20th WCP: Kants Reply to Hume in the Second Analogy

In the Second Analogy, Kant argues that we must presuppose, a priori, that each event is determined to occur by some preceding event in accordance with a causal law. Although there have been numerous interpretations of Kants argument in the Second Analogy, we have not been able to find an argument that we can show valid. The modest title of a recent article, Another Volley at Kants Reply to Hume, (1) suggests that the problem of finding a valid argument in the Second Analogy, and an adequate response to Hume, is still with us.

In this paper I will present an argument I have found in the Second Analogy for the necessity of presupposing the causal determination of each event. I will begin by briefly describing Robert Paul Wolffs interpretation of Kants argument, making clear why the argument he eventually attributed to Kant is invalid. As I develop my own interpretation, I will borrow an insight offered by Wolff. In Kants argument, our need to presuppose the causal determination of each event rests not upon our need to impose a necessary and irreversible temporal order upon representations of the states of an object, but upon our need to generate a comprehensive representation that includes a certain a priori conception of events in the world around us. Although the argument I attribute to Kant is valid, it cannot compel the Humean skeptic to accept the necessity of presupposing the causal determination of each event. I believe that Kant has not successfully responded to Hume within the Second Analogy.

Wolff claims that, according to Kant, experience of an event is possible only as we generate an objective order of representations that we can distinguish from all subjective representations within consciousness. This objective order of representations refers to successive states of an object undergoing change, while subjective representations do not refer to objects or events in the world. Wolff is especially concerned with our ability to distinguish an objective order from those representations that arise within original apprehension prior to any synthesizing activity of the understanding. (2) We can generate such an objective order, Wolff continues, only through the synthetic reworking of the manifold of representations. We must reproduce and combine representations in imagination, that is, in accordance with a rule that specifies the order in which they are to be reproduced. The temporal order we impose upon representations as we reproduce them must be necessary, Wolff emphasizes, in the logical sense that comes with being in accordance with a rule. We must apply a rule, Wolff claims, that specifies a temporal order of the states of an object. Through specifying a succession of states, the rule specifies the temporal order in which we are to reproduce representations of the states. Wolff interprets Kant as straightforwardly identifying an object with our representations of the object. [T]he object cannot be a distinct entity from the representations of it, Wolff declares. (3) An event is considered identical to ordered, rule-governed representations of the constituent states. A rule that specifies a temporal order of states specifies as well a temporal order of representations of the states, for an event is here no more than ordered, rule-governed representations. As we experience an event we must presuppose that the event arises in accordance with a rule that specifies a succession of states.

The argument proceeds by adding the claim that the schema of cause and effect is our concept of succession in accordance with a rule. We are able to generate a distinguishable, objective order of representations, then, and so experience an event, only as we apply the schema of cause and effect in our thought of the event we experience, presupposing that this event arises in accordance with a rule that specifies a succession of states. Kants conclusion is that we must, in fact, apply the schema of cause and effect in our thought of all events, presupposing that each event arises in accordance with such a rule.

In Wolffs account, the rule we apply as we reproduce representations must have a special form. This rule must do more than specify a temporal order of states. It must specify a temporal order of events. This rule must assert that an occurrence of an event B follows an occurrence of an event A in specified standing conditions. This rule specifies the temporal order of the constituent states of event B, and asserts that this succession of states follows an occurrence of event A in specified conditions. A rule of this form is a causal law. According to Wolff, to conceive a regular succession of events A and B in accordance with a rule of this form is to conceive A as the cause of B and B as the effect of A. (4) For Wolff, causal determination is merely the regular occurrence of B given A under this special sort of rule. In our experience of a ship sailing downstream, for example, as we reproduce representations we must apply a rule that specifies a succession of states: the ship located downstream follows the ship located further up the stream. Our rule, a causal law, specifies the temporal order of the states of the ship and so the temporal order of representations of these states. According to our rule, this change of state regularly follows an event A in specified standing conditions.

