolve that). The point is two-fold: first, not all or even most of these explanations can be reduced to one another (things just don’t reduce down to physics like most used to believe); second, there is no clear sense in which any one of these can be more “right” or correct than another. How do we quantify the relative significance of being preoccupied while behind the wheel with the state of the bushes partially occluding the other traffic lane etc.? For van Fraassen, there are as many legitimate causal explanations as there are people to give them. This is the pragmatic conception of causal explanations (it need not reduce down to my causal minimalism, but I like to think it does).
So what do we make of this? If causation were some metaphysical structure to the world, then there should be “a cause” of that car accident? But what does it mean if there isn’t “a cause”?