CHAPTER VIII

Action and Time

§1. Necessities Across Time. In this and the following chapter I shall develop the view that time is based on action. It turns out that the actions required are components of physical individuals rather than separately existing entities. The action view of time is, then, neither an absolutist nor a relativist view. The “absolutist” gives instants a reality that is independent of what occurs in them. The “relativist” views instants as derivative from relations among occurrences. So construed, relativism requires that relations exist, either temporal relations or other relations to which temporal relations might be reduced, such as causal relations. The advocate of the action view, like the relativist, rejects the independent existence of instants, but, unlike the relativist, rejects the existence of relations. The action view of time is a special case of the general view of relations described in the preceding chapter. In this view, there are only temporal relational properties, and they are grounded by actions.

Is it a consequence of the action view of time that actions belong to the required ontology? I contend that it is for the following reason: among the necessities required for the practice of action on prior experience will be necessities between conditions at distinct times. How else could the practice of action on prior experience be what I called in Chapter I a fundamental practice? The reason there would be no communication or no theorizing without such a practice is that communication and theorizing suppose that people have beliefs as to how speakers and hearers and simply bits of matter react after being submitted to certain stimuli. Forming these beliefs is a matter of action on prior experience.

But now suppose action on prior experience involved merely the projection of the coexistence of conditions on the basis of prior experience. The practice of action on prior experience could then no longer be called fundamental. Other human practices would be conceivable without it. Necessities between conditions at distinct times cannot be rejected without rendering the practice of action on prior experience in its role as a fundamental practice incoherent with our view of the world. This need for necessities across time, taken together with the action view of time, puts actions into the required ontology.

But how do necessities across time fit into the picture of limited variety (or of levels of limited variety) drawn in Chapter V? Suppose a’s being F is not just followed by b’s being G at a later time, but that if a is F then b is G later. Thus a has the conditional temporal property (F_→ Gb later). Also suppose a has the property H. So, given the assumptions for limited variety, there will be a chance that □ a(Ha → (Fa→ Gb later)). If induction is to allow us to assert this universally of individuals other than a that are also H, there must then be a chance that H is a kind property, and thus a chance that □ a(Ha). It follows immediately that there will be a chance that □ a(Fa→ Gb later), that is, a chance that there is a certain necessity across time.

However, there are two reasons why one might deny that the limited variety picture yields a chance for such a necessity across time. First, it might be held that a’s conditional property (F_ → Gb later) is a relational property. It would the׳n fail to come under the properties considered by the limited variety principle. Recall that relational properties are not just correlative, as this one is, but also have foundations. Yet (F_→ Gb later) can be a property of a even though no properties of a and b are foundations for it.

Second, it will be pointed out that there are possibly an infinite number of conditional temporal properties in any individual. How then can a requirement of limited independent variety be satisfied? This is a crucial question at this point. So far, I have ignored time in that I have not indicated whether the necessities needed were merely necessities among coexistent features or also necessities across time. Now, however, it is clear that the practice of action on the basis of prior experience requires necessities across time. We have just seen that, for there to be such necessities, conditional temporal properties should be among the properties that individuals have. Yet if such properties are to have a chance of being necessary, their presence should not lead to an unlimited variety.

It might seem, though, that any individual could have, as a result of all of its states, an infinite number of consequences, and that these consequences could be mutually independent. As a result, the principle of limited variety could not be invoked to say there is a chance any of these connections is necessary. For the principle to be usable at all, conditional temporal properties would have to be excluded from consideration, just as relational properties were─and for the same reason. The structure of individuals required by limited variety would become irrelevant to the practice of action on prior experience, because none of the necessities posited by such a structure would be necessities across time.

To meet this difficulty, two assumptions must be accepted. First, it is to be assumed that conditionals relating a single condition of a given individual to temporal consequences of the same kind for other individuals are not independent. If a’s condition of being F gives rise to b’s condition of being G sometime later, then it will also give rise to c’s being G sometime later. (Fa→ Gb later) and (Fa → Gc later) are not then independent, and the thrust of the objection is partly deflected. For this assumption to be important, G will generally be unspecific in a number of ways. A match’s flaming does not imply specific temperature increments for bodies warmed by it. Rather G would here be expressible by a formula relating variables for distance, intervening medium, size, and material to change of temperature. To repeat, our first assumption is that similar consequences of a given state do not arise independently.

This first assumption is not of itself sufficient. There could still be an unlimited number of independent temporal conditional properties for a given individual. This would be the case if the conditions of an individual had, jointly, an unlimited number of different kinds of consequences and if the connections between the conditions and the consequences were independent. So our second assumption is that the temporal conditional properties of an individual with different kinds of consequents are not of an unlimited independent variety. I am able to speak here about the number of kinds of consequents, rather than about the number of consequents, since, by the first assumption, conditional properties with the same antecedent and with consequents of the same kind for distinct individuals are not independent.

