Monday, November 25, 2013

A Case for Causal Dispositions

It seems to be something we almost take for granted. But it is notoriously difficult to come up with a theory that describes what it is, exactly. What is causation, exactly? This article was inspired by Causation: A Very Short Introduction.

David Hume

One popular analysis of causation is that it consists of an event, followed by another event. Since we can't see, or detect, the actual causal connection between the two events, and all we detect are the events themselves, then this is all causation boils down to. This theory was popularized by David Hume, who was an empiricist and thus believed all knowledge comes through sense experience.Through the regular succession of events, we form a habit in our minds that event A will be followed by event B, and this is all causation is.

Problems with Hume

The theory cannot deal with accidental or coincidental correlations. When we do a  randomized control trial on some new drug, and we see a percentage of patients improve, this event might be correlated with many different factors, such as a different diet eaten by the patients that improved. We seek to find out if the drug itself was the causal factor, and not just a correlation. As it is often said, "correlation is not causation."

In further addition, the Humean theory cannot deal with one-time causes. Perhaps there is some unique cause and effect that occurs only once in the history of the universe. But since Hume's theory requires regular correlation for something to be considered a cause/effect relationship, then single unique events cannot count as causation, even though after examining such a case we might conclude that it is in fact a cause/effect relationship.

Also, Hume's theory seems to presuppose some form of real causation, and not just correlation, in that it theorizes that we see an event followed by another event. In other words, the event caused us to see it. So this theory presupposes something more than just correlation.We need a better causal theory than this one.

Counterfactual Theories of Causation

Perhaps causation amounts to this: saying that A causes B means that if A had not occurred, B would not have occurred. Saying that the drunk driver caused the crash means that if the driver had not been on the road, the crash would not have occurred. This "counterfactual" theory of causation was popularized by David Lewis and is generally classified as "semi-Humean."

But it too faces problems.

Problems with Counterfactuals

One problem is that this account relies on a fiction: if A had not occurred then B would not have occurred, but how do we know what would have happened if A had not occurred? We can't empirically observe the event not occurring, because the event did occur. Lewis's theory relies on a fiction, so Lewis himself supports the theory that these other worlds actually exist. An infinite number of alternate worlds. Never mind the fact that this might be a bit much to believe, it also still faces problems. We can't observe these alternate worlds, so we still can't observe what would have happened had the drunk driver not been on the road that night.

And there are more problems. An effect could have multiple causes. Consider a firing squad, all shooting at the prisoner at the same time. Any single bullet would have been enough to kill the prisoner, but he was shot by multiple shooters at once. So if a shooter did not shoot, the effect (the prisoner's death) still would have occurred.

Also, we can have a case where if event A had not occurred, event B would not occur, but we still say that A did not cause B. Consider that if you had not woken up this morning, you would not have walked faster in your way to work. But your getting out of bed was not the cause of you walking faster. It was a necessary but insufficient condition for faster walking. So here again we have a counterexample to the causal theory at hand.

Is Causation the Transfer of Energy?

We often like to think of physical science as dealing with fundamental reality, so perhaps causation is some basic event from physical science. Energy transfer seems to be the basic events that occur in physics, so maybe causation is nothing more than the transfer of energy.

Problems with Energy Transfer

But again we have counterexamples: if we inject money into an economy, it causes inflation. This is not energy transfer, as money these days is often in terms of just numbers in a computer. So not all causation can be described as the transfer of energy.

Is Causation Primitive?

Perhaps the problem is that causation is a basic feature of the universe, and cannot be broken down into simpler concepts. Just like right now we think quarks are the basic particle of matter, and cannot be explained in terms of anything simpler. They are the bottom level, so to speak. Perhaps causation is the same. Remember Hume, above? His theory seemed to presuppose causation at a very basic level, as it relied on seeing or sensing an event, followed by seeing or sensing another event. So perhaps this is why none of the above theories can capture what we mean by causation: it is just a basic feature of the universe, unable to be broken down anymore.

Causal Dispositions

But even if this is the case, we can still examine causation and describe it. The idea of dispositions is this: an object has certain hidden properties or powers that dispose it to behave a certain way. For example, an object that is elastic is disposed to "snap back", even if this effect never actually occurs. It is disposed to behave one way rather than another way. An object that is "elastic" is disposed to snap back, but not disposed to catch fire. An object that is "flammable" is disposed to catch fire, but not disposed to snap back. And so on. This seems to capture what we mean by causation better than any of the theories above.

Historically, empiricists have not embraced the dispositional theory of causation, because "dispositions" are ephemeral. They cannot be measured, observed, or quantified. You can observe and measure a property that an object actually has, such as mass, or charge. You can put numbers on them. But you can't do the same with dispositions because they don't even really exist yet, in a way. They are a propensity to behave a certain way, which is not a property that can be observed.

Yet, if this analysis of causation is plausible, then it is still correct, despite our inability to observe them.

No comments:

Post a Comment