See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/236817423

Hume and the Limits of Reason

Article in Hume Studies · April 1996

DOI: 10.1353/hms.2011.0152

CITATIONS READS
7 872

1 author:

Michael P. Lynch
University of Connecticut
66 PUBLICATIONS 1,262 CITATIONS

SEE PROFILE

All content following this page was uploaded by Michael P. Lynch on 26 August 2014.

The user has requested enhancement of the downloaded file.

Hume and the Limits of Reason
Michael P. Lynch
Hume Studies Volume XXII, Number 1 (April, 1996) 89-104.
Your use of the HUME STUDIES archive indicates your acceptance of HUME
STUDIES’ Terms and Conditions of Use, available at
http://www.humesociety.org/hs/about/terms.html.
HUME STUDIES’ Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue
of a journal or multiple copies of articles, and
you may use content in the HUME STUDIES archive only for your personal,
non-commercial use.
Each copy of any part of a HUME STUDIES transmission must contain the
same copyright notice that appears on the screen or printed page of such
transmission.
For more information on HUME STUDIES contact humestudies­
[email protected]

http://www.humesociety.org/hs/
HUMESTUDIFS
Volume XXII, Number 1, April 1996,pp. 89-104

Hume and the Limits of Reason

MICHAEL P. LYNCH

In all demonstrative sciences the rules are certain and infallible; but
when we apply them, our fallible and uncertain faculties are very apt
to depart from them, and fall into error.’

So begins Treatise I iv 1, “Of Scepticism with Regard to Reason.’’ The rules

of reason may be perfect, but we are not. From this truism Hume cultivates a
notoriously obscure argument to the effect that we should never trust our
reasoning. In characteristic fashion, Hume considers the argument sound but
unpersuasive. He believes that no one who reflects on it will lose faith in their
reason and sets out to explain why. In this essay, I am concerned to make two
main points. First, the argument is more successful than often thought; and
second, reflecting on Hume’s explanation for its lack of force illuminates the
depth of his naturalism and contains an important insight into the limits of
human reason.2
Hume actually presents two distinct arguments in “Of Scepticism with
Regard to R e a ~ o n . “The
~ arguments are distinct insofar as they are aimed at
different conclusions; they are related in that they share a premise concerning
our natural fallibility as reasoning agents. The first argument (T 181) is meant
to show that any belief formed in the “demonstrative sciences“-any a priori
belief about (say) mathematics-cannot be held with certainty. At best, such
beliefs can be taken as probably true. This argument can be sketched as fol-
l o w ~ In
. ~ performing any set of calculations, no matter how simple, we are

Michael P. Lynch is at the Department of Philosophy and Religion, University of

Mississippi, University MS 38677 USA. email: [email protected]
90 Michael P. Lynch

susceptible to error. This we know from past experience. We will, of course, be

more confident in our conclusions if we recheck our calculations, or allow
others to check them for us. But while these methods can increase the
subjective probability that we are right, they can never increase it all the way
to 1. For the possibility that we have made a mistake always remains.
Therefore, we should never be completely certain of any of our beliefs, even
of those concerning rudimentary mathematics, and hence “all knowledge
degenerates into probability.”
Using the above line of reasoning as a stepping-stone, Hume then presents
a second, more shocking argument. Here, the conclusion is a strong
scepticism: reflection on our natural fallibility shows us that we shouldn’t
even trust our reasoning in everyday life; if we follow the argument, we will
see that we actually have n o justified beliefs at all. Hume thinks the arguments
of each stage are epistemically related-accepting the first argument should
lead one to accept the second. In this essay, I will only be concerned with the
second argument. Thus, when I refer to Hume’s “sceptical argument” I am to
be understood as referring to the stronger, more radically sceptical
hypothesis.
The paper is broken into two major sections; each corresponding to a
question. The first question is: What is the sceptical argument and does it
work? The second is: What does Hume take the sceptical argument to show,
or: what is Hume’s point in raising the sceptical argument? One might expect
that the answer to this second sort of question would already be contained in
an answer to the first sort. But as is often the case, things aren’t simple where
Hume is concerned. As I’ve already noted in the introduction, Hume doesn’t
think anyone will-or can, for that matter-believe the sceptic’s conclusion,
even if her argument for that conclusion is sound. Therefore, in what follows,
we must keep these two issues in mind; we must separate Hume’s presentation
and defense of the sceptical argument, and his intended use of that argument.