For a given event, we may not know exactly which event is the antecedent event A or which standing conditions must be present. In this case, we must presuppose the existence of something, Kant writes, upon which our event follows invariably, that is, in accordance with a rule, a causal law (A198/B243). (5) We must presuppose that the event we experience regularly follows some event or other in accordance with a causal law. Although we do not know many details of the appropriate causal law, as we experience an event we must reproduce representations in accordance with the part we do know, the part that specifies a succession of states. According to Kant, our experience of an event is possible only as we presuppose that each event arises in accordance with a causal law.

Wolff claims that the schema of cause and effect is our concept of rule-governed succession, but this is too vague a characterization. There are different kinds of rules. Within Wolffs interpretation this schema is actually our concept of succession in accordance with a causal law, a special sort of rule according to which an event B regularly follows an event A in specified standing conditions. Kant is arguing that we must apply this concept in our thought of all events.

In Wolffs interpretation, Kants reply to Hume focuses upon unsynthesized representations within original apprehension. Hume seems to have been unaware, Wolff claims, that there is a difference between the order in which we perceive events or objects and the objective order which we subsequently ascribe to them. (6) We must move beyond original apprehension to generate a representation of an event we experience. We must impose upon representations of the states a temporal order that is rule-governed and so necessary. According to Wolff, Kant adds that the schema of cause and effect is our concept of rule-governed succession. Within Wolffs interpretation, however, cause and effect is actually a richer concept than this. We must apply a special sort of rule that asserts a succession of events in specified standing conditions.

We should question whether the argument Wolff has provided so far can possibly be valid. Can Kant argue validly from our need to impose a rule-governed temporal order upon representations to the necessity of applying the schema of cause and effect, conceiving that each event arises in accordance with a causal law?

The most pressing difficulty with the above argument is that it does not compel us to apply the schema of cause and effect in our thought of any event. We can impose upon our representations a rule-governed temporal order by bringing to bear a weaker concept. According to Kant, in order to experience an event we must formulate a judgment in which we classify the event as being of a certain kind. We must conceive that a ship is sailing downstream, for example, or that a man is growing old. We must apply an empirical concept in our thought of the event. Kant considers an empirical concept a rule for reproducing representations. He provides as an example our empirical concept of body, which directs our reproduction of representations of extension, impenetrability, shape, etc. (A106). This concept does not specify the order in which we are to reproduce these representations, however. Our empirical concept of a house does not specify an order in which we are to reproduce representations of the parts of a house. Our empirical concept of sailing downstream does specify an order of representations: our representation of the ship downstream follows our representation of the ship further up the stream. With this concept, we impose upon representations a necessary, rule-governed temporal order, which Wolff claims is the mark of objectivity. We need conceive only a single instance of a succession of states. We could conceive that the ship sails upstream and reverse the order of our representations, but this is to apply a different concept. (7) We are not compelled to presuppose that this event or any other arises in accordance with a causal law. We are not compelled, then, to apply the schema of cause and effect. In fact, we are not compelled to presuppose that any event arises in accordance with any rule that specifies a succession of states. Kants argument is therefore invalid.

The challenge Wolff faces is to show how our need to impose a rule-governed temporal order upon representations compels us to presuppose that any change of state arises in accordance with a causal law. (8)

In fact, Wolff suggests an essential role for the schema of cause and effect.

What, in general, distinguishes an actual event in the real world from a mere fancy of imagination? Putting to one side all considerations of force and vivacity of perceptions, as Hume would say, the mark of the real is its unalterability, its demand that we adjust to it, rather than it to us. ... we must always so represent it. Thus, the objective reality of the temporal succession of A and B is expressed by a necessary and universal rule for their representation. (9)

According to Wolff, Kant assumes that experience of an event in the world is possible only if we are able to distinguish this experience from a mere fancy of the imagination, from illusions and dreams. As we experience an event, we must reproduce our representations of the states in accordance with a necessary and universal rule. We must apply a rule that specifies a succession of states of an object. We must therefore apply the schema of cause and effect, which is, Wolff says again, our concept of rule-governed succession. Wolff adds that with this schema we conceive that the event we experience arises within a wider succession of events in which each event is causally determined by some preceding event. As we experience an event B under this schema, we place B into the objective sequence ABC. We assert that B is the effect of A and the cause of C. (10)

What, in general, Wolff asks, distinguishes our experience of an event in the world from mere fantasy, from illusion and dreams? Wolffs insight is that, according to Kant, this distinction is possible only as we presuppose the causal determination of each event. As we experience an event in the world, we must place it within the proper, rule-governed succession of events. It is our thought of causal determination that enables us to distinguish reality from mere fantasy.

