Reza N ega resta n i
Req u i e m fo r Detective F i ct i o n
85
:::0
m
N
)>
z
m
Ci)
)>
:::0
The Pledge: Requiem tor the Detective Novel is a novel by the Swiss dramatist
Friedrich DOrrenmatt published in 1958. It tells the story of a super -detective
named Matthai who is now insane and working at an isolated gas station in a
small forsaken town. Years earlier. Matthai had come to this place to solve the
case of a child murder. From the start. Matthai begins to assemble a highly
detailed profile of the criminal. I n effect. he solves the mystery right after taking
the case. Having solved the puzzle. Matthai then devises an elaborate trap to
catch the suspect. to confirm his sure theory and close yet another case. But
the serial killer never arrives. Shocked by the failure of his method. Matthai
succu mbs to alcoholism. Years later. with Matthai now insane. it is revealed
to the new chief of police that the serial killer was indeed heading to the trap
on the day Matthai expected him, but. in a sudden twist of fate. died in a car
accident on his way to bite Matthai 's bait .
This is a brief but sufficient sum mary of DOrrenmatt 's requiem for the
detective novel . We could even condense it into a simple description: This is a
story about the logic of detection and its failure owing to its faith in the prior
reliability of law-like generalities. Now I would like to make the unsurprising claim
that The Pledge is a philosophical thriller about epistemic inquiry, in particular an
old yet still open wound precipitated by epistemological scepticism. The nature
of this wound is precisely what I would like to address here. But before that. let
us briefly say that its nature is the problematizing and unresolved entanglement
of deductive and inductive inferences in epistemological investigation. particularly
that of science.
What turns DOrrenmatt's novel into a requiem for detective fiction is the way
in whi ch , in his story, the problem of induction i mplodes the logico-deductive
framework of detective fiction from within. It has long been suggested that
the traditional detective fiction-and the figure of the detective-represents
a method of investigation that is, in essence, logical in the sense that it primarily
m
(/)
-t
)>
z
'
:::0
m
V)
c
m
�
"'Tl
0
:::0
0
m
-t
m
()
-t
<
m
"'Tl
()
-t
0
z
86
focuses on the deductive rules of thought . I nductive gener aliz ations of obser
vations-patterns, evidence. etc.-are only secondary in their import . They are
:::0
m
N
)>
z
m
Ci)
)>
:::0
m
en
....
)>
z
'
:::0
m
0
c
m
�
'Tl
0
:::0
0
m
....
m
(')
....
<
m
'Tl
(')
....
0
z
aspects of police work rather than the defining characteri stics of a tru e detective .
Therefore, in the classical framework. the archetype of the detective is not a
scientist but a logician proficient in following deductive chains of thou ght no
matter how convoluted they might be and regardless of how far beyond salvage
the crime scene is contaminated . The triumph of the detective. the solving of
the crime, owes not to the forensic expertise of the detective. the sophistication
of laboratory instruments, and certainly not to the competence of the police
force or its information-gathering capabilities but . first and foremost. to the
logico-deductive skill of the detective. whose disinterested and mechanical
nature is often whitewashed by his eccentric and ingenuity-feigning charisma.
The contemporary hard realist detective genre. on the other hand , identifies
itself in complete contrast to this logical -deductivist structure of the traditional
detective fiction. Examples of the hard realist detective genre would be Forensic
Thrillers and Police Procedural Dramas. franchise series such as CS/ and Law
and Order. What distinguishes the hard realist genres from the traditional form
of detective fiction is the strong emphasis on the insufficiency of the logical
method of analysis. Instead they underline-often to the point of ideological
valorisation-empirical methods of crime solving a nd the technoscientific
adequacy of laboratory instruments, forensics, and police profiling . A half-burnt
cigarette in an ashtray is no longer a mere hypothesis in the detective's chain
of deduction whose truth can be determined by procedurally exhausting all
possible contradictions and incompatibilities. It is instead a subatomic field
teeming with empirical footprints, DNA traces. structural patterns that reflect
the muscular and sensory-motor habits of the criminal, and chemical molecules
rich with geographical metadata out of which the complete profile of the culprit.
their scientific image-both biological and psychological structures-can be
put together. It is in this sense that the new genre replaces the frigid logical
framework of the traditional detective fiction with the cold realism of science.
capable of uncovering the hard naturalized and empirical facts of physical laws
behind any inconspicuous sociocultural, political. or quotidian situation.
