<< . .

. 16
( : 30)



. . >>

types rather than primarily between event-tokens. He does not so
obviously take regularity in the sense of repetition to be a necessary
condition for causal relations. The general causal principle entails
lawlike connections between event-types of the kind that support
counterfactuals, but it does not entail laws that cover recurrent event-
types of the kind involved in a strict regularity theory of causation. Thus,
although the general causal principle requires the existence of causal
laws in nature that make changes recognizable, it does not obviously
entail that we are actually capable of determining what those laws are.
There is nothing in the argument of the Second Analogy, according
to Allison, which shows that particular causal laws, even if they must
exist, must be knowable.± This claim deserves to be contested. For if
particular causal laws were unknowable, then we would not be able to
know whether a change in a particular substance had occurred. Since
the principle that changes must be empirically signi¬cant provides the
basis for the objective necessity that Kant assigns to causal relations, he
cannot allow for the possibility of causal laws that have no empirical
signi¬cance.
The Second Analogy does not itself provide any criteria for ident-
ifying the speci¬c causal laws in terms of which events are actually to be
ordered in time. However, the argument of the A-Deduction directly
links a defense of the general causal principle to the existence of
recognizable uniformities in nature. Kant states quite explicitly that an
empirical rule of association must always be presupposed when one says
that there is always some earlier event that a later event succeeds (
±±“±±). Not only does association generally presuppose the recurrence
of events, but it is di¬cult indeed to make sense of an empirical rule of
association that is not based on the recurrence of events in experience.
Kant makes this point in a context in which he stresses the indepen-
dence of causal necessity from mere inductive generalizations.
The possibility that we might be able to provide a justi¬cation of the
general causal principle without being able to justify the legitimacy of
particular causal laws has suggested to a number of interpreters that the
necessity of particular causal laws might be purely a function of their
position in a systematic uni¬cation of nature.±µ Particular causal judg-
ments have their tentative standing as statements of causal laws because
they occur within the best systematic uni¬cation of experience available
to us. But they have their status as statements of causal laws because they
are instances of the general causal principle. All speci¬c causal laws are
± Kant and the demands of self-consciousness
metaphysical, because they cannot be derived from the transcendental
conditions governing experience, but presuppose some metaphysical
assumption about the world in which we exist (Ak. , p. ±±). They
derive their status as causal laws from the presumption that they provide
our best understanding of the necessary relations between events in
virtue of which we are able to identify and reidentify events in time.

 °  ©   © « ®  · ¬ ¤ §   ¦ °   ©  µ ¬     µ  ¬ ¬ ·  ?
Kant™s attitude towards how we know speci¬c causal laws is somewhat
ambiguous. On the one hand, he certainly wants to give an important
place to empirical investigation. On the other hand, laws are only
genuine laws for Kant to the extent to which they can be known to apply
to all individuals in their domain with necessity and strict generality,
otherwise they amount to mere empirical generalizations. Such empiri-
cal generalizations as ˜˜all swans are white™™ may have the provisional
status of laws. But as soon as an appropriate counterexample is dis-
covered, such as a black swan, their claims to be laws must be given up.
Given Kant™s evident interest in the importance of empirical knowl-
edge in science, some commentators have argued that Kant thought
that speci¬c natural laws cannot be known a priori. Thus Guyer claims
that ˜˜Kant speci¬cally denies that individual causal laws are known a
priori (see  ±“±·/ ±“, or more generally  ±).™™± The pas-
sages cited by Guyer do not support his assertion.  ± does not make
reference to causation or causal laws at all. Kant does note at  ± that
the judgment ˜˜bodies are heavy™™ is empirical and contingent even
though the verb to be in this judgment expresses a necessity that is based
on the transcendental unity of apperception. This suggests that even
empirical generalizations can purport to be laws only insofar as they are
supported by a priori principles.  ±“ ¬rst notes that Kant™s idea
that we must presuppose a necessary and strictly general causal prin-
ciple seems to contradict our practice of making generalizations con-
cerning temporal sequence from di¬erent experiences of sequence, but
he then goes on to claim that such a strategy in isolation would give us
only a kind of imagined universality and necessity (i.e. Humean general-
ity and necessity). As opposed to this view, Kant claims that causation is,
like other concepts, a priori in that we can only derive it from experience
because we have already constituted experience in accordance with it.
Thus, if anything, the passage in question would seem to support the
view Guyer rejects. For it suggests that we can only derive a necessity
±
Causal laws
and universality claim from nature by antecedently investing nature
with that very necessity and universality.
There is nothing in the Critique that suggests that causal laws could be
knowable purely a priori, although the Critique does seem to require
some a priori element to support an empirical generalization™s claim to
be a bona ¬de law, which Kant takes to be characterized by necessity
and universality. At  ±µn in the Deduction, Kant does say that
particular laws cannot be completely derived from the categories be-
cause they concern empirically determined appearances. But then he
asserts that experience is necessary in order to be acquainted (kennen)
with these appearances. The fact that we need experience to become
acquainted with certain regularities does not preclude the laws that we
postulate from being a priori. However, this fact does suggest that we
would ¬rst have to become acquainted empirically with regularities of
the kind involved in causal laws. Kant can consistently argue that, even
though the necessity and strict generality of laws is something that one
would know independently of experience, it is still not possible for us to
demonstrate the necessity of such necessary and strictly universal laws.
In other words, such rules governing objects of experience would be
necessary, but not necessarily necessary.
Now there are passages in Kant™s later work that encourage the view
that he came to regard all causal laws as knowable a priori:

Even the rules of consistent appearances are only called natural laws (for
instance the mechanical ones), if one knows them either really a priori or at
least assumes (as in the case of the chemical ones), that they would be known a
priori based on objective grounds, if our insight went deep enough. (Critique of
Practical Reason, Ak. , p. )

Despite some of Kant™s claims about the knowability of causal laws, no
one has argued to my knowledge that he was committed to the deriva-
bility of all causal laws from the category of causation. It has been
suggested by Friedman that the basic principles of Newtonian mechan-
ics are thought by Kant to be a priori. According to Friedman, Kant
also thought that all more speci¬c physical laws would, in principle, be
derivable a priori.±· Even if Kant was attracted to such a view, he
realized that the project of articulating a set of fundamental physical
laws would have to count as a regulative principle of inquiry, rather than
a goal that was already completed. It thus seems that the necessity and
universality of laws would ultimately depend on the necessity for us of
±µ° Kant and the demands of self-consciousness
projecting a certain systematic unity of nature under laws if we are to
regard our judgments about occurrences in nature to be justi¬able and,
hence, empirical truth to be attainable at all:
One can also not say that it [reason] has previously taken this unity according
to principles of reason from the contingent character of nature. For the law of
reason to look for it [the unity] is necessary, since without it we would have no
reason, without that no connected use of the understanding, and without that
no su¬cient mark of empirical truth, and therefore in respect to the latter we
must presuppose the systematic unity of nature as objectively valid and necess-
ary. ( µ±/ ·)

Kant™s attempt to ground particular laws of nature in regulative syn-
thetic principles a priori is based on the interest understanding has in
postulating an intelligible order of natural laws. In the Critique of Judg-
ment, Kant acknowledges that there are an in¬nite multiplicity of causal
laws that we cannot know a priori, although he also maintains that the
necessity involved in speci¬c causal laws is a necessity that we must think
of as a priori. These two statements are consistent to the extent that one
takes causal necessity to be a priori while allowing that we may be
incapable of determining a priori what the particular form of that
necessity may be in the case of many individual laws. Indeed in the case
of the laws of biology, Kant thinks that, in principle, it is impossible for
our insight to go deep enough to understand the connection between
biological laws and the mechanical laws of physics:
It is quite certain that we can never get a su¬cient knowledge of organized
beings and their inner possibility, much less explain them, according to mere
mechanical principles of nature. So certain is it, that we may con¬dently assert
that it is absurd for human beings to make any such attempt, or to hope that
maybe another Newton will some day arise to make intelligible to us even the
production of a blade of grass according to natural laws that no design has
ordered. Such insight we absolutely deny to humanity. (Ak. , p. °°)

Elsewhere in the Critique of Judgment, Kant also makes the point that
organisms resist a purely physicalistic explanation. In this passage, he
connects the point directly to the status of di¬erent kinds of causal
explanation:
It is utterly impossible for human reason (even for any ¬nite reason that might
resemble ours in quality, however much it may surpass it in degree), to hope to
understand the production even of a blade of grass from merely mechanical
causes. For the possibility of such an object, the teleological connection of
±µ±
Causal laws
causes and e¬ects is quite indispensable for judgment, and even if only to study
it under the guidance of experience. (Ak. , pp. °“±°)