Representations within illusions and dreams are, Wolff says, subjective representations within consciousness, as are unsynthesized representations. According to Wolff, the problem of how we distinguish experience of an event in the world from mere fantasy is another aspect of the general problem of how we generate an objective order of representations that is distinguishable from all subjective representations within consciousness. Actually, Wolff sees that Kants problem is even broader than this. Another insight Wolff offers is that Kant is concerned with our ability to generate an objective order that is generally distinguishable from all other successive representations, both subjective and objective, including representations that are objective yet refer to coexisting objects. (11) Within Wolffs interpretation, our need to generate such a distinguishable, objective order compels us to apply a rule as we reproduce representations, and so we must apply our concept of rule-governed succession, the schema of cause and effect. Even as he brings in consideration of illusions and dreams, properly broadening the problem Kant addressed, Wolff rests our need to apply the schema of cause and effect upon our need to impose a rule-governed temporal order upon representations. Kant did not argue in this way, however. Wolff conflates arguments that should be separated: the argument for our need to apply merely a rule as we reproduce representations, and the argument for our need to apply the schema of cause and effect.

Kant indicates the general problem he attempts to solve in the Second Analogy when he writes that we must somehow distinguish the apprehension of an event from other apprehensions (A192/B237). He is not limiting his problem to how we distinguish experience of an event from experience of coexisting objects, although this distinction is featured in the following passage.

Every apprehension of an event is therefore a perception that follows upon another perception. But since, as I have above illustrated by reference to the appearance of a house, this likewise happens in all synthesis of apprehension, the apprehension of an event is not yet distinguished from other apprehensions. But as I also note, in an appearance that contains a happening B can be apprehended only as following upon A; the perception of A cannot follow upon B but only precede it. For instance, I see a ship move down stream ... . The order in which the perceptions succeed one another in apprehension is in this instance determined, and to this order apprehension is bound down ... . But in the perception of an event there is always a rule that makes the order in which the perceptions follow upon one another a necessary order. (A192-3/B237-8).

We must be able to distinguish experience of an event from other apprehensions, he writes, for example, our experience of the coexisting parts of a house. As we experience an event, we must conceive that the temporal order of our representations of the states is bound down and necessary in the sense of being rule-governed.

Thus only can I be justified in asserting, not merely of my apprehension, but of appearance itself, that a succession is to be met with in it. ... I cannot arrange the apprehension otherwise than in this very succession. In conformity with such a rule there must lie in that which precedes an event the condition of a rule according to which this event invariably and necessarily follows. (A193/B239)

In conformity with such a rule, referring to the rule we apply as we reproduce representations, an event must invariably and necessarily follow another event that serves as the condition of a rule.

Kant does not wish to suggest that as we reproduce representations we must apply, as our rule specifying an order of representations, a causal law. We may apply a more minimal rule, perhaps simply an empirical concept. Rather than say, In conformity with such a rule, Kant should say, In conformity with such a rule-governed order of representations. (12) He wishes to say that corresponding to a rule-governed order of representations referring to a succession of states, we must presuppose that each event follows some event or other (the condition of a rule) in accordance with a causal law. His view is that successive representations genuinely refer to an event in the world only if we place the event within the proper, rule-governed succession of events.

