Download author`s proof!

AUTHOR’S PROOF!
UNCORRECTED PROOF!
Journal of Computer-Aided Molecular Design, 0: 1–13, 2002.
KLUWER/ESCOM
© 2002 Kluwer Academic Publishers. Printed in the Netherlands.
A STUDY ON FLUORIDE SORPTION
BY MONTMORILLONITE AND
JAMES SAGE
KAOLINITE
A comparison of bacterial and human prolyl oligopeptidases: 3D-QSAR
analysis of amino
acid
and
thioxo amino
acid pyrrolidides and
TRUTH-RELIABILITY
AND
THE
EVOLUTION
OF HUMAN
MEETU AGARWAL, KAVITA RAI, ROHIT SHRIVASTAV and SAHAB DASS∗
FACULTIES
inhibitors
of prolyl
oligopeptidases
Department ofthiazolidides
Chemistry, FacultyCOGNITIVE
of as
Science,
Dayalbagh
Educational
Institute,
Dayalbagh, Agra, and examination of
India
binding sites in the homology models of prolyl oligopeptidases
∗
( author for correspondence, e-mail: [email protected], fax: 562 351845)
Ki H. Kim
(Received
4 May 2000;
accepted
25 March
2002) 100 Abbott Park, IL 60064-6100, USA
Department
of Structural
Biology,
Abbott
Laboratories,
I. accepted
INTRODUCTION
Received 27 February 2001;
24 January 2002
Abstract. Fluoride (F) sorption by acidified montmorillonite, montmorillonite and kaolinite has
been investigated as a function of period of agitation, pH, initial fluoride concentration and clay
Many philosophers
have claimed
that evolutionary
theory
providesit is
amount. In caseKey
of montmorillonite
F sorption
insignificant
at pH and
> 7, while
words: amino and
acidkaolinite,
and thioxo
aminois acid
pyrrolidides
thiazolidides, CoMFA, COMSIA, 3D-Q
good
reason
to
believe
that
human
cognitive
faculties
are
maximum at pH
4. Enhanced
F sorption
acidified montmorillonite is noted at pHreliable.
< 10, maxhomology
model,
prolyl by
oligopeptidases
and silica are
released from
acidified
imum being
at pH 6. Significant
concentration
Al3+ , Fe3+
Specifically,
they claim
that ofreliable
cognitive
faculties
enjoy
a
montmorillonite
lattice
(pH
≈
2;
1:10
w/v),
which
decreased
considerably
owing
to
the
formation
selective advantage over unreliable cognitive faculties. Thus, it is of
hydroxy (Al and Fe) silicates on subsequent base addition. Soluble fluoro-complexes at pH ≈ 2 and
Summary natural selection favors cognitive faculties that are
concluded
chemical
interaction ofthat
F with hydroxy (Al and Fe) silicates at pH > 4, are potent sink for F in case
capable
of fulfilling
thefluoro-complexes
epistemic goal
having
mostly
of acidified
montmorillonite.
Soluble
areof
also
detected
in case
oftrue
montmorillonite
3D-Quantitative
structure-activity
relationships
of amino
acid
andbeliefs.
thioxo amino acid pyrrolidides and thiazol
and kaolinite
at
pH
≈
4.
Saturation
indices
(SIs)
for
acidified
montmorillonite,
over
the
studied
In this
paper, Iofargue
that
fact that human
faculties
as inhibitors
bacterial
(F.the
meningosepticum)
and cognitive
human placenta
prolylpHoligopeptidase were investigated
range (2–10) has
also
been
computed.
CoMFA and COMSIA approaches. For the bacterial POP, the best CoMFA model obtained from 16 inhibito
have an evolutionary origin provides no reason to accept the claim
= 0.852 While
and RMSEcv = 0.213 for the cross-valid
five-component
model
the following
statistics:
R2cv beliefs.
thatclay,
they
are
reliable
withwith
respect
to generating
true
Keywords:
defluoridation,
kaolinite,
montmorillonite,
saturation
indices
2
and R = 0.954 and RMSE = 0.119 for the fitted. The best COMSIA model obtained is a two-component m
evolution might warrant the claim
that human cognitive faculties
with the following statistics: R2cv = 0.826 and RMSEcv = 0.203 for the cross-validation, and R2 =
are adaptive
(or =fitness-enhancing),
the the
claim
thatplacenta
human POP,
cognitive
and RMSE
0.142 for the fitted. For
human
the best CoMFA model obtained is ag
2 = 0.885 and RMSE
faculties
reliably
generate
true
beliefs
is
not
supported
by
an
appeal
five-component model.
This
model
has
the
following
statistics:
R
cv = 0.470 fo
cv
1.2 Introduction
to evolution.
cross-validation, and R = 0.966 and RMSE = 0.256 for the fitted. The best COMSIA model obtained is a
component
model with thean
following
statistics:
R2cv =
0.922 and
RMSEcv = 0.339 for the cross-validation
I begin
by toxic
constructing
argument
linking
natural
selection
Fluoride (F),Ra2potent
element,
enters
the
soil
by
natural,
as well
as,
anthropo=
0.959
and
RMSE
=
0.247
for
the
fitted.