I associated hard realist detective fiction with an ideological valorisation of
the empirical sciences and the efficacy of technoscientific techniques. This is
_ -
- '"" '-' '"1ua1,..; y UT empirical
methods of investigation- by contemporary technoscience. In this sense, hard
realist detective fiction is fully i n tandem with a liberal naturalizing program
whose aim is to u nveil the ice-bound universe of empirical and naturalized facts
behind every potential terrestrial crime scene. be it social, political, or economic.
It i s this uncritical affirmation of the foolproof adequacy of empirical methods,
employed within an overarching naturalizing program whose limits have been
left uninvestigated and u ndemarcated , that makes hard realist detective fiction
a shining example of neoliberal capitalism's appropriation of the sciences. rather
than an exemplar of the modern empirical sciences as such . New detective fiction,
accordingly, derives its legitimacy from the reliability and efficacy of technosci
entific methods. But the problem is that. just because something reliably works,
that does not make it essentially u nshakable at its foundation, nor does it confer
epistemological legitimacy upon it. The efficacy of technoscientific methods in
the epistemological i nvestigation of every possible crime scene-Le. the claim
that it always works-is exactly the blind spot of hard realist detective fiction.
My contention is that Durren matt's novel underlines just this precarious
blind spot. It distils the superacid of epistemological scepticism that corrodes
the foundations of both the traditional and new realist detective genres. It is as
much a requiem for the traditional form of detective fiction as it is a requiem
for contemporary hard realist forensic fiction based on the organon of true
detection . science. As mentioned above. one of the main points of this episte
mological scepticism that Durrenmatt utilizes in his novel is that every genre
of detective fiction is built on an epistemolog ical system within which we are
not only dealing with logical deductive procedures (which are overemphasised
in traditional detective fiction) . More importantly and fundamentally, we are
also dealing with inductive inferences employed to derive law-like generalities,
patterns, and predictions on the basis of observations or past experiences
pertaining to different occurrences, clues, a nd footprints. Now, to the extent
that any epistemological investigation has a n underlying inductive component
(call it the inductive profile of the crime scene) . regard less of how sound the
::::0
m
N
)>
z
m
G)
)>
::::0
m
en
-i
)>
z
....
::::0
m
(j)
c
m
�
11
0
::::0
0
m
-i
m
()
-i
<
m
11
()
-i
0
z
-
88
of inductio n can at any time encroach u pon the procedu res of i nvestig ation and
disrupt the anticipated resolution .
:::0
m
N
)>
z
m
(j)
)>
:::0
m
CJ)
-l
)>
z
'
:::0
m
CD
c
m
�
It is not that M atthai 's logical inference is u nsound, nor is it that his method
of profiling the criminal 's featu res, behaviours, and predatory routines is flawed.
He is as gapless as a logical machine and as effective as a high -tech forensic
laborato ry. It is just that the problems of induction (of singling out patterns and
law- like regularities . of forming predictions and confirmations) have caught up
with his methods. And it is precisely the old and new problems of induction -the
corrosive substance of epistemological scepticism-to which both detective fic
tions and scientific inquiry qua true detection are in one way or another vulnerable.
But what are the old a nd new problems of i nduction ?
.,,
0
:::0
0
m
-l
m
()
-l
<
m
First let us look at a few definitions:
1. Deduction can be roughly u nderstood as a form of reasoning that links
premises to conclusions so t hat if the premises a re true, following step- by-step
logical rules. then the conclusion reached is also necessarily true.