I
Suppose that upon mulling over some issue in your mind, you reach a
decision, PeSSince you have made mistakes in reasoning in the past, however,
you find yourself (quite prudently perhaps) doubting P. Now according to
Hume,

...we are oblig’d by our reason to add a new doubt deriv‘d from the
possibility of error in the estimation we make of the truth and fidelity
of our faculties. This is a doubt, which immediately occurs to us, and
to which, if we would closely pursue our reason, we cannot avoid
giving a decision. But this decision, tho’ it should be favourable to our

HUMESTUDIFS
 Hume and the Limits ofReason 91

preceeding judgment, being founded only on probability, must

weaken still further our first evidence, and must itself by weaken’d by
a fourth doubt of the same kind, and so on in infinitum; till at last
there remain nothing of the original probability ... (T 182)

While the basic idea here is perhaps clear enough, the details of the
argument Hume is presenting are not. The basic idea is that an awareness of
our fallibility should, “if we closely pursue our reason” lead to a succession of
doubts which undermine our original belief. Given that we realize this, the
implication is that we should not believe P to start with. Thus our natural
fallibility, together with the fact that, “[iln every judgment, which we can
form concerning probability ...we ought always to correct the first judg-
ment ...by another” would seemingly place us in a sceptical dilemma (T 181-
182). And yet, while the argument’s general structure seems clear enough at
first, a second look at the details shows that it is ambiguously presented-a fact
most easily seen by considering two different interpretations of the passage.
The first interpretation is similar to the one presented by Robert Fogelin; the
second traces its lineage to some unpublished notes on the subject by
Jonathan Bennett.6
(A) Suppose upon reflection I come to the conclusion P and assign it a
probability of 0.8. Nevertheless, I might recall that I have made errors in rea-
soning in the past, and thus feel a bit uneasy about the reasoning described in
the previous sentence. So, I assign a probability of 0.7 to the proposition that
my original assignment was right. Again, I recall that my reasoning is not al-
ways perfect, and hence assign only a 0.6 probability to the proposition that
the assignment made in the last sentence was correct. And so on. If 1 continue
on in the same vein, I will eventually reduce the probability of the original
proposition to “nothing”; that is to 0. Assuming that I recognize this fact, 1
should assign that probability to P straightaway, and hence should not believe
that P.
(B) Suppose upon reflection I come to the conclusion P and assign it a
probability of 0.8. Nevertheless, I might recall that I have made errors in
reasoning in the past, and thus feel a bit uneasy about the reasoning described
in the previous sentence. Therefore, I lower the probability of P slightly, say to
0.7. But of course, I realize that I have made errors in reasoning before, so I
lower the value of P’s probability again, this time to 0.6, and so on. If 1 con-
tinue on in the same vein, I will eventually reduce the probability of the
original proposition to “nothing”; that is to 0. Assuming that I recognize this
fact, I should assign that probability to P straightaway, and hence should not
believe that P.
Each interpretation picks out a different feature of Hume’s own presenta-
tion of the argument. Interpretation (A) focuses on what Hume calls the
“reflexive” nature of the sceptical doubts (T 184), while (B) has the additional

Volume XXII, Number 1, April 1996

92 Michael P. Lynch

virtue of incorporating in a more direct fashion Hume’s assertion that each

higher level doubt lowers the probability of the original judgment. Both
interpretations, however, are unsound. As Fogelin has noted, a trivial reason
for this is that they are both invalid as they stand; both arguments are missing
an important step.7 In order to work, both versions must rule out the case
where the diminutions of probability become progressively smaller and hence
reach a finite limit. A premise must be added to the effect that for any arbitrary
level of doubt D, the diminution of probability at D must be greater or equal
to the diminution at level D-1. The addition of such a premise requires some
discussion; but there are more pressing problems to be dealt with first.
According to Fogelin and Hacking, (A) fails because of a confusion of
levels. Version (A), unlike (B), holds that the target of one’s second-level doubt
is not one’s belief that
(1) the probability of P is 0.8,
but the belief that
(2) I am correct in assigning the probability of 0.8 to P.

What’s the difference? Well, lowering the probability of (2) just by itself
won’t lower the probability of (1). An admission that I could be wrong in the
first assignment (the possibility that (2) is false) does not entail that (1) is false.
As Fogelin says: “However certain or uncertain we are about our ability to
calculate probabilities, if a proposition has a certain probability, that
(tautologically) is the probability it has.”*
While clearly correct, this point only holds if Hume is concerned with
objective, rather than subjective, probability. If our concern is how confident
we are in our beliefs, not how probable they are in fact, then a lowering of
confidence in (2) will result in a lowering of confidence in (l).9If this is
Hume’s point, then (A) will collapse into (B). And there is good reason to be-
lieve that Hume was concerned with confidence levels. Consider his repeated
use of the word ‘doubt‘, and the locution ‘degrees of assurance’ (see, e.g.,
T 184) as well as his assertion that the argument gives him “less confidence in
his opinions” (T 183). These turns of phrase suggest that the target is a more
subjective conception of probability. More simply, it is reasonable to think
that any argument worthy of being called sceptical would be concerned with
undermining confidence and instilling doubt.
Both versions of the argument actually fail for a different reason. There is
just as much reason to think that I initially gave P “too much” confidence as
there is to think I gave it “too little.” Errors in reasoning go both ways; some-
times we underestimate, sometimes we overestimate. Barring further evidence,
there is no way of telling which I may have done in this situation. So, there is
no reason to lower the original probability assignment; thus P’s probability
will not be reduced to nothing. To put it another way, if one and the same

HUMESTUDIES
 Hume and the Limits of Reason 93

reason supports both my having undercalculated and my having over-

calculated, why should I feel less confident about my original belief? It would
seem that I am back to where I started. As Mikael Karlsson has pointed out, this
point holds equally well against Morris’ statement of the argument, which is
essentially (A) cashed out in terms of confidence Morris says,

As I work through the iterated assessments of my probability claim,

and recognize that with each step the probability that I’ve correctly
assessed the original proposition diminishes, so should my
confidence, or the strength with which I hold the original belief,
diminish. As it does, 1 begin to think that the likelihood that the
opposing belief-the negation of that belief-is true increases.”

