Events and Processes
Something happened: a window broke, an election was won, a war was fought. Are these entities — items in the inventory alongside the window and the voters — or is talk of happenings a way of talking about objects and the properties they have at times?
The question is not idle bookkeeping. Causation relates causes to effects, and if the relata are events, then admitting causation admits events. Explanation, perception, action, and memory all appear to involve them.
Davidson's argument
The most influential argument for events is linguistic in form and ontological in conclusion. Consider:
Jones buttered the toast slowly, deliberately, in the bathroom, with a knife, at midnight.
From this any subset of the modifiers can be dropped and the result still follows. "Jones buttered the toast with a knife" follows; so does "Jones buttered the toast." That is a valid inference pattern, and a semantics must explain it.
Treating "buttered" as a five-place predicate does not: from a five-place relation nothing follows about a three-place one, and we would need a separate predicate for every combination of modifiers, with the entailments stipulated one by one. Davidson's proposal is that action sentences contain a hidden quantifier over events:
Now the entailments are simply conjunction elimination. The logical form explains the inferences, and by the criterion of ontological commitment the quantifier commits us to events.
Two further considerations support the same conclusion. Singular causal statements — "the short circuit caused the fire" — appear to relate two particulars, and the particulars are events. And anaphora: "the window broke; it happened at noon and was heard next door" picks up a thing to say more about, and that thing is not the window.
This is the section's cleanest example of an ontological argument built from logical form, and it is worth noting how much rests on the claim that no alternative semantics does the job. Adverbial approaches that treat modifiers as predicate operators have been developed precisely to resist it.
What is an event?
Granting events, their nature and individuation are disputed, and the accounts differ chiefly over how fine-grained they are.
Kim's property-exemplification account. An event is a structured entity — an object exemplifying a property at a time — with identity conditions to match: two events are identical only if they have the same constitutive object, property, and time. This makes events very fine-grained. Brutus's stabbing Caesar and Brutus's killing Caesar involve different properties and so are different events, even though there was only one plunge of the knife.
Davidson's own view. Events are unstructured particulars, individuated originally by their causes and effects (an account he later abandoned as circular) and subsequently by spatiotemporal region: events are identical if they occupy the same region. This is very coarse-grained. The stabbing and the killing occupy the same region, so they are one event under two descriptions.
Lewis's account. An event is a property of a spacetime region — equivalently, a class of regions across possible worlds. This allows intermediate grain, since properties can be more or less specific, and integrates events with his modal framework.
The disagreement matters because of causation. If the stabbing and the killing are one event, then citing either gives the same cause; if they are two, we must say which caused Caesar's death, and the answer that the killing caused the death is uncomfortably close to saying the death caused itself. Conversely, coarse individuation makes trouble for explanation: the flipping of the switch and the illuminating of the room may occupy the same region, but "he turned on the light because he flipped the switch" is not trivial.
Processes
A separate category, or possibly a rival one. Processes are ongoing rather than completed: the running, not the run; the melting, not the having melted.
The linguistic marker is the mass/count distinction applied to occurrences. Processes are atelic and homogeneous — any part of a running is itself a running, as any part of water is water — whereas events are telic and have natural completions, so that a part of a crossing of the road is not itself a crossing. This is the same structure as the distinction between stuff and things, transposed to time.
Process ontology (Whitehead, Rescher) takes the more radical step of making processes basic and treating apparently enduring objects as stable patterns in them. A river is the obvious case, an organism a better one, since it exchanges its matter continuously while maintaining organisation.
Two areas make this attractive. In biology, organisms are far better described as persisting processes than as objects that happen to change: metabolic turnover means the matter is not the same, and the identity is carried by organisation. In quantum field theory, the fundamental description is of field excitations, interactions, and scattering, and the "particles" are not stable localisable objects — which is a point developed in the section's treatment of the sciences.
The objection to process ontology is the one facing every attempt to make the derivative basic: it must explain what makes a process one process, and the natural answer refers to the object undergoing it.
Agents and events
A challenge from a different direction. If causation relates events, then every cause is an occurrence, and an agent's causing something must be some event involving the agent causing it. Agent causation denies this: on the libertarian view developed under libertarianism, an agent — a substance, not an event — can be an irreducible cause, not analysable as a causing by any event.
If agent causation is coherent, the inventory of causal relata includes substances as well as events, and the tidy Davidsonian picture is incomplete. If it is not, the free-will debate loses its principal libertarian option. The two sections make the same choice from opposite ends, which is why they cross-link.
Where events sit
This is an inventory question whose interest is that the argument for it is almost entirely formal. Nothing observational settles whether there are events; what settles it, if anything does, is whether the semantics of action sentences and causal statements can be given without them.
That makes it the best illustration of the Quinean method at work, and also of the method's limits. The criterion tells us that Davidson's regimentation commits us to events; it does not tell us that Davidson's regimentation is the right one, and the alternatives were constructed precisely to avoid the commitment. Which regimentation to accept is settled — if at all — by the usual theoretical virtues, which is exactly what the commitment page warned would happen.
The next page takes up the entities that stand to true sentences as events stand to happenings: facts and states of affairs.