2. Induction or inductive reasoning cannot be formulated as neatly as
.,,
()
-l
0
z
deduction. But roughly it is a form of reasoning in which premises provide
strong support (whether in causa l , statistica l , or computational terms) for the
outcome of the inference.1 Whereas the truth of the conclusion in deductive
reasoning is log ically certain, the truth of the outcome in inductive reasoning
is only probable in proportion to the supporting evidence. Hence, as evidence
piles up, the deg ree of supporting valid statements for a hypothesis indicate that
false hypotheses are-as a matter of generalization-probably false, and in the
same vein , that true hypotheses are-as a matter of generalization- probably
true. But this dependency on evidence also means that inductive inference is
contingent and non-monotonic. Non-monotonicity means that addition of new
premises can fundamentally change the truth of the conclusion. either drastically
raising or lowering the degree of support already established for the inductive
outcome. It is this epistemological ( rather than ontolog ica l ) contingency and
1.
' My reason for using the expression "outcome" rather than "conclusion" (which might seem a
more appropriate way of characterizing the terminus ad quern of an argument or inference) is that
although the whole point of a first order probability argument is to generate a terminal outcome. the
relation between a terminal outcome and the premises of the argument is radically different from that
which obtains between what we ordinarily call the conclusion of an argument and its premises.' W.
Sellars. 'Induction as Vindication', in Essays in Philosophy and its History (Dordrecht: Reidel, 1974 ). 370.
radical non-monotonicity that operates in the background of DOrrenmatt 's novel
89
and eventually throws Matthai ' s elaborate system of detection, and his mental
healt h , into disarray.
:;;o
Now back to the problems of induction : David H u me formulates one of the
strongest forms of the old problem of induction as follows: I nductive reasoning
is grounded on the principle of the uniformity of nature as its premise-that
is. unobserved events are similar to observed events, so t hat generalizations
obtained from past observed occurrences can be applied to future unobserved
occurrences. I n Hume's words, ' that instances of which we have had no expe
rience, must resemble those of which we have had experience, and that the
m
N
)>
z
m
G)
)>
:;;o
m
(/)
-I
)>
z
'
:;;o
m
(/)
c
m
course of nature continues always uniformly the same '.2 But this principle itself is
�
a conclusion reached by induction , and cannot be proved by the understanding
0
or by deductive reasoning . It cannot be proved by deduction because anything
that can be proved deductively is necessarily true. But the principle of uniformity
is not necessarily true since the deductive framework admits, without logical
contradiction , counterexamples for events which have not yet been experienced
in which a true antecedent ( past patterns of events) is consistent with the denial
of a consequent (future patterns of events not similar to the past) .
If the principle of u niformity cannot be proved through deduction, and if
therefore the validity of i nduction cannot be established deductively, then it
must be proved by causal-probabilistic or inductive reasoning . Yet the validity
of such reasoni ng is precisely what we sought to prove. To justify the principle
of uniformity and i nduction by inductive reasoning is simply question-begging ,
i.e. a fallacy i n which the conclusion is granted for the premises. Therefore, it
follows that induction cannot be proved inductively either, because this would
count as vicious circularity.
I n short. H ume's probl em of induction boils down to the idea that experience
ca nnot provide the idea of an effect by understanding or reason (i .e. deductive
and causal inferences) , but only by the impression of its cause which requires 'a
3
certai n association and relation of perceptions '. U nderstanding cannot produce
cause-effect relations or matters-of-fact since such relations are obtained by
i n d u ctive g enerali z ation of observations. Matters of fact rely on causal relations
2.
3.
D. Hurne , A Trea tise of Human
Ibid .
Nature ( London: Green and Co. , 1874), 390.