The assumption lurking here is that once I begin to think I have made a
mistake about the original proposition, I must lower the (subjective) prob-
ability of that original belief. But this does not strictly follow. While I will in-
deed lower my confidence in the proposition that the original assignment was
right, I could in fact raise my confidence in the original belief. For all I know,
the mistake I might have made was one of underestimation of the evidence for
P; so being less confident in the proposition that my first assignment was right
doesn’t necessarily mean that I should be less confident in believing P. Of
course, it doesn’t show that I should be more confident about P either; what
it shows is that we are in something of a stalemate. The mere fact that I could
have made a mistake in reasoning doesn’t show me which type of mistake I’ve
made. And thus it is hard to see why I should, on the basis of these arguments,
lower my confidence in the original belief.
Hence, none of the above interpretations are sound. But the very point
which is their undoing points us toward a subtly different, and more success-
ful, reading of the argument. To repeat, Karlsson’s point against Morris is that
the mere possibility that I’m mistaken in estimating P’s probability doesn’t
show which type of mistake-over- or underestimation-has been made;
epistemically speaking, I’m stalemated. Of course, once I’ve realized this-that
is, once I’ve realized at level 2 that I might have bungled at level 1-then a
little reflection should show me that I have no idea how big-and what
type-of a bungle I might have committed.
For example, imagine that upon looking throughout the house, I become
fairly sure that I left my wallet at a restaurant. Reflection reveals that 1 have
often reasoned poorly about such things in the past, however, and hence I
conclude that the degree of confidence I have in my belief is misplaced. I re-
alize that I may have overlooked something, and if so, should feel less sure in
my belief that my wallet was left at the restaurant. On the other hand (as a
frustrated friend might point out) I have looked throughout the house, and
thus perhaps I should be even more confident in my belief, and call the

Volume XXII, Number 1, April 1996

I
94 Michael P. Lynch

restaurant immediately. A little further reflection shows that the reasoning

just described may be wrong as well, and so on, ad infiniturn, until 1 literally
don’t know what to believe. More precisely:
(C) Suppose that upon reflection I come to a conclusion, P, and assign it
a probability of 0.8. Nevertheless, 1 might recall that I have made errors in
reasoning in the past, and thus feel a bit uneasy about the reasoning described
in the previous sentence. I realize that I might have overestimated or under-
estimated the original probability. Thus, at level 2, 1 should “if I pursue my
reason closely,” widen the range of P’s probability from 0.8; for example, I
should conclude that the subjective probability of my level 1 belief-the de-
gree of confidence I have in that belief-is somewhere between 0.7 and 0.9.
But of course, at level 3, I realize that I have made errors in reasoning before,
so I widen the range of P’s probability again, and so on. Assuming that at each
level the amount of widening was the same or greater, then continuing on in
the same vein will eventually leave me in the position of thinking only that
the subjective probability of P is somewhere between 0 and l! Hence, 1 have
no reason for thinking P as any more probable than any other belief,
(including not-P) and thus, assuming that I recognize this, I should not believe
P--I should withhold belief altogether.
How faithful is this interpretation to Hume? An obvious criticism is that
the very point which would allow Hume to sidestep Karlsson’s criticism
implies that the target-the conclusion of the argument-cannot be a
reduction of the original belief‘s probability, but rather an elimination of any
subjective probability at all. For (C) does not result in P‘s probability being
reduced to “nothing”-that is, to zero-but rather in a general state of
befuddlement. Our confidence disappears altogether. We are left with no
more reason to think that P is true than to think that it is false. Thus (C) would
seem to be at odds with Hume’s talk of a “weakening” of evidence.
This problem is not as serious as it looks at first; (C) is entirely compatible
with Hume’s putting the point in terms of a reduction or weakening of prob-
ability. For recall that Hume is concerned here with subjective probability. The
concern is the confidence we have in our beliefs. But that confidence is
diminished under (C): specifically, as we go through the argument, what we
become less and less confident about is the belief that our initial assignment
of subjective probability to P is correct. This loss of confidence to the effect
that we were right the first time (that we didn’t over- or underestimate) is
precisely what forces us to widen the probability spread. The more we widen
it, the more apparent it becomes that we don’t know how much confidence
we should have in P; which in turn, leads us to not assign P any probability at
all-we literally don’t know what to believe. The belief whose (subjective)
probability is reduced to zero is the belief that P‘s original assignment was
correct; and of course, this doesn‘t entail that P’s probability is zero-instead,
we should simply withhold from believing anything, whether P or not-P.