In sum, the practice of action on prior experience requires necessities across time. Specifically, the necessities across time that are needed are given by the requirements that (1) individuals have to have conditional temporal properties, that is, properties such as the property a has when (Fa → Gb later), and that (2) individuals have conditional temporal properties in such a way that the principle of limited independent variety (or the principle of levels of limited independent variety) is not upset. To insure that it is not upset, it must be assumed that (a) conditionals relating a single condition of an individual to temporal consequences that are of the same kind but are for distinct individuals are not independent conditionals, and that (b) the independent temporal conditional properties of a given individual with different kinds of consequents are not unlimited. But if action is taken among the foundations of temporal relational properties, then the existence of conditional temporal properties─that inevitably imply the existence of temporal relational properties─must rest, in part at least, on action.

§2. Temporal Atomism and Temporal Holism. The action view of time contrasts with two classical views of time, the atomistic and the holistic views. The atomistic view is simply the application to time of the doctrine that relations are entities distinct from their relata. In this view of time, there are, in addition to conditions such as being red and being green, temporal relations among these conditions. It was seen in the last chapter that relations among distinct entities are external. And, in particular, temporal relations are external to the conditions they stand between. Hence no condition can be necessarily followed by another, or necessarily preceded by another.

There will, however, be such connections if there are necessities across time. For, if it is a necessity of a that if Fa then Gb later, then, quite obviously, when the condition of a’s being F obtains, it is a necessity of a that A’s being F is followed by b’s being G. Thus the existence of temporal relations abolishes necessities across time. The conditions in a network of temporal relations must, then, be like isolated atoms.

The holistic view, on the other hand, does not regard temporal relations as fundamental. The fundamental factor in regard to time is, in this view, a kind of action that encompasses a multiplicity of different entities as a whole encompasses its parts. The conditions that we speak of as temporally separated are among the entities encompassed by the surge of action. The encompassed conditions are not then independent unities; they have the status of components of action. Temporal relations among conditions encompassed by an action will themselves be components of the action.

So the holistic view is a special case of the view that Russell called the “monistic״ view of relations. A relation, on this view, is a component of an entity that also has the relata of this relation as its components. So the relation is internal in our second sense. Clearly then, in the holistic view, the conditions that are components of an enduring action are not distinct from one another or from the encompassing action. ”In short,” says Bergson, an advocate of the holistic view of time, “it is necessary to admit two types of multiplicity, two possible senses of the word ‘distinguish’, two conceptions, the one qualitative and the other quantitative, of the difference between same and other.”¹ Bergson overcomes the atomistic isolation of conditions in time by dissolving their numerical distinctness in a whole where they are only qualitatively distinct, or, in our terms, where they are only different components.

It has often been supposed that temporal holism is the only alternative to temporal atomism. Holism is definitely on the side of an ontology of components; actions are not simple entities but entities with different, though not distinct, components. On the other hand, temporal atomism does not require that there be entities with components. It might then seem that we should side with holism against the ontology of external relations of the atomistic view of time. There is, though, a non-holistic alternative to temporal atomism.

But why not be satisfied with holism? Holism seems most plausible for actions that do not have effects outside their agents. But consider the case of Jones's sanding a block of wood. The conditions encompassed by the action are not only conditions of Jones, but also conditions of the block of wood. Both the condition of Jones’s forearm as he begins a stroke and the condition of the wood's surface after Jones has completed the stroke are dissolved into the unity of the sanding.

It is a minimal requirement for the sameness of conditions that conditions of distinct physical individuals be distinct. But for the holist, the condition of Jones’s forearm and the condition of the wood’s surface are not distinct, but at best different. It then follows that Jones’s forearm and the block’s surface are not distinct, but at best only different. Yet at the same time, the holist would want to be able to claim that some things are genuinely distinct; for example, Jones’s sanding and Jones’s walking would be distinct actions. On the face of it, if anything is distinct from something else, then surely it is absurd to claim that Jones's forearm and the block's surface are the same particular. There is the additional objection that holism displaces individuals by actions as the ultimate determiners of sameness. But only in Chapter XI will the basis of this objection be established.

The non-holistic alternative to atomism─to be prepared for by the discussion of action in this chapter and to be presented in detail in the next chapter─is a special case of the theory of relational properties described in the preceding chapter. It can be described briefly as follows. Suppose an individual performs a certain action. There will be conditions that are the results of this action. These results are not swallowed up by the action; they remain distinct from it. The action and any component corresponding to one of these results are, together, the foundations for a temporal relational property. Since the foundations imply the temporal relational property, there is no isolation of entities at different times as there is in the atomistic view. There is a genuine togetherness supplied by the foundations, which may be components of distinct entities.