'T1
:;;o
0
m
-I
m
()
-I
<
m
"Tl
()
-I
0
z
90
and causal relations can only be obtained i nductively. But the validity of inductions
themselves cannot be corroborated deductively, nor can they be explained
;;a
m
N
)>
z
m
Ci)
)>
;;a
m
Cf)
-I
)>
z
'
;;a
m
ID
c
m
�
Tl
0
;;a
0
m
-i
m
()
-i
inductively. Therefore, w hat is problematic is not only the derivation of uncertain
conclusions from premises by way of i nduction, but also. and more gravely, the
very inductive principle by which such u ncertain conclusions are reached.
Argument about law-like regularities from the perspective of the numerical
asymmetry of recurring evidences or corroborating observations represents a
Humean version of the problem of inductio n . This is the idea that numerous
recurring i nstances of the same observations in the past are sufficient for estab
lishing law-like regularities that can in turn be used as explanatory components
for a ph enomenon that is i n need of explanation. Such inferences generally take
.
the form of proportional probability a rgume nts constrained by limit functions
which specify the parameters satisfying the degree of confirmation:
<
Given the hypothesis h expressible in the language L capable of accommodating
Tl
descriptions of space-time order and elementary number theory.
m
()
-i
0
z
- the proportion of n-items in the class of properties P is NIM.
- most of the m-membered subclass of P contain a proportion of n-items which
approximates that of P such that the next m individual instance ( xn +1. xn +2. .. xn+ m )
are all P and the degree of confirmation of the effective hypothesis P( xn+m +1 )
depending on the parameters of the system of inductive logic approaches 1 as
limit. or converges on or remains greater than o.g, or i n the weakest scenario.
becomes or remains greater than 0.5 irrespective of the property of the first
n individuals (e.g . , in the case of 0.5 s P( xn +
m+1
) . if n=8, then there must be
an m-let us say m=107-such that we can state that x . x
9
all green. then it is probable more than one-half that x
10
1000
. ... . x
being
1000
1 will also be gree n ).
S is an m-membered subclass of P.
- therefore, it is probable that the proportion of n-items in S approximates NIM,
once the above approximation is satisfied, the truth of h shall be accepted,
The acceptance of the probability-truth of h can be said to be the inductive
counterpart of a deductive conclusion.
However, just because up to the present observations have been manifested
in the form of a law-like regularity-in the sense t hat no i nstance of observation
within an acceptable limit has violated the pattern-this by itself does not
91
suffice to confirm the truth of the hypothesis. nor does it suffice to establish
the truth of its regularity. In other words, it is not just that . given a very long
time, more pattern-violating exceptions might be observed , but rather that we
cannot prove at any point the definition of the degree of confirmation even in
the weakest scenario (0.5 � P(xn+m+,) ) i n the sense that . save for few instances
at the beginning , one cannot establish that it is more likely than not that the next
individual m will conform to the proportional probability of observed n individuals.
The numerical asymmetry of supporting observations or confirming evidences
is precise ly that which is to be explained , the explanandum. Thus it cannot be
posited as that which explains, the explanans of its own law- like regularity.
Hume's problem of induction, accordingly, challenges the validity of our
predictions-the validity of the connection between what has been observed
in the past and what has not yet been observed . We cannot employ deductive
reason ing to justify such a connection since there are no valid rules of deductive
inference for predictive inferences. Hume's resolution to this predicament was
that our observations of patterns of events-one kind of event following another
kind of event-creates a habit of regularity in the mind . Our predictions are
then reliable to the extent that they draw on reliable causal habits or regularities
formed in the mind through the function of memory that allows us to correlate
an impression with its reproduction and anticipation. For example, if we have
the impression (or remember) that A resulted in B. and if we also witness at a
later time and i n another situation that 'an A of the same kind resulted in a B of
the same kind', then we anticipate a nomological relation between A and B: B is
the effect of A. as the cause of which we have an impression.
However, this resolution is also undermined by the problem of the reliability
of our memories. Bertrand Russell has formulated the strongest version of
the problematic nature of the reliability hypothesis (of memory) . According to
Russell 's ' five minutes ago' paradox .