Causal laws involve more than generalizations based on accidental
regularities. They involve nomic necessities. But we can only know such
nomic necessities, according to Kant, when we understand the kind of
thing that we are investigating. We must know the dispositions of things
belonging to a certain natural kind. That is, we know lawlike necessities
when we know what things of a certain kind do under various physically
possible circumstances. And we have no way of understanding such
dispositions independently of what a thing of a certain kind would do
under di¬erent circumstances. To understand the behavior of a certain
kind of thing under di¬erent circumstances we need to be able to
formulate a counterfactual conditional of the kind supported by causal
laws.
Kant argues that without the supposition that nature divides into
natural kinds in a manner susceptible to explanation in terms of the
concepts we have, such knowledge of dispositional properties will not be
forthcoming. That nature has a systematic unity susceptible to the
formulation of hypotheses leading to true theories is an ideal guiding the
formulation of empirical concepts of dispositional properties and lawlike
causal connection. This transcendental postulate of the systematic unity
of nature allows the truth of theories to be tested by a combination of
internal coherence and empirical adequacy.
The inductive inference from singular causal judgments to corre-
sponding causal laws must be supported by regulative principles of
reason. If the lawlikeness of causal judgments must be supported by
regulative principles which are antinomial, if treated as objective laws of
unrestricted scope, then this will allow a role in inquiry for probabilistic
laws. Kant certainly believes that laws make an implicit claim to unre-
stricted generality. But he also believes that this claim cannot be made
good by us. Although we must strive to unify di¬erent causal laws under
common principles, Kant believes that there are certain inherent limita-
tions to our capacity to carry out this enterprise. Such an interpretation
is suggested by the introductions to the Critique of Judgment.


   ¬ ¬   µ  ¬ ¬  ·  ¤    © ® ©   © ?
Kant™s argument from time-determination to the a priori validity of the
causal principle is compatible with the existence of indeterministic
±µ Kant and the demands of self-consciousness
causal laws, as well as indeterministic applications of causal laws. Both
Melnick and Guyer argue that the existence of statistical laws of nature
does not threaten the universality and necessity that Kant attributes to
causal laws.± This is clearly correct, since he countenances the existence
of indeterministic natural laws:
Whatever concept one may hold, from a metaphysical point of view, concern-
ing the freedom of the will, certainly its appearances, which are human actions,
like every other natural event are determined according to universal laws. Since
the free will of human beings has obvious in¬‚uence on marriage, births, and
deaths, they seem to be subject to no rule by which the numbers of them could
be calculated in advance. Yet the annual tables of them in the major countries
prove that they occur according to natural laws as stable as the unstable
weather, where we cannot predict individual events, but which, in the large,
maintains the growth of plants, the ¬‚ow of rivers, and other natural events in an
unbroken, uniform course. (˜˜Idea for a Universal History from a Cosmopoli-
tan Point of View™™ [±·], Ak. ©©©, p. ±·)
Kant regards probability as something subjective, roughly the plausi-
bility of a certain hypothesis relative to a certain set of evidence, rather
than as an objective property of nature, i.e. chance: ˜˜Probability is to be
understood as a holding true on the basis of insu¬cient reasons, which
however, have a greater relation to the su¬cient ones than the grounds
of the opposite™™ (Logic, Ak. ©, p. ±). Given this view of probability there
is no reason to think that natural laws cannot be probabilistic. Our
judgments about temporal relations and their correlative causal rela-
tions are uncertain, because they are subject to error, and so we can
expect most of the laws that we formulate concerning causal relations to
be inherently probabilistic. It is true that Kant seems to have thought
that the fundamental laws of nature would be inherently unprobabilis-
tic, but this is only because he thought they could be known a priori, and
he seems to have thought they could be known a priori because he was
convinced that we could know a priori that objects have completely
determinate trajectories through space and time. This is a plausible
assumption. But it is in con¬‚ict with the best theory of physical objects
that we now have available to us. In quantum mechanics, the classical
notion of a trajectory with a completely determinate position and
momentum for a particle breaks down at the microphysical level. While
quantum mechanics may eventually be replaced by a more fundamen-
tal theory that does involve the sharp trajectories of the kind Kant
postulates, it is at least equally likely that this indeterminacy will remain
in place in future physical theory.
±µ
Causal laws