Atomists have typically seen their own position as a response to a holistic destruction of the distinctness of instants. Their commitment to a certain view on the nature of distinctness has prevented their consideration of the action view of time just outlined. Descartes, for example, seems to have thought that if moments of time are distinct, there can be no causal action between the occurrences that fill them: “The present time has no causal dependence on the time immediately preceding it.”² It seems probable that his argument for this was as follows:

(1) If one moment can be clearly thought to be distinct from another, the former could exist without the latter.³

In other words, if one moment is necessarily connected to another, they cannot be thought to be distinct, and hence, for Descartes, it is not possible that they are distinct. But:

(2) A causal action between moments would be a necessary connection between them.

So if moments were joined by a causal action they could not be distinct. Hence, since distinct moments can be clearly thought to be distinct:

(3) If moments are distinct there is no causal action between them.

So in Descartes’ view of distinctness, if there were causal action, it would have to be like the action of temporal holism. It would dissolve the distinctness of moments within it into a unity. Nonetheless, the distinctness of moments is for Descartes compatible with God’s causal action in respect to them, provided that the content of each moment is produced by a distinct action of God’s.⁴

The crux of Descartes’ argument is premiss (1), that is, the implication from distinguishable to separable. All along I have insisted that this premiss cannot cohere with human practice. If then we deny this premiss, causal action need not be interpreted as a whole within which there can be no distinctness of moments. That is, if we deny (1), the acceptance of causal action between moments does not commit us to saying, with the holist, that those moments are not distinct. And so by denying (1), we get beyond the dichotomy of atomism and holism.

Hume adds little to Descartes’ reasoning when he argues that the idea of a duration is not the idea of a whole encompassing successive objects, but is rather the succession of ideas of objects. Since the distinct is treated by him as separable, atomism is the only alternative he sees to the holistic view he rejects. Since five notes played on a flute are distinct, they are, for Hume as for Descartes, separable. But then if there is a whole─say, the act of playing the flute─that encompasses them all, one would be faced with an absurdity. For this playing is not separable from these notes being played, and hence this playing is not distinct from the notes. Yet since there are five distinct notes, the one playing must, contradictorily, be five playings. So there is no act of playing at all but only the succession of notes.⁵ That is, the alternative to holism is the view that relations of succession stand between isolated events. However, if the distinct need not be separable, the notes might be distinct from the playing even though each of them is an intermediate result of the playing. Their temporal relations to one another need not be taken any longer as irreducible. Rather, the foundations of these relations are to be found in the stages of the action of playing (Chapter IX, §3, ad fin).

I shall tie this section to §1, by the following observations. First, the atomist’s conflation of the distinct with the separable is incompatible with the requirement that necessities across time not be denied if one is to engage in the practice of action on prior experience. For necessities across time are necessities connecting distinct conditions. But the conflation of the distinct with the separable is the chief premiss for temporal atomism, which is the view that relations, not actions, are needed for temporal relatedness. So the direct argument for temporal atomism given by (1)-(3) has to be rejected as unsound.

Note, second, that this argument can be turned around. That is, starting with temporal atomism, one can attempt to show that there are no necessary connections across time, and hence that what is distinguishable in time is separable. In fact, this reverse argument has already been constructed. This was done in Chapter VII, §1, when it was shown that if there are relations between distinct entities they are external. So, in particular, if there are temporal relations, they are external, and hence there are no necessary connections across time. The argument is valid, and it is the backbone of Humean and post-Humean anti-necessitarianism. It relies on temporal atomism as a premiss, which in turn relies on the existence of relations between distinct entities. But we saw that such relations make relatedness impossible. Since, however, there is temporal relatedness, temporal relations and hence the premiss of temporal atomism must be rejected. Temporal atomism will be deprived of any residual appeal if we can establish below the action view of time as a positive alternative.

Third, let us note that the conflation of distinct with separable that is behind the temporal atomist’s argument (1)-(3) is not the same as the conflation of distinct with different that characterizes the ontology of simples. Thus the temporal atomist’s relata may be complex in the sense of having different components. However, the prime motivation for a component ontology is, as we saw in Chapter III, the need to integrate natures and things. If necessary connections across time are denied by conflating the distinct with the separable, then the role of natures is greatly limited. Once natures have in this way lost their importance, the problem of integrating them with things will be ignored. The step of conflating the distinct with the different then seems the obvious one to take, since there is no apparent need for entities with components.