There is no logical impossibility in the hypothesis that the world sprang into being
five minutes ago, exactly as it then was, with a population that 'remembered' a
wholly unreal past. There is no logically necessary connection between events
at different times: therefore nothing that is happening now or will happen in
:;o
m
N
)>
z
m
G)
)>
:;o
m
Cf)
-i
)>
z
....
::0
m
0
c
m
�
"Tl
0
:;o
0
m
-i
m
()
-t
<
m
"Tl
()
-t
0
z
92
the future can d isprove the hypothesis that the world bega n five minutes ago.
Hence the occurrences which are called knowledge of the past are logically
independent of th e past; they are wholly analysable into present contents, which
::u
m
N
l>
z
might . theoretically, be just what they are even if no past had existed .4
m
(j)
l>
:::0
The gist of the ' five minutes ago' paradox consists of two parts: (1) memo
m
CJ)
-i
ry-beliefs are constituted by what is happening now and not by the past time
l>
z
to which the said memory- beliefs appear to refer. I n so far as everything that
....
:::0
forms memory-beliefs is happening now, there is no logical necessity that what
m
"
c
is being remembered (the reference of the memory-belief) shou ld have actuaHy
m
occurred or even that the past should have existed at a ll . (2) There is no logical
�
"T1
0
reason to expect that memory states are in one-to-one correspondence with
0
m
-i
m
correlations between memory states and external states of affairs. Therefore
:::0
the rest of the universe. There can be both one-to- many and many-to- many
()
-i
<
what we remember as the impression of a cause, a past event . or an observa
m
tion . may very well be a false memory-either a different memory or a memory
'Tl
()
-i
of another impression of a cause. Accordingly, our knowledge of the past or of
0
z
the impressions of causes can also be problematic at the level of logical plausibil
ity and statistical improbability, which does not imply impossibility. Consequently,
it is not only the justification of our predictions regarding events wh ich have not
bee n experienced or obse rved yet that face difficulty, but also our memories
whether understood as physical states or irreducible mental states-of past
i mpressions which have shaped our regularities and habits of mind.
The Humean problem of induction is further sharpened into new riddles or
predicaments of induction by Nelson Goodman in his work Fact, Fiction and
Forecast (1955) , and then reformulated in the most radical way by Hilary Putnam
in Representation and Reality (1988) :
Let us imagine that before time t (e.g . , a hypothetical future time, say 2050)
we have observed many emeralds recovered from a local mine (or in Goodman's
examples. well-watered grass) to be green, and no emerald to be of another
colour. We thus have the following statements based on successful observations.
4.
8. Russell, The Analysis of Mind ( London: George Allen & Unwin Ltd, 1921 ) , 159-60.
Eme ra l d a i s g r e e n ,
eme r a l d b
i s g r e e n , etc.
Such evidence statements then support generalizations of the kind ,
93
:;o
m
N
)>
Al l eme r a l d s a r e g r e e n (not just in the local mine but everywhere) .
Here the predicate green can be said to be a projectable predicate or a predicate
that is confirmed by its instances (emerald a. emerald b, etc. ) and can be used
in law-like generalization for the purpose of prediction.
Now let us introduce the predicate grue. An emerald is grue provided it is
green and observed or (disjunction) blue and unobserved before the year 2050,
i.e. if and only if it is green before time t and blue afterwards. Here, the predicate
grue does not imply that emeralds have changed their colour, nor does it suggest
that. in order for emeralds to be grue, there must be confirmation or successful
observation of its instances. We call such a predicate a non- projectable or
grue-type predicate.
In the case of grue emeralds, we then have non-projectable generalizations,
Eme r a l d a i s g ru e ,
eme r a l d b . i s g rue , etc.
