Facts, Events, and States of Affairs

Brutus stabbed Caesar. Caesar died. The stabbing caused the death, and that Brutus stabbed Caesar is why the Senate descended into chaos. Each of these sentences seems to be about something, and the somethings are not objects or properties. They are not Brutus, not Caesar, not stabbing as a universal. They are the sort of thing that has a date, or is true, or is the reason for something else. Metaphysicians have proposed three main candidates for what these things are, and the disagreements among them decide how causation, truth and explanation can be stated at all.

Two ways of nominalising a sentence

Natural language lets us turn a sentence into a noun phrase in more than one way, and the results behave differently. Zeno Vendler drew attention to this in Linguistics in Philosophy (1967). Compare:

1. Brutus's stabbing of Caesar was sudden and brutal.
2. That Brutus stabbed Caesar is well documented.

The first phrase, a perfect nominal in Vendler's terminology, names something that can be sudden, long, slow or interrupted. We can say where and when it happened, and we can ask how it unfolded. The second phrase, a that-clause, names something that cannot be sudden and does not unfold. It can be well documented, surprising, or denied. It does not happen at 11 o'clock in the Senate House. It is the case.

The contrast is a clue and not a proof, because ordinary speech is loose. But it suggests that two different kinds of entity are in play. The one is datable and has a career: it begins, goes on, and ends. The other is closer to a way the world is, and the question is whether it needs to be reified at all.

Two features of facts are worth fixing at once. Facts are not located in the way events are. We say that the stabbing occurred in the Theatre of Pompey, but it is odd to say that the fact that Brutus stabbed Caesar occurred anywhere. And there are no false facts. A false proposition is not a fact that fails to obtain, since a fact that fails to obtain would be nothing at all.

Facts and states of affairs

P. F. Strawson, in "Truth" (1950), argued against J. L. Austin that facts are not things in the world waiting to be matched with statements. "Fact" is simply what a true statement states, and the idea of a fact as an item in the world separate from the thing described is a muddle. Austin's view was that a true statement corresponds to a fact, an item in the world. Strawson's reply is that when we ask what the fact is, we can only repeat the statement.

That deflationary line is still available, but a large part of the literature has resisted it for a reason that is metaphysical and not linguistic. If truth depends on how things are, there must be something about how things are that does the depending. Wittgenstein's Tractatus (1921) put the thought in its most famous form: the world is the totality of facts, not of things. A catalogue of objects would not tell you whether the cat is on the mat, since the cat and the mat are there either way. The world is also how they stand.

It is here that a second term enters the discussion.

On that usage a fact is an obtaining state of affairs. Brutus's being a sailor is a state of affairs, but not a fact, because it does not obtain. Plantinga's usage lets us talk about the way things might have been without conjuring false facts. D. M. Armstrong, whose A World of States of Affairs (1997) is the fullest development of the idea, concentrates on states of affairs that exist, and does not need the non-obtaining kind for his main purposes. Terminology varies from author to author here, so it is worth checking which sense a given text intends.

What the notion buys is a candidate for the thing that sentences are about when they say how things are. It also gives truth something to depend on. A sentence is true because a certain state of affairs obtains, and false because it does not.

The unity problem

If states of affairs are in the world, they need an account. The tempting thought is that Brutus's being a conspirator is just Brutus together with conspiracy, or the whole made of the two. Armstrong's central point is that this cannot be right.

The mereological sum of Brutus and the property of being a conspirator exists whether or not Brutus is a conspirator. If Brutus had been loyal, the property would still exist, Brutus would still exist, and so would the sum. Yet the state of affairs would not. So the state of affairs is more than its constituents. What it adds is the way they are connected, the fact that the object has the property. Armstrong called this a non-mereological mode of composition.

This is the old problem of predication in new clothes, and it threatens a regress. Suppose we say Brutus and conspiracy are united by a relation of instantiation. Then Brutus, conspiracy and instantiation are three constituents, and what unites them? F. H. Bradley's regress arguments in Appearance and Reality (1893) press just this. A further relation to tie the first to its terms will need yet another, and so on.

Armstrong's answer is to refuse the regress, not to answer it. The tie between a thing and its property is not a further constituent that needs tying. States of affairs are primitive unities, and the relation between constituents and the whole is internal, a necessary consequence of the state of affairs existing, and therefore no addition of being. Critics reply that this relocates the mystery and does not remove it. The dispute is a standing one, and for present purposes what matters is that anyone who accepts states of affairs takes on a decision about their unity.