HUMESTUDIES
 Hume and the Limits of Reason 95

Further, the state of general befuddlement in which (C) leaves us is exactly

what Hume maintained was the overall aim of the sceptical argument. For
instance, after giving his version of the argument, Hume summarizes what he
takes to be the point:

When 1 reflect on the natural fallibility of my judgment, I have less

confidence in my opinions, than when I only consider the objects
concerning which 1 reason; and when I proceed still farther, to turn
the scrutiny against every successive estimation I make of my
faculties; all the rules of logic require a continual diminution, and at
last a total extinction of belief and evidence. (T 183; emphasis mine)

With its mention of “diminution,” this passage again reflects Hume’s tying of
the argument with a reduction of probability. But most importantly, it offers
a construal of the target of the argument: a total extinction of belief and
evidence.I2 At many points in fact, Hume talks this way, saying that the
argument results in “total uncertainty” (T 184) and “subverts all belief and
opinion” (T 183). Thus, (C) has the virtue of being aimed at Hume’s desired
conclusion. Given this, and given that it is more plausible, I submit that this
interpretation is a serious contender for being what Hume had in mind. It
certainly goes farther towards explaining why Hume would put so much stock
in the argument in the first place.
Finally, it is worth noting that our new interpretation of the argument
reveals a connection between it and more traditional sceptical arguments.
Hume was certainly aware of such arguments; although the degree to which
he was familiar with them is a matter of contention. Hume himself sometimes
encourages the view that he is presenting, in an updated and improved form,
the arguments of the ancient sceptics.13 Ancient sceptics would indeed, it
seems, have been pleasantly disposed toward Hume’s argument in I iv 1. As
Julia Annas and Jonathan Barnes note,

The ancient sceptics did not [just] attack knowledge: they attacked
belief. They argued that, under sceptical pressure, our beliefs turn out
to be groundless and that we have no more reason to believe than to
di~be1ieve.l~

The point is that, Hume, like the ancient sceptics, wasn’t content with show-
ing that scepticism undermines knowledge while leaving justified belief in-
tact. Rather, sceptical arguments leave us with no reason to believe anything.
As I have been presenting it, this is precisely the goal of Hume’s argument: as
I said above, argument (C) results in our having no more reason to believe P
than not-P.
Is argument (C) sound? Obviously, we must first determine whether it is
valid; and the question here is the aforementioned assumption, which all

Volume XXII, Number 1, April 1996

96 Michael P. Lynch

three interpretations must make, that at each level, the degree of doubt
involved must be equal or greater to the degree involved at previous levels.
There is at least one good reason I can think of to make that assumption, and
it is that there is no good reason not to make it. Whatever amount of doubt
one starts with, no matter how small, there is no principled reason to dower
that amount as the argument proceeds. In fact, it seems that if one did do so,
then the sceptic would be within her rights to cry foul, for if (she might say)
you doubted your reasoning to degree n at level 1, then, given that at level 2
you have no more or less reason to believe you have erred, then for
consistency’s sake, if for nothing else, you should lower your confidence by a
degree of n at level 2 as well. In other words, since one’s epistemic situation
vis h vis one’s own reliability does not change during the course of the
argument (how could it?) then clearly one should have the same degree of
suspicion towards one’s reasoning all the way through.
So the argument seems quite plausible. Of course, anyone who draws a
sharp distinction between the activity of justifying and the state of being
justified may reject the argument. Suppose that one can be justified in
believing that P even if one cannot justify that belief. If so, then even if I
cannot prove (to myself or others) that my reasoning is correct (i.e., that my
calculation of probabilities is correct), it may be correct, and hence, I may be
justified. However, this response can be less persuasive against the version of
the argument that does not aim at objectively undermining our beliefs, but
only at making us doubt those beliefs. Even if I am justified in believing my
reasoning to be correct, I won’t be able to prove it to myself, and hence a
niggle of doubt must remain.ls The argument would seem to yield a
ineliminable degree of doubt about our reasoning.

I1
Hume’s sceptical argument in I iv 1 has turned out to be surprisingly
resistant to criticism. But even if one does end up rejecting Hume’s sceptical
argument, there remains much that is of interest in I iv 1. I now want to
examine two interrelated questions. First, why did Hume believe that
reflection on the argument could not cause anyone to lose faith in their rea-
son? Second, given that he didn’t believe that the argument could convince
anyone of scepticism, what was his purpose in presenting it?
Hume is quick to point out that “neither I, nor any other person was ever
sincerely and constantly [such a total sceptic]” (T 183). This is apparently a
matter of physical necessity:

Nature, by an absoIute and uncontroulable necessity has determin‘d

us to judge as well as to breathe and feel... (T 183)

HUMESTUDIES
 Hume and the Limits ofReason 97

Thus, humans just naturally reason, and no amount of sceptical

argumentation, no matter how compelling in the abstract, will get around
that fact. Anyone who follows the sceptical argument just won’t lose faith in
their reason: they will continue to have beliefs and will continue to take some
of these beliefs as more reasonable than others. Indeed, since he takes the
sceptical argument as sound, Hume does not argue against it at all. Rather, he
is interested in explaining why it is that no one “is a complete and total”
sceptic. He believes that the explanation for this fact provides evidence for his
theory of mind. As he says,

