Kim's aim in this article is
(1) to give an account of the relationship between macro- and micro-events/properties
(2) to explain causation on the macro-level
(3) to model mental causation on macro-level causation
(4) to show that epiphenomenalism is not a serious problem

Epiphenomenal causation

a sort of apparent causal relation, e.g, a mirror reflection of an object at t1 does not cause the mirror reflection of that object at t2. Both images have the same cause and so are causally related in a certain sense, but one does not cause the other, even if it might appear that way. Eg, a reflection of two billiard balls.

another example: the succession of symptoms of a disease, appear to cause one another, but really something else which underlies the symptoms and is doing the causing.

Why call it "causation" at all if only appearance?
- Because they are in some sort of causal relationship with one another, even if that is not the cause-effect relationship. Two successive events which appear to cause one another are themselves effects of the same underlying causal process. So they have some important causal relationship with one another, even if it is not cause-effect.
- Because epiphenomenal causation is pervasive - most of the causal relations we see are actually of this kind, e.g., all macro-causation.

Kim wants to argue that
(1) all macrocausation should be considered to be epiphenomenal causation
(2) macrocausation is a specific type of epiphenomenal causation, namely supervenient causation.
(3) mental causation is a type of macrocausation and so a type of supervenient causation.

So what is macrocausation?

macro-micro is relative: chairs and tables are macro and atomic level is micro, but relative to the particle/sub-atomic level, the atomic level is macro.
All medium sized objects, anything we can see with the naked eye, are macro objects. Two billiard balls hitting one another would be an example of macrocausation. Any causal relation between objects we can observe, then, is macrocausation, and so a type of epiphenomenal causation.

Two reasons for holding this view

if we are physicalists, then we want to avoid any non-physical properties that are irreducibly so. therefore, irreducibily mental properties are a problem for physicalism, because "the world remains bifurcated" - it's a type of dualism. If the mental is irreducibly mental, then physical theory would lose its closed character - some causes of events or some effects of causes would be non-physical.

same goes for macrocausation - it is an assumption of physics that what goes on on the macrolevel is determined by what's going on on the microlevel. Water is liquid because of certain features of water molecules. Macro properties, like being liquid, or solid, or soluble... are supervenient or dependent on the micro properties.

Mereological supervenience = microdeterminism (top-down, bottom-up)

- the properties of the whole supervene on the properties of the parts
- the micro level determines the macro level.

If you believe this, as you should if you believe in physics, then you can not allow for the possibility of irreducibly macro causal relations, relations that do not depend in some way on the micro causal relations.

If two macro events, water being liquid and sugar being soluble, have a causal relation, and there is a macro law, like if you put something soluble in something liquid, then that thing will dissolve, this law ought to be determined by micro laws, or laws that governed events at the microlevel. Lawlike connections between macro properties, like solubility and liquidity,  should be dependent on laws connecting micro properties. Otherwise we would have to believe that water is irreducibly liquid and sugar is irreducibly soluble. If we believe that the whole is determined by the parts (mereological supervenicne + microdeterminism), then we cannot believe in irreducibly macro properties.

The kind of supervenience Kim has in mind is Strong Supervenice, or supervenience which supports counterfactuals (has modal force).

The macro property F supervenes on the micro property m(F) or the microproerty m*(F) and so on.

Example: macro property of fragility, in one object may have certain micro properties that make it fragile, in another object may have different micro properties that make it fragile.

In this model, F and G are in a mereological supervenient causal relation - the causal relation is supervenient upon an underlying causal process because the macro properties supervene on the micro properties.

Two claims:

(1) macrocausal relations should be viewed as in general reducible to microcausal relations
(2) the mechanism of the reduction involves identifying the microstates on which the macrostates in question depend, or with which they are correlated, and showing that a proper causal relation obtains for these microstates

If there are causal relations between macroevents that are not microdetermined, then these relations are "causal danglers" and are "an accute embarrassment to the physicalist view of the world."

Using this model for mental causation

What makes epiphenomenal causation "real"? There's nothing "unreal" about water causing sugar to dissolve, whereas there is no real causal connection between the symptoms of a disease. Why?

Mereological supervenience. Macro-micro - relation of supervenience, so that makes macrocausation perfectly "real". When there is no such supervenience, as in the case of symptoms, then there is no causation going on:

"some epiphenomenal causal relations are supervenient causal relations, and these are among the ones that are "real"; there are also cases of epiphenomcanal causation that do not invlove direct causal connections, and these include ones in which the events involved are succesive causal effects of some underlying process."

Mental causation is a type of supervenient causation, i.e, it does not have this stucture (symptoms structure).

Mental (macro) properties are muliply realized by physical (micro) properties. Whenever a mental event causes another event, this is because of the relation betwen the micro properties of those events, but that doesn't make it any less "real" - it is just a real as the two billiard balls. Mental properties are emergent properties, like liquidity or solubility. (Kim doesn't like this way of talking, though, because mental properties, like liquidity or solubility, can be reduced to other properties. "Emergentism" as Kim calls it, is a non-reductivist view. Emergent properties are irreducibly so.)

Davidson cannot help himself to this account because he does not think there is a nomological relation between mental properties and the physical properties that realize them - he believes in Weak Supervenience. Strong Supervenience is needed to make the mental properties causally relevant.

When a mental event m causes a physical event p, this is because the mental event is supervenient upon a physical event p*, and it is p* that causes p.
When a mental event causes a physical event, it has the structure of model above.

"Epiphenomenal causal relations involving psychological events, therefore, are no less real or substantial than those involving macrophysical events. They are both supervenient causal relations."