# Davidson, events, and states

Davidson argues that events are individuals. According to him, the meaning of Brutus stabbed Caesar is ∃e.stab(e,Brutus,Caesar), that is, there was a stabbing, Brutus did it, and Caesar underwent it. Likewise, Caesar died means ∃e.die(e, Caesar). Is there a reason not to view states as individuals, too? Something like ∃s.dead(s, Caesar) for Caesar is dead? One could then say, for example, that

die(e, Caesar) ≡ ∃s1,s2.change(e, s1, s2) ∧ alive(s1, Caesar) ∧ dead(s2, Caesar)

just as we can say

kill(e, Brutus, Caesar) ≡ ∃e′.cause(e, Brutus, e′) ∧ die(e′, Caesar)

Coordination can also be construed as an individual

||Brutus stabbed Caesar and drank a beer|| = ∃e,e1,e2.and(e,e1,e2) ∧ ...

which gives us the possibility to have complex events (I flew to Paris is an event but it can be refined as going to the airport, entering the plane, etc.). Do I miss something that prevents us from wanting to reify states?

## 2 Answers

Well there is a typical consideration , why to not reify concepts (like states). This is the age old Occam's Razor (http://en.wikipedia.org/wiki/Occam%27s_razor):

Entities must not be multiplied beyond necessity

In other words, we'd better not reify states, if we don't have to. Syntactically, we can still treat states as pseudo-entities using transformation rules, such as:

alive(x) <-> (Es)alive(s,x)

We seem to need reified events, more than we need reified states. Davidson's main reason for that, if I recall correctly, is that events are needed as the terms of causal relations. There seems to be no similar motive for reified states, because states are not ingredients of causal relations (at least in Davidson's account of causality ).

Your hair-splitting of events is no different from hair-splitting of other entities as well.

So you'd have:

go(e, Brutus, Paris) ≡ go(e, leg(Brutus,left), Paris) ∧ go(e, leg(Brutus,right), Paris) ∧ go(e, arm(Brutus,left), Paris) ∧ go(e, arm(Brutus,right), Paris) ∧ go(e, torso(Brutus), Paris) ...

You can split `Paris` in a similar manner too.

Davidsonian event semantics provides a formal way of notating the things you want to express. If you have good reason to split one event into several sub-events, do it. If you have good reason to split one entity into several sub-entities, do it.

States are just attributes of entities, and can be notated as ordinary predicates. The following statements refer to the same `Soup`, but in one event, it is a member of all `hot` things, and in another event, it is a member of all `cold` things.

∃e1.hot(e1, Soup)

∃e2.cold(e2, Soup)