My intention then in displaying so carefully the arguments [for

scepticism regarding reason] is only to make the reader sensible of the
truth of my hypothesis, that all our reasonings concerning cause and ef-
fects, are deriv’d from nothing but custom; and that belief is more properly
an act of the sensitive, than of the cogitative part of our natures. (T 183)

The “hypothesis” in question, for which the discussion of the sceptical

arguments supposedly provides evidence, is actually (as the “and” indicates)
two views wrapped into one. The two are both aspects of what we might call
his “theory of belief.” Unfortunately, the parts in question do not map neatly
onto the above conjunction; nonetheless, both are being invoked here, and it
is important not to run them together, even if Hume did not distinguish them
as carefully as he might have.
The first point Hume is making is that the sceptical argument, together
with the obvious fact that we continue to go on trusting our judgments any-
way, is evidence for his account of what a belief is-for his theory of the nature
of belief, in other words. A belief, o n Hume’s account, is distinguished from
a mere entertainment of an idea, in that the former is essentially more
“vivacious” and forceful. A belief that P is flashier, and sticks around longer
than a passing thought that P; without these features, which result from the
“sensitive part of our natures,” a belief would merely be “a simple act of
thought” (T 184). If it were, Hume reasons, then we could not explain why our
beliefs withstand the sceptical assault. Rather, Hume claims that a belief is a
particular “manner of conception”; and hence, that the argument fails
because as we ascend the ladder of the argument, our beliefs which express the
sceptical doubts become more and more “faint and obscure” until they finally
fade away to nothing (T 185). As Hume put the point

The attention is on the stretch: The posture of the mind is uneasy;

and the spirits being diverted from their natural course, are not gov-
ern’d in their movements by the same laws, at least not to the same
degree, as when they flow in their usual channel. (T 185)

Thus the length, complication, and downright weirdness of the sceptical

Volume XXII, Number 1, April 1996

98 Michael P. Lynch

argument is what prevents us from believing its conclusion, for by the time
we get there, the relevant beliefs which make up the stages of the sceptical
argument have lost so much vivacity as not to be beliefs at all. Hume is right
that it is psychologically difficult to follow the sceptical argument very far.
Further, one can certainly be sympathetic to Hume’s claim that this fact is
singularly explained by his account of (the nature of) belief. For (i) as we‘ve
just seen, his account of belief does at least supply something of an explanation,
and (ii) as Hume touts (T 623, 627) theories of the nature of belief were not
running rampant in the philosophical field of his day-hence one can’t really
expect him to discuss how other theories of belief might handle the problem.
Strictly speaking, however, the fact we can’t follow the sceptical argument
very far does not offer much support for his theory that a belief is an idea
associated with a certain feeling. This is because that fact is completely
compatible-as far as I can see-with any notion of what (ontologically)a belief
is. It certainly doesn’t imply that a belief must be a feeling of some sort. For
instance, I see no reason why a behaviorist could not say that folks simply are
not disposed to pay attention to such sceptical arguments for very long, and
then go on to explain this fact by appealing to certain environmental and
social factors.
Further, Hume’s actual reason for why we are not persuaded by the
sceptic’s argument is an empirical claim- even if his analysis of belief itself is
not. His claim is that we cannot sustain sceptical argumentation because the
ideas involved become less and less vivacious as we do so. Further, he gener-
alizes the point, concluding that “the conviction, which arises from a subtile
reasoning, diminishes in proportion to the efforts, which the imagination
makes to enter into the reasoning” (T 186). As I remarked above, none of this
is that implausible; but unfortunately, it is also not a question that we can
settle today outside of the domain of empirical psychology.16
Finally, this aspect of Hume’s explanation has an interesting
consequence. It implies that the sceptical argument always succeeds in low-
ering our confidence in our beliefs to some degree. For Hume’s point is only
that the regress of doubt, not every doubt itself, is halted by the mind. To see
this, assume that any version of the argument works. Then suppose that you
have reached some level of doubt, (say level 10) only to find that, due to the
unnatural posture of your mind, the level 10 doubt is drained of all its force.
According to Hume, that means that a belief is no longer involved; that is, you
don’t really believe at this point that your reasoning has been fallible etc. But
on any interpretation of the argument, this is not going to eliminate all of
your doubts. For instance, according to interpretation (B), where at each level
your doubt causes you to knock a bit off the probability of your original belief
P, your stopping at level 10 only means that you don’t knock off the bit of
probability you would have if you hadn’t stopped at level 10. Thus, you still
think your original belief is less probable (by 9 levels of doubt, we might say)

HUMESTUDIFS
 Hume and the Limits of Reason 99

than when you started.”

