Block wants to claim that one way of approaching this problem is by understanding functionalism and how causation works in scientific laws. One version of functionalism (see also the glossary) is the view that mental states or events are to be defined in terms of their connections with inputs, outputs, and other mental states. These connections "mirror" the physical and causal connections among sensory nerve stimulations, neural firing patterns, and motor behavior. But what we also need are for these functional interactions of the mental to be important in explaining behavior, for lots of things mirror or hitchhike other things, and most of them aren't relevant to the explanatory task at hand. So how do we know that the mental properties are required in understanding behavior? A popular answer is that mental properties are found in scientific laws and that is all the legitimation they need.

But that isn't enough yet. First, there are scientific laws that report mere correlations -- and not causes (particularly in the "softer sciences") -- so just being in a law doesn't guarantee you causal relevance. Second, the laws that do capture causal correlations sometimes articulate causal chains that we consider extraneous to the matter at hand (particularly when one effect co-occurs with another), so again being in a laws doesn't guarantee causal relevance. So, to take the bomb example: Block tells a story about a bomb that is triggered by an increase in thermal energy, which co-occurs with an increase in electrical activity. We want to say the thermal energy caused the bomb to explode, even though there is a law-like connection between the electrical activity and the explosion too. How do we know that mental states are analogous to electrical activity in the bomb story? Mental states, events, and properties need to be in the right sort of laws. How do we determine those?

The Macdonalds want to answer this question by leaning on the counterfactual aspect of nomological relations. They point out first that scientific laws are about types of events, classes of similar things (see also the glossary). However, most of the examples that Block discusses concern only token instances of something, a single exemplar of some type (see also the glossary). But if you want to know whether something is genuinely causally relevant, then you have to go beyond the individual instance and see how the case generalizes. If events relevantly similar to the one in question show the same pattern of correlations, then we can infer a genuine causal (and causally important) connection. Returning to the bomb example, they claim that if we generalize over the various instances of bombs, we shall see that bombs are triggered by heat and not by electricity. Hence, we will know that the thermal properties are the relevant ones. Similarly, if we generalize over our behavior, we shall see that what all the triggers for a certain type of behavior have in common is a mental property (or event) or set of properties (or events).

There are two difficulties with this sort of response, one of which Block gives. He notes that we don't really have any scientific laws that concern bombs, or anything else that specific. And our general physical laws wouldn't distinguish between the different possible causings of an explosion. If we created more specific laws, his so-called "apparatus" laws, then we have no reason not to generalize over "thermal bombs" versus "electrical bombs" versus "bombs in general." As long as we are going to be able to get as specific or as general as we want (or as suits our purposes), then we will be able to find any sort of abstraction we desire, some that make mental doodads causally relevant and some that make the same doodads irrelevant.

The second difficulty goes back to Nelson Goodman. He holds that how we "project our predicates" is completely undetermined. The Macdonalds are pretending that how things are generalized and abstracted is already known and accepted, when it in fact isn't. We could generalize over the notion of what a bomb is any way we want; nothing in the world determines how things are supposed to go counterfactually (which is what makes them counterfactuals), so how we devise the counterfactuals is up to us. Whether our generalizations about behavior should include the mind isn't determined by anything in the world.

Block tries a second tact in answering the problem using functionalism. He notes that (if the mind is physical and if it is understood functionally, then) there is no way for some mental state to have a different content (be a different mental state, in other words) and still cause the same behavior. Why? Because the mental state, and its content, are defined in terms of the behavior it is related to. Different outputs = different mental state. Very simple. It has to be relevant because if it were different, then we would get a different output.

But this doesn't answer the epiphenomenal worry (see the glossary for a definition of epiphenomenal). Block views functional properties as second order. That is, they are properties of having other properties. A functional definition of a mental state would be a statement concerning the sorts of interactions this mental state could have, which piggybacks on the physical interactions of the underlying nervous system. In essence then a functional description of the mind is a (covert) conjunction of the sorts of neural or whatever interactions that could underpin the mind. But then what is really causally important are the underpinnings and not the general second order description. That is just a useful shorthand.

But note that this view of functional descriptions and second order properties does away with all or almost all causes. We wouldn't want to say that neural firing patterns caused the behavior either, since that those are really shorthand for describing the chemical interactions that comprise the neural patterns. But causality doesn't bottom out in chemistry either, since that is just shorthand for the quantum interactions that underlie the chemical. But causality isn't at the quantum level either, since there all we get are functional descriptions (just lists of elegant correlations really) with nothing doing the instantiation. So causes don't exist!

I personally find this a bit hinky, though I don't believe that Block does. I prefer to say that causes are at all levels and the real question is determining which are the relevant ones to use in an explanation. I buy the point that it all depends on counterfactual understandings, which is completely underdetermined, but that doesn't bother me (though it makes Block hinky). I think of science as being largely pragmatic anyway. All this is spelled out in a paper of mine, which you can read for emjoyment (if you have an odd sense of fun) or to get a handle on how I think about these matters in greater detail before you come to class.

Valerie