Causation implies laws. A singular causal statement entails a general causal statement. If a caused b, we know that events like a will cause events like b. Thus universality is built into causation—the particular implies the universal. This puts causation in a very special class of relations: it is not generally true that a singular relational statement entails a general one. If a is to the left of b, it does not follow that everything like a will be the left of something like b: an apple can be to the left of a pear, but not all apples are to the left of pears. The same is true of all spatial relations: no general spatial proposition follows from the truth of a singular spatial proposition. Similarly for temporal relations: if a happens before b, it doesn’t follow that everything like a will happen before something like b—you might have dinner before going to a play, but it is not generally the case that dinners are followed by plays. Ditto for family relations: it doesn’t follow from my having a brother that everyone like me has a brother like mine. And the same seems true generally; only causal relations give rise to the kind of generality in question. This is because all singular causal relations are necessarily instances of general laws, whereas that is not the case for the other relations mentioned. The law need not be framed in the same terms as the singular causal statement, but some sort of description will exist under which the instance exemplifies a law. Everything happens by law; therefore all causal relations imply underlying laws.
We should view this as more surprising than we do. For how is it possible for the particular case to have implications beyond itself, covering indefinitely many other cases? How can we derive a universal statement from a singular statement? We can derive an existential statement from a singular statement, but how can we move from what is true in a particular instance to what is true in all instances? The causal relation between particulars seems to encompass causal relations between quite distinct and often remote particulars: if a certain causal relation holds on earth, we can infer that it generalizes to other galaxies. This gives us amazing powers of knowledge: we just need to know that this caused that and we thereby know that everything like the former causes something like the latter. Imagine if knowing that this cup is on the table enabled us to know that every cup is on a table! Yet causation seems somehow to condense the universal into the particular: if a really did cause b, then no matter where you go, whenever you have something like a it will cause something like b. Causation is not just the cement of the universe; it is a cement that repeats itself endlessly, holding things together in the same recurring pattern. Once you know one part of the pattern you know them all. The puzzle is how an individual instance of a relation can “contain” all the other instances. Generally, if a relation R relates individuals a and b, we can infer nothing about whether other similar individuals are related by R; but in the case of the causal relation, we can infer a universal proposition from a specific one. This is because every particular case is necessarily an instance of something more general. And that seems puzzling, almost miraculous, as if great tracts of the universe are coiled inside a particular localized case.
Consider two other relations that have generality built into them: logical and deontic relations. If a particular statement entails another particular statement, this is always an instance of something more general: the proposition expressed by the first statement entails the proposition expressed by the second, so that every individual statement will stand in the entailment relation. Similarly, if one person has a moral duty with respect to another, this implies that anyone relevantly like the first will have just such a duty to someone relevantly like the second. As Kant would say, particular moral maxims can be universalized. Is there a puzzle about how this is possible? If there is, it is surely superficial, since logical and deontic relations primarily hold between types not tokens—types of statement, types of person. Conjunctive propositions, say, have certain logical implications, which are inherited by particular expressions of them; and fathers and sons as general categories have certain duties to each other, which then apply to specific people. That is, the relations in question hold in the first instance between something other than concrete particulars and are understood as such. It is not a matter of inferring the universal from the particular but recognizing the universal in the particular. We know those relations to hold without having to inspect the empirical world of particulars. It might even be said, by way of emphasis, that logical and deontic relations don’t strictly hold between particulars at all—this is just a manner of speaking about more abstract relations, harmless enough if we don’t let it mislead us as to the true ontological situation.
This suggests an approach to the puzzle of causation that has some reassuringly familiar elements. What if we say that the causal relation holds primarily between types not tokens? Then generality will be built into it from the start. When token events causally interact this is an instance of a type-event interaction; the former is derivative from the latter. Thus universality is guaranteed because it is built into the nature of the basic causal relation: causation is a relation between event-types in somewhat the way logical and deontic relations are relations between types. This is familiar because it is commonly accepted that token events stand in causal relations in virtue of the properties they instantiate: it is not events tout court that stand in causal relations but events “under descriptions”, i.e. inasmuch as they instantiate causally relevant properties. It is the electric charge of a battery that causes an electrical device to work not the color of the battery, and events have causal powers in virtue of some of their properties though not all (being an event recorded in the history books, for example). Spatial and temporal relations relate things irrespective of their intrinsic properties, but causal relations between things depend entirely on their intrinsic properties (i.e. their nature). Thus the primary locus of causation is properties, which are inherently general. The puzzle arises when we think of cause and effect as particulars and then wonder how the particular can contain the general, but in fact causation is inherently general because of the essential role of properties in causation—they are the primary bearers of causal powers. Laws relate properties, and causation consists of laws in action. The universal is already present in the particular case. The form of a singular causal statement is, “a being F caused b to be G”, where the causally relevant properties figure essentially in the fact; so the singular instance already includes general properties as causal agents. The causal structure of the universe accordingly relates properties not just particulars. This is what makes the causal relation different from other relations, and solves the puzzle of causation. Causal structure is not the sum of isolated instances of causation between particulars but of general causal principles linking properties.
 There is an extensive literature on this question, with notable contributions from Davidson, Anscombe, and others; but I won’t get into this and simply assume a well-known position.
 Here we might think of Wittgenstein’s discussion of the way meaning seems magically to contain future use in Philosophical Investigations.
 It may be true that singular causal statements are referentially transparent statements about token events, but it doesn’t follow that causation itself works without reliance on selected causally relevant properties. No event has causal powers just by being that event.
 In a world of bare particulars, if such there could be, there could be no causation, because there would be no exemplified properties to do the work of causation. Bare particulars would have to be causally idle. In a slogan: no causation without exemplification.