Thus the first point Hume is trying to make, namely that reflection on the
sceptical argument provides positive support for his views about the nature of
belief, does not fare all that well under scrutiny. His second point fares better.
In saying that “all our reasonings concerning cause and effects, are derived
from nothing but custom,” Hume is saying that our causal reasoning is
something that has its origins in human nature. It is something which we find
ourselves naturally disposed to do by our primitive or animal nature; and
therefore, causal reasoning is not a process which some mysterious faculty
called “reason” engages in through acts of “wi11.”18 This is the point Hume is
making when he says that, as we move up the ladder of doubt, we have less
and less confidence in the sceptical argument despite the fact that “...the
principles of judgment, and the balancing of opposite causes be the same as
at the very beginning“ (T 185). By “opposite causes” here, Hume apparently
means “opposing reasons”; so the point is that we lose faith in the sceptical
argument despite the fact that the “perfect rules of reason” tell us that we
should not. Since these principles of judgment, these rules, derive from the
cogitative aspect of ourselves, and since they obviously fail to generate belief
in this case, Hume concludes that our causal reasoning, and the beliefs which
are the result of such reasoning, must themselves be ultimately grounded in
the ”sensitive” aspect of our nature.19
Note that this is a separate point from the first. The former concerns the
nature of any particular belief; the second concerns that in which causal rea-
soning is ultimately grounded. The two are logically independent; for while
Hume may be wrong in thinking that beliefs are vivacious ideas, he is at the
very least headed in the right direction when he insists that our reasoning is
grounded in our (all too human) nature and not in faculties which operate by
perfect and abstract rules. Nonetheless, both points are being made here; and
it would be a mistake to ignore either.
These points place me at odds with William Morris-whose work I
partially defended in section I of this paper. According to Morris, the sceptical
argument in I iv 1 is be understood as an extended reductio of a “prevalent
rationalist or intellectualist model of the way the mind works.”20 Morris takes
it that Part I11 of the Treatise is meant to “demolish the view that our causal
expectations are based in any way on reasoning.”21 It is meant to show that i f
we adopted the “rationalistic” model of the mind, we would be compelled to
be sceptics about causal inference. But Hume is not a sceptic: he instead re-
places the mistaken psychological model with his own, naturalistic account
of the mind. So given that it was his chief aim to undermine a certain model
of the mind, it would be a good strategy on his part to show, in Part IV, that
this model’s conception of reason crumbles just where we would think it
would succeed the most: i.e., in demonstrative reasoning. If this is so, then
perhaps we don’t need to take Hume as supplying any independent reason for

Volume XXII, Number 1, April 1996

100 Michael P. Lynch

adopting his theory of belief in I iv 1. Rather, what he is trying to do is remove

a final objection to a theory he already takes himself to have explained and
defended in Part 111.
My resistance to Morris’ suggestions arises from two sources. First, and less
important, I think that the support for his theory of mind Hume intends to
gather from I iv 1 is not purely negative. As I have argued above, Hume ap-
parently takes the fact that we cannot believe the sceptical argument to be best
explained by his own theory of what a belief is. Thus, that his theory accounts
for the data is a positive point in its favor. Second, and much more crucially,
I am not convinced that Hume is attempting to show that our confidence in
our judgment can never be based on reasoning. One problem is that Hume
himself takes it (in T 111) that algebra and arithmetic are sciences that “preserve
a perfect exactness and certainty” (T 71). In these sciences, we can compare
ideas and “determine their relations without any possibility of error” (T 71).
In other words, Hume takes it that when we perform arithmetic-during
which we are surely reasoning-we can and do form judgments in which we
are (rightly) confident. Further, he clearly holds that a belief formed when
doing arithmetic is subject to critical and rational evaluation: he says in the
first sentence of I iv 1 that the rules of logic are “certain and infallible”; it is
when we apply these rules that we fall into error. If such judgments weren’t
the product of some form of reasoning, there would be little sense in saying
that some of them could be false. Hence, if Morris were right, Hume would
have to be speaking in these passages in the voice of the rationalist, but it is
at best not clear that this is so. Further, these and other passages suggest that
Hume did think that we form judgments based on reason, and further, that
there are in fact, right and wrong ways to do so. As I see it, the point is not that
beliefs (and our confidence in them) can’t result from reasoning; rather, the
point is that the nature, ground and limits of reasoning itself must be re-
examined.
Another problem for Morris’ suggestion is that Hume admits that the
sceptical argument can also be run against someone who holds his view of
mind as well. After having criticized the notion that a belief cannot be a
“simple act of thought,” Hume imagines someone objecting in this manner:

But here, perhaps, it may be demanded, how it happens, even upon

my hypothesis, that these arguments above-explain’d produce not a
total suspense of judgment, and after what manner the mind ever
retains a degree of assurance in any subject? (T 184; emphasis mine)

On the preceding page, Hume uses the phrase: “my hypothesis” as a name for
his theory that belief is “properly an act of the sensitive, than of the cogitative
part of our natures” (T 183). As we’ve seen above, this hypothesis is actually
the sum of two parts: (a) Hume’s view that beliefs are vivacious ideas, and (b)