The generalizations 'All emeralds are green' and 'All emeralds are g rue' are
both confirmed by observations of green emeralds made before 2050. Before
2050, no grue emeralds can be observationally-Le. inductively-distinguished
from any green emeralds. Hence, the same observations support incompatible
hypotheses about emeralds to be observed after t-that they are green and
that they are blue. This is called Goodman's grue paradox. The paradox shows
that there can be generalizations of appropriate form which however are not
supported by their instances. So now the question is: What exactly is the dif
ference between innocent generalizations like 'All ravens are black' which are
supported by their instances, and grue-type generalizations which cannot be so
supported by their instances, but nevertheless are equally sound? Or, in other
words, how can we differentiate between healthy law- like generalizations based
on projectable predicates and grue-like (or not law-like) generalizations based on
non-projectable predicates? This is Goodman's new riddle of induction. which
z
m
G)
)>
:;o
m
(/)
-i
)>
z
....
:;o
m
0
c
m
�
.,,
0
:;o
0
m
-i
m
()
-i
<
m
.,,
()
-i
0
z
94
asks why it is that we assume that, after time t. we will find green emeralds but
not grue emeralds. given that both green and g rue-type inductions are true and
;;o
m
N
)>
z
m
G)
)>
)J
m
en
-I
)>
z
....
)J
m
ID
c
m
3:
,.,
0
)J
0
m
-I
m
(")
-I
<
m
false under the same set of conditions such that,
- Based on the observations of many emeralds, a miner using our common
language will inductively reason that all emeralds are green. The miner forms
the belief that all emeralds to be found in the mine or elsewhere are and will be
green before and after time t.
- Based on the same set of observations of green emeralds, a miner using the
predicate 'grue' will inductively reason that all emeralds observed after time t will
be blue. even though thus far only green emeralds have bee n observed.
Goodman's response to the paradox is as follows: the predicate green is not
essentially simpler than the predicate grue since. if we had been brought up to
,.,
use the predicate grue instead . it could very well be the case that grue would
(")
-I
0
z
by virtue of being green and blue. In that case, we could use predicates grue
no longer count as nonsensical or as more complex than the predicate green
and bleen ( i .e. blue before time t. or green afterwards) just as we now use the
predicates green and blue. An objection can be made that, unlike green . grue is
artificially defined disjunctively, and that therefore the natural predicate green
should be preferred . Per Goodman's response, there is no need to think of grue
and bleen-type predicates as disjunctive predicates. They can easily be thought
as primitive predicates such that the so-called natural or simple predicate green
can be defined as grue if observed before time t or bleen thereafter. Hence
even the predicate green can be shown to be disjunctive. To this extent, the
hypotheses we favour do not enjoy a special status because they are confirmed
by their instances but because they are rooted in predicates that are entrenched
in our languages as in the case of green. If grue and bleen were entrenched, we
would have favoured hypotheses of their kinds.
If projectable and non-projectable predicates are equally valid, then what
kinds of constraints can we impose on a system of inductive reasoning that will
exclude grue-type non-law-like generalizations? Goodman's response is that no
purely formal-syntactical constraints can be sufficient to distinguish projectable
from non-projectable predicates. The only way to tell apart healthy green-like
from grue-like properties is in terms of the history of past inductive inferences.
95
The reason we use green and not grue is because we have used green in our past
ind uctions. But equally, we could have been using the predicate grue rather than
green so that we could have now justified reasons to use grue and not green.
I n his radical version of the new problem of i nduction, utilizing Godel 's
incompleteness theorem, Putnam adopted and refined this argument to show
that inductive reasoning cannot be formalized -Le. , that there are no syntactical
or formal features of a formalized inductive logic that can be used to make the
aforementioned distinction.