§3. Progressive and Complete Actions. Whether it will be possible to establish an action view of time that avoids both atomism and holism will depend on whether we are forced to embrace what is apparently the most commonly accepted philosophical view of action. This is the view that what I shall call “complete” action is the only kind of action. Just as there is a contrast between a property and the having of it, so too there is a contrast between an action in what I call the “progressive” sense and the doing of the action. The doing of the action is the correlative complete action. In terms of our earlier distinction between components and conditions, a property and a progressive action are components of individuals, whereas the having of a property and the doing of an action are conditions of individuals. Conditions that are the doings of actions make up one of the various kinds of conditions that are generally called events. If we are forced to accept action only in the form of complete action then, indeed, as I shall presently show, the action view of time must fail.

Verbs in their gerundive form give us ambiguous referring terms for actions. They can be made to refer to progressive or to complete actions. But we also have some terms that unambiguously pick out complete actions. ‘Race’ is a complete-action term, but ‘running’ serves for both progressive and complete actions. If I say that the running took four minutes, I am referring to the complete action, to the performing of the running. If, however, I say that the running continues to be done or that the running is swift, I am referring to the progressive action.

It is important to notice that progressive action is not extensive and hence not extensively divisible. The running you are doing just does not have a first half and a second half. This is not because it is an extended whole that somehow cannot be divided. Rather it is because it is not an extended whole at all. Your doing the running, as opposed to the running you are doing, does have a first half and a second half. There are, then, various phases of your performance. The third lap is a phase in your doing the running, and it may have taken more time than either of the first two. But the progressive action of running does not have phases or stages; there is no earlier and later progressive action of running during the race.

Whitehead is perfectly right on this when he says that “in every act of becoming there is the becoming of something with temporal extension; but . . . the act itself is not extensive, in the sense that it is divisible into earlier and later acts of becoming.”⁶ The temporally extensive entity that comes to be is, in my analysis, the complete action. It is correlative to the progressive action. What Whitehead calls the act of becoming I am interpreting here as a generic expression for various progressive actions.

The condition of being red may last ten minutes. But the component red is not itself lasting or ephemeral. Likewise, the condition of doing cutting may be a condition I am in for a full hour. But the corresponding component, cutting, is, like red, neither lasting nor ephemeral. Not only does the condition of being red last ten minutes, but also I am in that condition at instants within that stretch. And so the condition of doing cutting that lasts a full hour is a condition of me at each instant in the hour. There is, of course, the peculiarity of speaking of conditions lasting through times, during parts of which the obtainings of the conditions were interrupted. My condition of doing cutting lasted a full hour, even though I rested several times during that hour. I shall ignore these peculiarities in identifying complete actions and treat what we speak of as unitary complete actions as though they were uninterrupted.

Why, then, must the action view of time fail if we admit only complete action? In the first place, since complete actions are, in general, extensive, they can hardly serve in an attempt to ground time. For a ground for time should give some account of how there can be entities, such as complete actions, that have temporally separate parts. And merely to say there are such entities is not to give an account of them. So if temporal relational properties have foundations, then the temporal structure of complete actions has the same foundations.

It will not be complete action but progressive action that can serve us here. Complete actions depend upon progressive actions in the same manner that any condition of an entity depends upon a corresponding component of that entity. But there is also a special dependence in the case of complete action, a dependence that has to do with time. In the absence of progressive actions, there are no foundations for temporal relational properties. Without temporal relational properties the conditions of entities would lack a temporal structure. So a complete action─an event, an actual occasion, an occurrence of action, a state of action, or a fact of doing something─exists through dependence on progrèssive action and has its extensiveness on the basis of progressive action.

In the second place, if we should try to ground time on complete actions without reliance on progressive actions, we would fail to ground it on components of individuals. But the action view of time puts forth precisely such a grounding on components. The difficulty is that by cutting ourselves off from progressive actions, there are no components of physical individuals that are correlative with those conditions that are complete actions. For what components are the correlatives of complete actions if not progressive actions?

When progressive actions are rejected, complete actions are cut off from their roots in physical individuals. Not only do we fail in this way to ground time in the components of physical individuals, but we erect a dualism of events and things, for an event that is associated with a thing is no longer dependent on there being an action that is a component of that thing. The things themselves are static; change comes into the picture by associating them with complete actions, which are numerically distinct from the things since they are conditions of those things.

Needless to say, this way of attempting to bring change into the picture is abortive. For, in effect, it is an attempt to have races without running, falls without falling, and dances without dancing. By excluding progressive actions, we have the absurdity of there being the complete performance without something performed. A parallel view would associate the condition of having red with individuals that lack the component red. There would be a condition of having a property without the property.