HUMESTUDIES
 Hume and the Limits of Reason 101

his view that our causal reasoning and beliefs are grounded in our animal
nature. Now after raising the above challenge, he goes on to rephrase the
sceptical argument again, this time in language more befitting his own theory.
His reply to this version is as revealing for what it doesn’t say as much as for
what it does. For Hume claims, as we’ve seen, that one doesn’t believe the
sceptical arguments because one can’t believe them. Further, as we‘ve seen, he
takes this fact to be best explained by his ’‘hypothesis‘’-that is, by (a) and (b).
But nowhere does Hume say that his views get us off the sceptical hook, as we
would expect if Morris were right. Hume thinks these arguments are sound,
period-and not just sound relative to a certain assumed model of the mind.
And here he is surely right; the sceptical argument, if sound at all, is sound
because-and not in spite of-the nature of reasoning itself. Hume’s point is
that despite this, we won’t believe it, and further, that this fact is best
explained by his own views about the mind.
So I am not persuaded that Hume’s purpose in section I iv 1 is to provide
a reductio of a certain conception of the mind. But the disagreement between
Morris and myself on this point may be nothing more than a family quarrel,
for I think that Morris is right in believing that Hume is neither a radical
sceptic nor a dogmatist. Further, he is trying to convince us to let go of a
particular conception of reasoning, even if he is not trying to get us to think
that no belief can be grounded on reason. This point is made most clearly by
Hume during his own summaries of his thinking about our topic. At the be-
ginning of the section after I iv 1, (the far more famous “Of Scepticism with
Regard to the Senses”) Hume sums up his point in regard to reason this way:

Thus the sceptic continues to reason and believe, even tho’ he asserts,
that he cannot defend his reason by reason ...“ (T 187)

And at the end of that same section:

‘Tis impossible upon any system to defend either our understanding

or our senses; and we but expose them farther when we endeavour to
justify them in that manner. (T 218)

And again at the close of Book 1:

...I have already shewn, that the understanding, when it acts alone,
and according to its most general principles, entirely subverts itself,
and leaves not the lowest degree of evidence in any proposition,
either in philosophy or common life. (T 267-268)

All of these passages purport to summarize the point of the section we have
been considering, and all express a common core: that reason cannot defend
its own reliability. Any attempt to show why reason is trustworthy-to prove

Volume XXII, Number 1, April 1996

102 Michael P. Lynch

that reason itself is a guide worthy of following-quickly and decisively

undermines itself. We cannot defend reason by reason without “exposing
ourselves still further” because, very simply, any attempt to do so is guilty of
epistemic circularity.22 When we try to employ the understanding “on its
own”-that is, when we try to use only our reason to defend reason-we end
up begging the very question being asked. Hence, if we cannot justify our
practice of reasoning, then we can only bite the bullet and appeal to what we
find “natural and easy” (T 268). As David Pears has said recently: “Reason
cannot defend the principles which we need to steer us through our lives, and
so nature takes over and engraves them on our minds.”23 Individual reasons
for individual beliefs might be questioned, but reason itself cannot be; we can
only say that we will continue to engage in it. But if we can give no reason to
trust our reasoning, then why should we? Hume’s answer is that the point is
moot. We can’t stop reasoning, and thus the question of whether it is rational
or not to engage in reason is an entirely empty one. The question has already
been settled, as Hume would say, by Nature.
Finally, this way of looking at things makes it a bit clearer how I iv 1 fits
into Hume’s overall project in Part IV. In particular, we can draw a precise
parallel between 1 iv 1 and the following section “On Scepticism with Regard
to the Senses.” In the first paragraph of that section, Hume announces that
just as the sceptic cannot defend his reason by reason,

...by the same rule he must assent tothe principle concerning the
existence of body, tho’ he cannot pretend by any arguments of
philosophy to maintain its veracity ....We may well ask, What causes
induce us to believe in the existence of body? but ‘tis vain to ask, Whether
there be body or not? That is a point, which we must take for granted in
all of our reasonings. (T 187)

Similarly, we can inquire: What causes us to think that reason is reliable?

But not: Is reason reliable? It may be, but we will not be able to explain why
it is without relying on reason itself. This point is simple and important. It
shows us that there is no foundation to which we can appeal in order to justify
our use of reason. We simply cannot answer the sceptic’s question. Thus
Morris is right in thinking that in I iv 1 Hume is separating himself from a
certain tradition. But his point isn’t that reason can’t generate belief-
obviously it can. The task is to explain how it can be that we have such beliefs
at all. In sum, since according to Hume, reason cannot ground itself, it cannot
supply the epistemically secure foundation it was originally thought to pro-
vide. It cannot do this because it cannot justify its own reliability. Our rea-
soning must therefore be grounded in turn on something else-the natural
mechanisms of the imagination: Custom and Habit.
 Hume and the Limits ofReason 103

Hume’s sceptical argument is better-and more threatening-than it

appears at first glance. Further, reflection on the sceptical argument
encourages us to adopt a more naturalistic standpoint toward the mind and,
in particular, toward what we might call our doxastic faculties. It advises us to
stop looking for an indubitable ground of those faculties and to get on with
the business of finding out how we use them.