Now since the epistemic traction of the human mind on reality ultimately
comes down to the validity of induction, the unformalizability of induction implies
that the mind-whether that of a detective or a scientist-can never even
provide a formal and hence modern scientific (namely, mathematized ) account
of itself. In other words, according to Putnam, 'if human minds have every con
ceivable computational description, it makes no sense to take the realist line that
human beings have a computational description, even though we can't discover
it'. 5 The claim that the human mind has a particular computational description or
formal model lacks explanatory power since such a computational description
would fall under Gooel 's i ncompleteness t heorem in the sense that there is no
way to correctly distinguish between the computational description of the mind
and any other computational description. Therefore, any inductive prediction
made by such a computational description can be verified save for one: that it is
the correct description of the mind. Whereas Hume and Goodman employ the
problem of induction to challenge the predictive-inductive descriptions of the
external world and events, Putnam reinscribes the problem of induction at the
level of the inductive mind itself: J ust as induction does not say anything about its
nature, the computational mind does not reveal anything about its nature either.
Since every method of epistemic i nquiry has to have an Inductive component.
this poses a serious problem for the ultimate legitimacy of our epistemic inquiry.
Putnam's resolution, i n the vein of H ume, is that even though induction cannot
be formalized , and can reveal nothing about its nature, we ought to trust our
intuitive and ordinary inductions, for they are the result of evolution (the evolution
of memory, . language, etc.) . What d istinguishes our ord inary inductions from
grue-type induction is their elegance or simplicity. But resorting to the principle
5.
J. Buechner. GOdel. Putnam, and Functionalism (Cambridge, MA: MIT Press. 2008), 27.
:::0
m
N
)>
z
m
G>
)>
:::0
m
en
-I
)>
z
....
:::0
m
0
c
m
3::
'Tl
0
:::0
0
m
-4
m
()
-I
<
m
'Tl
()
-4
0
z
of simplicity or Occam's razor is not sufficient to license our ordinary inductions
96
either, since the principle of simplicity is only a contextual and pragmatic matter
AJ
m
N
)>
of favouring one induction over another. Put differently, t here is nothing inherently
simple about the world or mind. If simplicity was sufficient to distinguish projectable
z
m
Ci)
)>
::0
m
C/J
-t
)>
z
from non-projectable predicates, ordinary induct ions from complex grue-type
inductions. false theories from true theories. then. as Adolf GrOnbaum has detailed ,
there would be no reason to endorse the Darwinian theory of life over the far
simpler theistic theory of creationism. 6 It is in this sense t hat the radicalization
....
::0
m
0
c
m
�
"Tl
0
::0
0
m
-I
m
()
of the problem of induction can be understood as the razoring of Occam's razor.
Moreover, the Putnamian appeal to evolution as a justification of our ordinary
inductions-just like Hume's resolution-simply defers the problem of induction
to a lower level, sweeping it under the carpet of evolution . To trust evolution
in order to trust our inductive capacities so as to trust the legitimacy of our
epistemic inquiries is only an act of faith in the blind god of evolution, whose
-I
<
gift of inductive reliability should not be mistaken for our epistemic birthright.
m
"Tl
()
-I
The cognitive lesson to be learned here is that only by distilling the superacid of
epistemological scepticism can we rescue the legitimacy of our knowledge and
0
z
the coherency of critical realism. Epistemology without scepticism about the
conditions of epistemic possibility is predisposed to dogmatism, and scepticism
without the rational ambitions of epistemic inquiry is doubt as debilitation. In the
end , the true detective is not someone who investigates the state of affairs by
being sure of the reliability of the methods of investigation, but one who brings
the reliability of the method of detection under the scrutiny of skeptikos or thor
oughgoing investigation. As Plato reminds us in Meno, the fact that there is doubt
is because there is a coherentist web of knowledge as its condition of possibility.
And where there is knowledge as a continuous web of truth-candidates- rather
than a collection of discrete facts or canonical given truths- methodolog ical
scepticism is an essential component of knowledge that ought to be embraced .
A realism that neither instigates the cognitive provocations of epistemological
scepticism nor is up to the task of taking on its challenges is not worth its name.
6.
A. Gri.in baum, 'The Poverty of Theistic Cosmology', British Journal of Philosophy of Science
55 : 4 (2004), 561-61 4.