Inevitably, this view requires modification. It is absurd to have races without running (cf. Chapter XI, §4). For an ontology of simples, there may be a running that corresponds to the race even though this progressive action, the running, is not a component of the runner but, rather, a distinct entity with the character of a universal. In such an ontology, the distinction between progressive and complete action would be a special case of that between universals and exemplifications of universals. Still, complete actions without progressive actions would be impossible. The natural course for those who do not countenance progressive actions is to deny that even complete actions are genuine actions. Complete actions are reduced to sets of conditions that are the havings of properties, where these conditions are strung out in unbroken sequence. A multiplicity of property conditions in sequence replaces a unitary action condition. So not only is the running rejected, but the race is rejected as an action. Each of the conditions in the sequence is correlative with a property that is a component of a physical entity. The dualism of things and events, resulting from breaking this correlativity between conditions and components in the case of conditions that are complete actions, is overcome at a level that does not involve action at all.

Ockham, who believed that everything is either a substance or a quality, claimed that words like ‘action’ and ‘motion’ do not stand for anything.⁷ Rather, for a body to be in motion “it suffices that . . . [it] . . . continuously─without any interruption of time, or rest─acquire or lose something, part by part, one after another.”⁸ But such a reduction of complete action to property conditions in sequence posits the atomistic view of time. The property conditions are in sequence, and any attempt to fasten on the sequential relations and discover the grounds of these relations will lead straight back to the actions that Ockham wished to eliminate. Relations must then be taken as irreducible. The property conditions in the sequence are separated by distinct in-between entities.⁹ Since there are no relations, however, the reduction of complete actions cannot be carried out. Nonetheless, this reduction was called for, once progressive actions were pulled out from under complete actions. We are not put into a position of needing to carry out this impossible reduction if we allow for progressive actions in the first place.

§4. Components and Conditions. The distinction between progressive and complete action has been drawn in terms of that between components and conditions. Further elaboration of this latter distinction, as it applies to actions, is desirable. This shall be done by illustrating some further consequences of the component-condition distinction.

Components, but not conditions, are distinguished numerically on the basis of the entities with them. The summer green and the autumn yellow of this leaf share the unity of the leaf, though they are different components. But “a changing thing changes from one definite condition to another (ek tinos eis ti).”¹⁰ The leaf changes from the condition of being green to that of being yellow. Now if this is, in turn, only a matter of difference within the unity of the changing being, we are faced with the consequence that the multiplicity in time of the conditions of a changing being is not a multiplicity of distinct moments and that the life of a single changing being is, in fact, only a single moment. Change does not then involve sameness through distinctness of states, but sameness with difference. Of itself, this consequence is not an impossible one to have to live with.

But if the condition of having green is, like green itself, not distinct from the green entity, then it, like green itself, will be a component of the green entity. Now each component has a corresponding condition. So the condition of having green will correspond to the condition of having the condition of having green, and so on endlessly. The sequence of requirements for green to be a component of any entity is then never complete. There will always be a further condition of having a condition. To avoid this objectionable regress, it suffices to regard conditions as entities distinct from the entities with the corresponding components. Since a condition is a distinct entity there is no condition of having a condition, and thus no regress. Granted, then, that conditions are distinct from their parent entities, are they distinct from one another when they have the same parent entity? Perhaps the cube’s conditions of being made of stone and of being grey are only different from one another. But this cannot be, for then─since neither condition would be a component of the other─they would have to be components of some entity different from each condition. And again there would be a regress of conditions of having conditions.

But what justification is there for putting performances, doings, events, and the like under the heading of conditions? The justification comes from treating progressive actions as components. For entities with components can be said, quite literally, to be in the condition of having those components. Now if progressive actions are components, then entities with these components are in the conditions of having them. But what is it to have a progressive action if not to do a progressive action? For an agent to have pulling as a component it is necessary for that agent to be doing pulling. So if having pulling is a condition, doing pulling is a condition. This accords with the idea that doing pulling is not two actions, a doing and a pulling, by interpreting the doing the pulling as merely the condition of having pulling as a component. But this all supposes that sense can be made of the notion that a progressive action is a component.

What else might be the pulling, cutting, teaching, or exploding that an entity does? Perhaps the progressive action is a universal. Its being exemplified would then be a complete action. One would be tempted by this interpretation only if one were antecedently committed to viewing properties as universals rather than as components. One would be strongly motivated to do this on one of only two assumptions. First, one would treat properties as universals if the individuals that have properties are simples. Properties would then be distinct from individuals and could be had by several individuals. Second, one would treat properties as universals if, in one’s ontology, there were no individuals but were instead conditions of having properties, actions, and parts. What is had when there is a condition of having a property is not the same particular as this having itself. Either it is the same particular as some individual, which is impossible here, or it is an entity with its own unity that could then be had in several conditions.