Events as exemplifications

The second candidate is the datable one. Events, unlike facts, are things that happen, and the main question is what an event is made of. Jaegwon Kim's answer, from "Events as Property Exemplifications" (1976), is that an event is an object's having a property at a time.

Two exemplifications are the same event only if they have the same object, the same property and the same time. Brutus's stabbing Caesar at a particular moment is an event with Brutus as its object, stabbing Caesar as its property, and that moment as its time. The account explains why events have the features facts lack: since the time and the object are constituents, the event is datable and located where the object is.

It also shows how close events are to states of affairs. An exemplification of this kind is a state of affairs with a time added, and many philosophers have treated the two as near relations. The difference lies in what each is asked to do. A state of affairs is invoked to say how things are, and an event is invoked to say what happened, and what caused what.

This is where the question of grain arrives, because Kim's identity condition is demanding. On it, Brutus's stabbing Caesar and Brutus's killing Caesar are different events if they involve different properties, since stabbing and killing are different properties.

How finely to cut

Donald Davidson defended the coarse-grained view in "The Individuation of Events" (1969). His criterion was that events are identical if and only if they have exactly the same causes and effects. Brutus's stabbing is the killing, since whatever caused the one caused the other and whatever the one caused, the other did. The different descriptions, "stabbing" and "killing", pick out the same event under different aspects. Davidson wanted this because it let him say that a single action could be described in many ways, as flipping a switch, turning on a light and alerting a burglar, without there being a swarm of distinct actions.

The fine-grained theorist has a reply, and the best-known case against the region criterion shows it. Take a metal ball that is rotating and, at the same time, heating up. The rotation and the heating seem to be two happenings: the rotation could stop while the heating went on, and they have different effects, the rotation making a pattern in a coating of dust and the heating melting the wax on the surface. Yet they occupy exactly the same region of spacetime, so a region criterion counts them as one event, and a single event cannot cause the pattern but not the melting. Kim's account separates them at once, since rotating and heating are different properties.

Davidson's causal criterion also drew a charge of circularity from W. V. Quine in "Events and Reification" (1985). If events are to be the relata of causation, then identifying events by their causes and effects presupposes that we already know which events there are. Quine proposed instead that events are identical if they occupy the same spacetime region, and Davidson accepted the proposal. The ball case is then a difficulty for the position Davidson ended up in.

The cost of the fine-grained view is the reverse. It multiplies events beyond need. If every difference of description yields a difference of event, then a single stab gives rise to the stabbing, the killing, the assassination, the betrayal and the political act, and a world with a single action seems to contain a crowd. The fine-grained theorist answers that the crowd is harmless: the events are tightly related, share a location and are not independent additions to what exists. Whether that settles the matter depends on how much one minds a profusion of coincident entities, much as a statue and its clay seem to be two things in one place.

A third option, due to David Lewis in "Events" (1986), takes an event to be a property of a spatiotemporal region, or more exactly a class of regions across possible worlds, and so makes grain depend on how much the event's essence is held fixed. The account lets the choice of grain be flexible where the others fix it.

What causes what

The question of grain is not idle, because we have a reason to want events or facts to be the terms of causal claims, and the choice interacts with the choice of grain.

Davidson, in "Causal Relations" (1967), argued that causation holds between events. His reasoning was about logical form. Singular causal statements are extensional: if the short circuit caused the fire, then so did the event that occurred in the kitchen at midnight, if that is the same event. That is how a relation between particulars behaves. Statements of the form "because" are different, since "the fire started because there was a short circuit" is not preserved under every substitution of a co-referring term. Davidson took this to show that the causal relation itself relates events, and that the "because" idiom is a way of explaining.

D. H. Mellor, in The Facts of Causation (1995), argued the opposite: what stands in causal relations are facts. That the match was struck is what causes the fact that it lit. His reasons include that causes can be negative or absent, as when a failure to water the plant kills it, and there is no event of the not-watering to serve as a cause. Facts accommodate such cases because they include negative and general facts, where events need something to happen.

The positions are not forced to be exclusive. One might hold that events are the relata and facts the contents of explanations, so that "Caesar died because Brutus stabbed him" cites a fact and "the stabbing caused the death" relates events. Something like this division of labour is common, although it leaves the question of grain open.

The upshot is a division of labour, with open questions about how finely each of the three should be cut. Whichever way those questions go, the constituents of a state of affairs are properties, and the next article turns to the tool philosophers use to state how one family of properties is tied to another: supervenience.