NOTES
I want to thank the following people for helpful comments on earlier drafts of
this paper: Jonathan Bennett, Paul Bloomfield, William Morris, R. M.
Sainsbury, and two anonymous referees.
1 David Hume, A Treatise o f H u m a n Nature, edited by L. A. Selby-Bigge, 2nd
ed. revised by P.H. Nidditch (Oxford: Clarendon, 1978), 180.
2 For a sampling of the recent literature on I iv 1, see John Passmore, Hume‘s
Intentions (London: Gerald Duckworth,l952), 513-516; Ian Hacking, “Hume’s
Species of Probability,” Philosophical Studies 33 (1978): 30; Fred Wilson,
“Hume’s Sceptical Argument Against Reason,” Hume Studies 9.2 (1 983):
90-129; Robert Fogelin, Hume’s Skepticism in the Treatise of Human Nature
(London:Routledge & Kegan Paul, 1985), and “Hume’s Scepticism,” The
Cambridge Companion to Hume, edited by David Fate Norton (Cambridge:
Cambridge University Press, 1993), 90-1 1; William E. Morris, “Hume’s
Scepticism About Reason,” Hume Studies 15.1 (1989): 39-60; and Mikael M.
Karlsson, “Epistemic Leaks and Epistemic Meltdowns: A Response to William
Morris on Scepticism with Regard to Reason,” Hume Studies 16.2 (1990):
121-130. All but Morris find the argument deeply confused.
3 Morris, Karlsson, and Fogelin (1993) all make this point.
4 For a more detailed presentation of Hume’s first argument, see Morris,
40-46. Morris supplies what I see as one of the sharpest discussions to-date on
I iv 1. Morris takes what I call “two arguments” to be different premises in one
argument aimed at the stronger conclusion noted in the text: namely, as
Morris says, that “our use of reason can’t even generate belief” (39). No doubt
that this was Hume’s intention, but it still seems to me that one might
acknowledge that no belief is certain without believing that no belief is
justified.
5 As far as I can see, “P” could be a place-holder for any proposition. One
might think that, since this second argument concerns only beliefs based on
probability, the domain in question would be restricted to only those
propositions. But recall that Hume, at this point in the section, believes that
he has shown that all “knowledge degenerates into probability” and hence, it
would seem that the candidates for P are wide open.
6 I refer specifically here to Fogelin, 1985. Fogelin himself refers to Hacking,
30. I wish to thank Bennett for allowing me to cite his work, and for several key
discussions of these issues which greatly helped to sharpen my thinking about
them.

Volume XXII, Number 1, April 1996

104 Michael P. Lynch

7 Fogelin, 1985, 17.

8 Fogelin, 1985, 18.
9 Morris, 52, makes a similar point. Notice that I don’t say: a lowering of
confidence in (2) will result in us lowering the probability of (1) further than
0.8. In recognizing that we could be wrong at level (2) we will be less confident
that the probability of P is 0.8. But we might begin to think it is now 0.9! See
below for further comment.
10 See Karlsson, 1990, 126.
1 1 Morris,50.
12 1 take it that when Hume mentions “evidence” here, as in many other
parts of the Treatise, he means “evidentness”; that is, being evident to the
subject.
13 On Hume’s familiarity with traditional scepticism see Fogelin, 1985, 173
n2; Barry Stroud, “Hume’s Scepticism: Natural Instincts and Philosophical
Reflection,” Philosophical Topics 19.1 (1992);Richard HPopkin, “David Hume:
His F’yrrhonism and His Critique of Pyrrhonism,” in Hume, edited by by V. C.
Chappell (Indiana: University of Notre Dame Press, 1968); and Passmore,
134136.
14 Julias Annas and Jonathan Barnes, The Modes of Scepticism
(Cambridge:CambridgeUniversity Press, 1985), 8.
15 ‘Niggle’ is a technical term for a tiny bit of doubt (Fogelin, 1985).
16 His general point seems particularly vulnerable. Note that if true, then
given the complex and subtle reasoning found throughout the Treatise (in fact,
in this very section!)we could never find ourselves in strong agreement with its
author !
1 7 It is unclear whether Hume was aware of this point. On the one hand, he
points out in the beginning of I iv 1 that, “In the man of the best sense and
longest experience, this authority [of our confidence] is never entire; since
even such-a-one must be conscious of many errors in the past, and must still
dread the like for the future” (T 182). On the other hand, he later says that after
encountering the sceptical arguments, one still “continues to believe, and
think and reason as usual” (T 184, emphasis mine).
18 See Barry Stroud, Hume (London: Routledge and Kegan Paul, 1977), 75ff.
19 Thanks to an anonymous referee for bringing this passage, and this point,
to my attention.
20 Morris, 56.
21 Morris, 55, emphasis mine.
22 I take this term from William Alston’s “Epistemic Circularity” in his
Epistemic Justification: Essays in the Theory of Knowledge (Ithaca: Cornell
University Press, 1989).
23 David Pears, Hume’s System (Oxford: Oxford University Press, 1990), 99.

Received July 1994

Revised January 1996

HUMESTUDIES
View publication stats