Neither the first nor the second assumption can be accepted by the ontology of components, since it is incompatible with an ontology of simples and since for it conditions are dependent on individuals. It is reasonable for us, then, to consider the doing of cutting as having the component cutting rather than as participâting in the universal cutting. (Even so, resorting to universals allows one to treat doings as conditions. For participating in the universal cutting is just being in the condition of having cutting, though not of having cutting as a component.)

Discussions of action are often vitiated by failing to distinguish progressive from complete action. Consider, for example, the problern raised by the fact that actions recur. Jones runs after breakfast and then again after lunch. The recurrence could be explained if the running were a universal. But it might be objected that this explains too much. Smith’s running and Jones’s running would be one running, not two. Yet there is an obvious reply. Is it the progressive or the complete actions of Smith and Jones that must be distinct? Even if their progressive actions are the same universal, their complete actions are distinct instantiations of that universal. This is sufficient to account for the obvious fact that Smith and Jones are engaged in two performances.

Suppose actions are said to be propositions in the realistic, not the intentional sense. Thus, proposition here means what fact-of or condition means. Jones’s running after breakfast is one proposition, but his running after lunch is another, for otherwise it would be impossible to say Jones ran twice. But this seems to make it impossible to explain the fact that the second running is a recurrence of the first.¹¹ Since the propositions are distinct entities, we are left with no basis for saying it was a repetition.

Again the answer to the conundrum lies in the distinction between progressive and complete action. As distinct, the propositions serve well in the role of complete actions. For, indeed, there were two performances. But to say that the running recurred is to make a claim about progressive action. The progressive actions of an individual are not entities distinct from that individual. This fits with the fact that recurrence does not seem to imply distinctness. To go on to say that we should call actions that recur “quasipropositions” is to make the mistake of supposing that, since complete actions are propositions, progressive actions are sufficiently like them to be similar to propositions. Far from being similar to propositions, progressive actions are the components that by being had give rise to propositions in the present realistic sense.

Still one wonders how, in fact, the component view of progrèssive action handles recurrence. The problem is like that of how redness of the skin recurs with a new exposure to the sun. In the conceptualist theory of difference─that if a is different from b their concepts are not the same─the redness in June and that in August are not different, though the corresponding conditions are distinct. But we have correlated difference with distinctness of conditions. Since I lived through two sunburns, the color component I had in June is different from that I had in August. Analogously, Jones’s two runs involve different, though not distinct, components of running. In this view, there is no universal entity instantiated by Jones during each run that makes the second a recurrence of the first. But can we explain the recurrence if the two components are different? The key to the matter is similarity. The June redness and the August redness are different entities, but they are similar entities. There is recurrence of a component when it is followed by a different but similar component of the same particular. Jones’s running recurs since his different components of running are similar. Jones’s running does not recur in Smith since Smith’s running, though similar, is not a component of the same particular.

Whether actions are treated as progressive or complete, there is a problem associated with the use of adverbial and prepositional modifiers in statements reporting actions.¹² If I say that Jones ran to the door, am I saying that Jones has the component running-tothe-door? If so, is this component different from running simpliciter? (Or, in terms of conditions rather than components, am I claiming that Jones is in the condition of having running-to-thedoor? If so, is this condition distinct from that of having running?) And is there, over and above the component running and the component running-to-the-door, also the component running-to-the-door-in-a-bathrobe?

There seem to be three options here. If we take the first, all components but the running simpliciter are rejected. In the second, all components are admitted. That is, we admit running simpliciter, running-to-the-door, running-to-the-door-in-a-bathrobe, and so on until we arrive at a component that contains the full complexity of the situation. In the third option, all components short of the one that reflects the full complexity of the situation are rejected. They are rejected as unreal abstractions from the component with full complexity. Taking the second option in֊ volves unnecesary multiplication of actions if either the first or the third is defensible. The third prohibits our saying anything true about actions, since a true statement would doubtless have to have an endless number of modifiers as constituents. Is then the first defensible? It is only if one can explain how a modifier makes a difference in a statement to which it is added without requiring a more complex action in the world.

How this is indeed the case may be explained as follows. In general, a modifier in an action statement serves to indicate the role of the component action in regard to the individual with it. Thus, to say Jones ran to the door is to say (i) that running is a component of Jones and (ii) that this running puts Jones in the direction of the door. The putting of Jones in that direction is not a component of the running, for components do not have components. Rather, to say the component puts Jones in that direction is to say the component is the basis in Jones for his direction.