As we experience an event, we must generate a comprehensive representation that consists of two parts. We must impose a necessary, rule-governed temporal order upon our representations of the states. As Henry Allison emphasizes, Kant believes that our need to distinguish experience of an event from experience of coexisting objects compels us to impose a rule-governed temporal order upon representations. (13) In contrast to our experience of a house, Kant writes, in our experience of an event there is always a rule that makes the order in which the perceptions follow upon one another a necessary order. Yet our application of a rule as we reproduce representations leaves open the possibility that our experience, of a ship sailing downstream perhaps, is actually mere fantasy. Even in illusions and dreams we can apply a rule that directs our reproduction of representations. We can apply our concept of sailing downstream, for example, ordering our representations in accordance with this rule, and our fantasy is still fantasy. We can experience an event in the world, Kant believes, distinguishing such experience from experience of coexisting objects, only within the context of presupposing that each event is causally determined. This is the second part of the comprehensive representation we must achieve. We must place an event we experience within a succession in which each event arises in accordance with a causal law. We must, therefore, apply the schema of cause and effect in our thought of all events. Kant claims that we are compelled to apply the schema of cause and effect by our need to distinguish experience of an event in the world from mere fantasy. He states that if we were not to presuppose the causal determination of events, we would have only a play of representations, relating to no object (A194/B239). In a later passage, Kant adds that if we were not to presuppose the causal determination of events, we would be left with a merely subjective play of ... fancy, perhaps a mere dream (A201/B247).

Kants argument begins, then, from our need to generate an objective order of representations that is generally distinguishable from all other successive representations, including those that are objective yet refer to coexisting objects, and those that arise within illusions and dreams. Our need to distinguish experience of an event from experience of coexisting objects compels us to impose a necessary, rule-governed temporal order upon representations. Yet this must occur within the context of presupposing that each event arises in accordance with a causal law, for this presupposition is necessary if we are to distinguish experience of an event in the world from mere fantasy. A causal law asserts that an occurrence of an event B follows an occurrence of an event A in specified standing conditions. This is a valid argument for the necessity of applying the schema of cause and effect in our thought of all events, given the assumption that we are able to distinguish an objective order of representations from subjective representations within illusions and dreams only as we place an event we experience within the proper, rule-governed succession. This argument does not rest our need to presuppose causal determination upon our need to impose a rule-governed temporal order upon representations, although Kant believes we must impose such a necessary order upon representations.

Lewis White Beck has argued that Kant does not use the categories in general to separate experience of the world from illusions and dreams. Rather, the categories are used to separate consciousness or awareness of objects, including subjective objects in illusions and dreams, from dumbly facing chaos without even knowing it--less even than a dream. (14) According to Beck, at least some categories are obviously applied in illusions and dreams. The mathematical categories are applied, for example, whenever we represent the sun in imagination, for the brightness we represent has intensive magnitude. Generally speaking, our consciousness or awareness of objects, even of subjective objects within illusions and dreams, requires our application of categories as we formulate judgments. In the passages I have discussed above, Kant is not saying that without the schema of cause and effect we would be left with no judgments whatsoever and no application of categories. Nor is he saying that we would be left without at least some schemata in our thought of subjective objects. We would be left, however, without our thought that each event arises in accordance with a causal law. We would be left with illusions and dreams, Kant says, but not with a total chaos of which we cannot be conscious or aware.

A difficulty with the argument I have attributed to Kant is that, of course, we are under no obligation to accept it. Hume would refuse to analyze our ability to distinguish experience of an event in the world from illusions and dreams in terms of our thought that each event arises in accordance with a causal law. At issue for Hume is our warrant for conceiving that any event arises in accordance with such a rule, since experience alone cannot definitely establish the truth of such a rule. Hume is not compelled to adopt a crucial premise of this argument and accept this easily the conclusion that we must apply the schema of cause and effect. As Wolff expresses it, Kant assumes, against Hume, that force and vivacity of perceptions is not sufficient for distinguishing experience of an event in the world from mere fantasy. We must turn to our thought that each event in the world is causally determined. Only if Hume agreed with Kants view on this would he be compelled to accept the conclusion of Kants argument. As it stands, this argument is not successful against Hume.

There have been numerous interpretations of Kants argument for the necessity of presupposing causal determination. I have dealt with several of these in earlier papers. I have come to believe that in the Second Analogy Kant did not successfully respond to Hume.