   °  ©  © ¬ ©    ¦ ®   µ   ¬   µ    ©  ®  ® ¤
 µ  © ® ¦     ®
At the level relevant to the explanation of human action, even Kant
seems to think that the causal laws that apply are, at best, probabilistic.
At least this is suggested by the passage from the essay on history cited
above. Allowing for probabilistic causal laws governing actions helps
one to see how he could argue that our actions might all have causal
covering laws, and yet be such that they allow for a kind of causation by
reason that is independent of natural causal laws. Kant™s formulations of
the causal principle in both the ¬rst and second editions suggest that
they are designed to provide leeway with respect to the problem of the
compatibility of non-physical causes and physical causal determination.
In the ¬rst edition of the Critique ( ±), he asserts as a ˜˜principle of
generation™™ that everything which happens or comes to be presupposes
something which it is the consequence of according to a rule. This does
not entail the existence of a temporal cause for every occurrence,
although that could be understood to be his meaning. It certainly does
not entail that we can have knowledge of what the speci¬c cause of an
event is. Di¬erent causes may be independently su¬cient for bringing
about a certain event-type, and hence no cause need be a necessary
condition for the occurrence of a given event. In the second edition at 
, the principle is restated in somewhat tighter terms as a principle of
temporal succession according to the law of causality. All changes are
now said to occur according to the connection of cause and e¬ect.
Again, every change must be describable in terms which allow one to
identify a cause and an e¬ect. Yet the formulae are compatible with an
action being uncaused under a more fundamental description of that
action as long as there is a temporal description of that action under
which it is caused.
In causation from freedom, or from reason, an intellectual state
causes a physical state but is not in turn caused by any antecedent
physical state: ˜˜if reason has causality in respect to appearances, then it
is a faculty through which the sensible condition of an empirical series of
e¬ects ¬rst begins. For the condition that lies in reason is not sensible
and does not itself begin™™ ( µµ/ µ°). The uncaused event (the event
which is the e¬ect of causation through reason) is uncaused under its
purely reason-based description, and caused under its description as an
event belonging to a determinate temporal order. This leads to the
thesis that incompatibilism and compatibilism are compatible so long as
±µ Kant and the demands of self-consciousness
we distinguish the two fundamentally di¬erent aspects of how things
are, namely, things as they must appear to us, and things as they are
grasped by reason independently of the way they must appear to us.±
Kant™s peculiar form of compatibilism is clearly consistent with the
idea that every instance of singular causation entails the existence of a
causal law which is true of it. Things would be more di¬cult if Kant
were committed to the stronger claim that each individual cause entails
a particular causal law. Given that individual causes do not entail the
existence of particular causal laws, singular causation can, in some cases
at least, consist in the temporal appearance of causation based on
reasons for choice that are independent of the agent™s causal history.
The thesis that every physical event has a su¬cient physical cause is
compatible with the existence of non-physical events that serve as
overdetermining causes of physical events. That is, it allows for the
possibility that the occurrence of a physical event could equally well be
explained in terms of a non-physical cause.
Interpreters who have been strongly in¬‚uenced by Donald David-
son™s account of free agency have argued that the key to Kant™s account
of the compatibility of free will and causal determination is that reasons
serving as causes, or what Kant calls intelligible causes, provide ˜˜ration-
alizations™™ for what we do, but are not subject to causal laws.° On the
view in question, mental events are anomalous, but token“token ident-
ical with physical events that are subject to causal laws. Initially, it would
seem that this view of mental causation threatens to make all mental
events and intelligible causes into mere epiphenomena of physical
events. However, this is not so, for the view in question treats causation
as an extensional relation. While mental events are only subject to
causal laws under a physical description, they are causally e¬cacious in
virtue of their nature as events. This, however, raises the following
obvious problem for such a view of mental causation. On the view in
question, reasons will be causally e¬cacious insofar as they are part of a
person™s psychological make-up. Now, insofar as these reasons are
causally e¬cacious they will also themselves be subject to causal deter-
mination by antecedent causes. It is true that we will not be able to come
up with causal laws governing such determination under the mental-
event description of the event that instantiates a reason. However, such
a mental event will belong to the causal order as an event, and,
moreover, be token“token identical with a physical event that is subject
to causal laws and thus, in principle, subject to deterministic causal laws.
So it is hard to see how any progress has been made in rescuing Kant™s
±µµ
Causal laws
intuitions about the compatibility of deterministic physical explanation
with indeterministic reason-based behavior.
Kant argues that, as long as we are determined by causes that provide
the conditions under which we are able to determine the relations
between events at di¬erent times, we have only ˜˜psychological and
comparative freedom™™ which would be nothing better than the ˜˜free-
dom of a turnspit, which also, once it is cranked up, goes through its
motion on its own™™ (Critique of Practical Reason, Ak. ©, p. ·). The
anomalous monism interpretation of mental-event causation does no-
thing to help Kant out of the dilemma that psychological freedom seems
to be nothing but the freedom of a turnspit or a spiritual automaton.
Unlike the anomalous monist, Kant insists that reason has to be genu-
inely causal in its own right, and not merely in virtue of being realized in
the form of the psychological states of some agent, if freedom is to be
saved. Now, from the second- or third-person perspective, Kant argues
that we must attribute to human beings an empirical character that
provides the basis for their choices, where such choices are themselves to
be understood as a causal power of reason. But, since the empirical
character is itself something that allows of a causal explanation, there
can be no freedom:

[I]f we could research all appearances of choice to their ultimate ground, then
there would not be a single human action that we could not predict with
certainty and could not know with necessity from its preceding conditions. In
respect to this empirical character there is therefore no freedom, and we can
alone consider human beings according to this character when we merely
observe, and physiologically investigate the moving causes of this action, as
we do in anthropology. ( µµ°/ µ·)

The important point to note is the conditional character of Kant™s
claim that all human actions would turn out to be completely predict-
able. We would have to be able to seek out the ultimate grounds of
choice in a complete understanding of the person™s empirical character
and the other relevant causal circumstances, if we were to be able to
explain a person™s behavior in causal terms. Such a causal explanation is
a regulative ideal governed by a principle of reason in our observations
of ourselves and others ( µµ/ µ). But this does not mean that, even
in principle, we could ever have su¬cient knowledge to fully carry out
the required causal explanation.
A comprehensive explanation of the world is something that we are
forced to strive for as rational beings, but it is also something to distrust
±µ Kant and the demands of self-consciousness
because we are inherently incapable of realizing it given our ¬nitude.
Our reason brings with it the transcendental illusion that, in principle,
we could attain such a comprehensive explanation. This illusion that
one can provide such a comprehensive explanation is at the basis of his
critique of transcendental realism.
Although it looks merely like a chain of causes here that does not allow an
absolute totality in the regress to its conditions, this concern does not stop us at all;
for it is already dealt with in the general consideration of the antinomy of
reason. If we give way to the deception of transcendental realism: then neither
nature nor freedom remains. ( µ/ µ·±)

The transcendental realist thinks that, in principle, he or she can
presume to have all objects as they must appear to us in experience
available to him or her, and then goes on to make inferences about what
the world is independently of the way it must appear to us in our
experience on the basis of that presumed closure in our knowledge of
objects of experience. Kant maintains that the transcendental realist
view cannot account for our knowledge of the necessity that laws of
nature are supposed to have, and thus cannot account for nature. This
might seem to leave the transcendental realist in a better position to deal
with the problem of freedom. However, randomness is no more com-
patible with free agency than is complete causal necessity.
In contrast to the transcendental realist story in question, Kant argues
that because our e¬orts to identify the physical causes not only of what
we do, but of anything at all, are always going to be relatively incom-
plete, there is still always space for an alternative account or description
of our action under which the occurrence of an action is explained by
the rational principles that we adhere to, rather than by appeal to past
facts about our causal history or about the causal history of the world.
Causal overdetermination is su¬cient to provide an alternative expla-
nation for what the cause of a given event is.
Given the inherent incompleteness of our knowledge of causes, there
will always be room both for an account of what we did that is
determined by antecedent causal conditions, and an account that treats
our adoption of the reasons for action that we adopt as genuinely
independent of our past causal history and thus as inherently indeter-
ministic. Kant insists that it is only in the latter case that we can
legitimately regard the reasons for our action as ones that initiate a

<< . .

. 16
( : 30)



. . >>