Conversely, to say Jones necessarily runs is to imply that running is a component of Jones and that this component is due to his nature. Here the role of the progressive action is that of being dependent on something about Jones, his nature, whereas in the previous case the action was the basis of something about Jones, his direction. When Jones runs fast, his running enables him to be at a greater distance from a given point in a given time than the typical member of his kind would be. (Depending on the context, ‘Jones runs fast’ can mean other things as well.) Again, the speed is not a property of the progressive action. Rather, the modifier ‘fast’ indicates what the progressive action does for the individual with it. Similarly, a bright blue vase has a component that gives the vase a bright appearance. So in the case of properties as well as actions, modifiers serve to indicate the roles these components play for the individuals with them.

Is a new entity─a role─being added in order to understand modifiers? Not at all, for roles can be treated as conditional properties, which are already familiar enough in the required ontology. To say the component puts Jones in a certain direction, enables him to be in a certain place, or gives the vase a certain appearance is to say─beyond implying that the individual has the component ─that if the individual has the component then it will be in that direction, be enabled to be at that place, or be given that appearance. (For the modifier ‘necessarily’, reference to the component will occur in the consequent, not the antecedent, of such a conditional.) The same pattern can be used for modifiers in relation to conditions, rather than components. When Jones runs to the door, having running is a condition that puts him in the direction of the door. That is, he has this condition only if the door is the direction he is headed in.

Of course, only if these components and conditions are conceived in the manner of our present ontology can they have such implications. If running were a universal, then its being instanced in a certain individual would not imply any particular direction for that individual. But in our ontology, running is not distinct from the runner, and two different runnings of the same runner may well be dissimilar. Dissimilar components, or conditions, will have different implications.

Not all modifiers serve to indicate the role of a component. Some, in fact, serve to deny that a certain component is present.¹³ For example, nearly running is not running at all. My concern, however, is solely with cases in which the truth of the statement with the modifier requires the presence of the corresponding component. Moreover, adjectival modifiers of kind nouns do indeed signify components rather than simply indicate the roles for components. Thus ‘red’ in ‘red ball’ signifies a component of a ball and not just the role of the kind-property signified by ‘ball’ in relation to the individual ball.

Though the distinction of component from condition is not crucial for understanding modifiers, it is of greatest importance as regards causation. If a causes b, then it seems reasonable to say that a causes any particular that is the same as b. However, acting beings and their progressive actions are the same particulars. So if a causes the running that Jones has, then, by this principle, a would also cause Jones, which, of course, a need not do. However, if a causes the corresponding condition─the complete action of doing running─then, since conditions are not the same particulars as the individuals with them, it does not follow that a causes Jones as well. A similar line of reasoning would support the view that it is the having of properties and not properties that are caused. “In making the bronze round one makes neither the round nor the spherical, but something else; one puts this form into something other than itself.”¹⁴ What is caused has a unity of its own, whereas components have their unity from individuals with them.

But if a causes Jones to be doing running without causing the running itself, one seems forced to conclude that the running existed prior to Jones’s doing it. The cause, aג merely brings Jones and a pre-existent running together.¹⁵ Running could no longer be viewed as a component. Both actions and properties would have to be entities distinct from the individuals they come to be associated with. This retreat can be halted only by appealing to a variation on the Aristotelian notion of “potential forms.”¹⁶ If a succeeds in causing Jones to run, then it was true of Jones that, if he were subject to a certain influence, he would run. The cause, a, provides this influence, and then it is only a matter of logic that Jones will run.

It is not, then, necessary to appeal to a running that existed apart from any individual. It is only necessary to appeal to a conditional property, the property an individual has when it will run provided there is a certain influence. As was observed earlier, to have a conditional property it is not necessary to have the components associated with either the antecedent or the consequent. So, given conditional properties, causes can cause conditions without causing components even though the components do not preexist. Notice that if a causes the condition of Jones’s existing, then clearly the relevant conditional property is not Jones’s but rather it is a’s property of being such that, if it is in certain circumstances, Jones will exist.

§5. Action as Required by Temporal Separation. So far we have seen that atomistic and holistic views of time are inadmissible. The atomistic view eliminates the possibility of necessities across time; the holistic view destroys the distinctness of the agent and what is acted upon. We have also seen that if an action view of time is feasible, it must be progressive action that grounds time rather than complete action, since the latter is already temporally extended. Now it is time to ask if there are any reasonable alternatives left to the view that actions are among the foundations of temporal relational properties.

Perhaps the temporally separate property conditions of entities are such that their corresponding properties are the grounds of their temporal separation. The two different sizes of a balloon are the ground of the temporal separation of its conditions of being of the two sizes. This view has all the advantages of temporal atomism without the disadvantages. For it avoids introducing actions as components additional to properties without having to resort, as temporal atomism must, to temporal relations. Let us call this the property view of time in order to contrast it with the action view, where here and henceforth I shall mean by action progressive action unless a qualification is added.

