The axiom of perfections pre-existing in agent causes

Someone googling around about agent causation (in order to make an objection at Randall Rouser’s site) found me saying that:

[A]ny perfection possessed by an effect must be possessed in equal or greater degree by its agent cause.

He responded:

But this is not true. IBM’s Deep Blue computer can play chess more perfectly than any of its human creators. Aquinas can be forgiven for not knowing that, but we know it now.

The causal axiom in question needs to be explained in four stages. First, it will seem almost universally false, second, it will seem vacuous and tautological, third it will seem false again, and last it will show itself as a useful insight about causality.

1.) It seems universally false. Why stop at Deep Blue – why even take such a complicated instrument? True, the programmers of Deep Blue couldn’t beat Kasperov, but a person using a screwdriver also couldn’t drive in a screw more perfectly just using his hands; a person driving a car couldn’t move that fast without it; a surgeon couldn’t do the surgery just as elegantly with his bare hands, etc. Isn’t the whole point of making tools to accomplish some task we couldn’t accomplish without them? Shouldn’t we make the axiom that an agent cause must lack any perfection it brings about?  In this sense, it seems not just that the axiom is false, but that its contrary is axiomatic!

But this is not how the axiom was meant to be taken. By “equal or greater degree” what is meant is that the agent has the thing it bestows on the effect either (a.) in the same way the effect has it, or (b.) in a virtual way. Examples of (a.) are when the sun causes something else to heat up, or when the fridge cools down food. Examples of (b.) are a bit more subtle – a knife causes something to be cut not because it is cut, but because it has something that is ordered to cutting (sc. its honed edge). This leads us to…

2.) The claim seems vacuous and tautological. If this is all one means, then all the axiom comes to is that an agent either is the very thing it causes, or it is merely ordered to causing it. Fire has a power to heat and is itself hot, matches have the power to heat even if they are not hot. Matches are, on this account “virtually hot”, knives have cuts not by being cut, but by having a power to cause them.

3.) But then the claim seems false again. What sense is there to describing a knife as “virtual steak” because it can butcher a cow? Why call malaria parasites “virtual fever”? Is there any sense to this at all?

Perhaps there is. Consider the malaria parasites again. True, fever results from them, but is it the parasite that does it? Is causing a fever an essential part of its life cycle, or is it rather an accident of its development in the host? These are interesting and perhaps even helpful things to know. Say that the malaria parasite causes a fever as an essential part of its life cycle. In order to do this, it would need some chemical trigger. If you discovered such a trigger in research, you actually would  name it after its effect. In fact, such a thing has already been named after its effect – it is a pyretic, i.e. “fever causing” or “a virtual fever”. My guess, however, is that the parasite only has fever as a side effect – though a necessary one – of its life cycle in a host. If this is the case, then it is not the parasite as such that causes the fever, even if its action is necessarily correlated with it.  If you were doing research on malarial fever, this is exactly the sort of problem you would need to solve. But the light you would be following to solve the problem would be the idea that finding the cause of the fever means locating what exactly is pyretic – is it something in the virus as such, or something else (viz. an immune system response, an overabundance of choleric humors, whatever). But this leads us to seeing…

4.) How the causal axiom is an illuminating, subtle, and yet a priori truth about causation. We tend to know effects before causes, and identifying causes in the fullest sense means locating precisely what contains the effect virtually. Sometimes we settle for what is less than a cause, sometimes we have no name for precisely what formality the cause corresponds to, but any systematic inquiry into agent causes is only complete when it locates some locus where the effect we wanted to understand was virtually, causally contained.