The property view of time conflicts with a fundamental fact relating change and difference. It is this fact that vindicates the inclusion of actions along with properties and parts among the components of individuals. In the property view, an individual's being in conditions of having properties that we would normally think of as incompatible properties implies the temporal separateness of these conditions. But will these conditions be temporally separate if the individual does not lose one of the properties and gain the other? Of course not. Saying that it has both properties implies just such changes. Otherwise, one would be making an impossible claim by saying that it has both properties. The obvious fact is that not only do the incompatible conditions exist, but also the changes involving the losing of one property and the gaining of another exist. The property view of time supposes erroneously that time can be accounted for by ignoring these changes and considering only the properties. It is clear, though, that without these changes a given individual would not even have the incompatible properties.

There are two ways of combating this argument against the property view. The first accepts the general framework of this discussion, whereas the second poses a radically different ontological framework. Neither way is entirely successful.

The first way is to treat losing or gaining a property in the same way the property view of time treats any supposed change. In this view, a change is a multiplicity of havings of different properties. In particular, losing a property, F, would be a pair of conditions, the condition of having F and the condition of having a property, Gy incompatible with it. But this response fails to take into account the generality of the above argument. For, according to that argument, even this pair of conditions requires a losing and a gaining of properties. Since temporal relations are not to be assumed, any pair of incompatible properties can be had by the same individual only if there is a losing and a gaining. The losing of F cannot be reduced to the pair of conditions, for losing F and gaining G keep the obtaining of the pair of conditions from being impossible. Since the objection leaves this principle untouched, it is groundless.

The second and more radical way is to adopt an ontology that does away with individuals and retains conditions in their place. The effect of this is to eliminate the entity that could lose one property and gain another. The changes appealed to in order to refute the property view are eliminated. I shall show in Chapter XI that such an ontology of conditions, events, or states of affairs is incompatible with the requirements of necessities for induction.

But now I wish to show only the more limited fact that, in the present context, such an ontology does not eliminate change. It only shifts the substratum of change from individuals with properties to what were formerly conditions of these individuals. Instead of the individual losing a property, we now have the condition ceasing to obtain. If there are individuals, then their ceasing to have a property might be said to be the only genuine change involved when we say truly that a corresponding condition ceases to obtain. But without individuals, the condition itself becomes the bearer of this change. With temporal relations assumed, there is, of course, no need to appeal to the condition’s ceasing to obtain, since incompatible conditions can then be separated by a temporal relation. But the property view assumes no temporal relations. So it would be impossible for there to be a pair of incompatible conditions unless one of them ceases to obtain.

Granted that changes are needed for time. How is this relevant to the action view of time? It is relevant only if changes are actions. So a broad conception of action is required. A component is an action of an individual if the individual “does” or “undergoes” that component, as it might do exercises or undergo modifications. Hence, an individual with an action need not be a genuine agent, for when it does or undergoes something it need not bring anything about. Losing a property is an action even when the entity that loses it does not bring about this loss. In this broad sense, action signifies what Aristotle called praxis.¹⁷ Walking and building are types of praxis, and their subjects bring about the doing of them and are thus agents. But coming to be and being moved are also types of praxis even though their subjects need not bring about the undergoing of them.

Finally, action in our broad sense need not be directed toward a result, even though there will be, as in other cases, a result of its having been done. Seeing, for example, is not seeing toward a result of seeing in the way that a linear motion is motion toward a certain place. Nonetheless, there is a result of seeing in that the brain is in a different state after it is done. But seeing qualifies as praxis simply because to see is to do something. Anything that qualifies as action in this broad sense can be a foundation of temporal relational properties.

If it is true that there is no transit from one time to another without action, then we reach an important conclusion about how natures function. To reach this conclusion let us begin by comparing non-temporal with temporal necessities. First, suppose a has F and G all at once and that its having F necessarily implies its having G. In this case, the nature of a makes its having F sufficient for its having G by being such that an individual with that nature has G given that it has F. Second, suppose a has G after it has F and that the connection is necessary. The nature does not make a’s having F sufficient for its having G just by being such that an individual with it has G given that it has F. For then the nature would only provide that a has G if it has F, but it would not provide that a has G after it has F.

What is needed is that the nature should guarantee that a changes to a G once it is F. Thus the nature must be manifested in changes. There are necessities across time only if there are changes that occur by nature under certain conditions. The growing, falling, and corroding that are components of individuals are not mere intermissions between an antecedent and a consequent that follows necessarily. They are as much by nature as it is by nature that the antecedent implies the consequent.¹⁸ If G is a property a has by nature once a is F, then the process of arriving at this result, which result is by nature, is itself a consequence of the nature of a.