A
comparison
of
the
homology
models of the human and the bac
with reliability.
I argue,
however,
that by
such an beings,
argument fails
genic sources.
Thereveals
excessive
intakeand
of this
element,
POP
similarities
differences
in the human
binding sites of results
the twointo
proteins and provides some clues f
because
it exploits anforambiguity
between two distinct notions of relifluorosis,
a disease
observedknown
selectivity. worst form of bone disorders and body crippling
ability:
fitness-reliability
and
truth-reliability.
While
it is plausible
to in
(National Research Council, 1993; John, 1998). Fluorosis
disease
is prevalent
think and
thatAfrican
our cognitive
faculties
fitness-reliable,
our dependent
epistemicon
many Asian
countries,
in areas,are
where
people are mostly
goals
require
them
to
be
truth-reliable.
After
discussing
reasons
for
Introduction
daseand
P and
prolyl carboxypeptidase, and the end
naturally occurring
groundwater for potable use (Chinoy, 1991; Teotia
Teotia,
tidase
prolyland
oligopeptidase (POP).
1991).doubting
Since, F enters
human
mainly
through
the consumption
ofbeliefs,
food
whether
truebody,
beliefs
are more
adaptive
than false
Many that
biologically
proteins
andimportant.
peptides
have to suppose
POP, that
previously called prolyl endopeptida
water, Iits
interactions
with
soilactive
components
are
conclude
natural
selection
provides
no reason
or more
prolineF,residues.
Because
of the
uniqueby its post-proline
enzyme, is a member o
Thehuman
fate one
ofcognitive
water
soluble
soils,
is largely
decided
adsorption cleaving
on
facultiesinare
truth-reliable.
proteases
are not
able (Bar-Yosef
family of etserine
clay mineralsstructure
present of
in proline,
soil andcommon
subsequent
chemical
reactions
al., proteinases and inactivated b
to hydrolyze peptide bonds before or after a proline
hibitors of both cysteine and serine proteinases
1989). Murray and Lewis (1985), found pH to be a major factor governing soil – F
residue. Therefore, specific enzymes participate in the
cleaves peptide bonds at the C-terminal side o
chemistry. Sorption of F by acidic soils is reported to be nearly five times more than
cleavage of such bonds. These proline-specific proline residues, and its activity is confined to acti
alkaline soilsteases
(Moore
Ritchie,
1988; Simard
Lafrance,
According
are and
of great
importance
in the and
regulation
of 1996).
oligopeptides
of less than 10 kD. It has an ab
to Omueti and
(1977), active
F sorption
by clay
accompanied
by release
theJones
biologically
peptides
[1].minerals
Most of isthese
requirement
for the trans-configuration of the pe
−
) ions
the mineral surface, which supports thebond
theory
of lig- proline and shows sequence sim
of hydroxyl (OH
proteases
arefrom
exopeptidases.
preceding
and exchange with
co-ordinated
investigating
F sorption
Proline
specific hydroxyl
proteases groups.
belong Further,
to different
ity with
dipeptidyl peptidase IV and acylamino
Philosophical Studies 00: 95–106, 2003.
families:
thePublishers.
N-terminalPrinted
exopeptidases
peptidase. The catalytic triad order (Ser-Asp-H
©structural
2003 Kluwer
Academic
in the Netherlands.
Water, Air,
and Soil Pollution
00: 1–15, 2002.P (APP) and dipepiminopeptidase,
aminopeptidase
POP differs from the triad order of chymotrypsin
© 2002 Kluwer
Academic Publishers.
Printed
in the Netherlands.
tidyl peptidase
II (DPP II),
dipetidyl
peptidase IV
ily (His-Asp-Ser) and of subtilisin family (Asp
(DPP IV), the C-terminal exopeptidases carboxypeptiSer). Although the action of POP on biological
PDF-OUTPUT
WEB2C
PDF-OP
Disk, CP
VICTORY: PIPS No.: 5151188 (philkap:humsfam) v.1.2
INTERLINIE/PC1/Data/wate3/DISK/new/Pipsnr.:
5087788
(watekap:spacfam)
v.1.0
philpa25.tex;
2/10/2003;
18:14; p.1
wate3127.tex; 4/07/2002; 13:23; p.1
PDF-OUTPUT
CP; DISK Gr.: 201010344, JCAM 236 (jcamkap:bio2fam)
jm407132.tex; 26/03/2002; 9:2
96
JAMES SAGE
II. EVOLUTIONARY RELIABILISM
The evolutionary reliabilist claims that evolution by natural selection shows that human cognitive faculties are reliable. Evolutionary
reliabilism is the conjunction of four theses:
K
S knows that p only if S’s true belief that p is generated
by reliable cognitive faculties.
R
Human cognitive faculties are reliable.
N
Human cognitive faculties result from evolution by
natural selection.
Cond The conditional probability of R given N is greater than
the probability of R.
In order to have reason to think that we meet the conditions for
knowledge mentioned in K, we must have reason to accept R. In the
face of this skeptical challenge, the evolutionary reliabilist tries to
argue for R by appeal to scientific findings. If any scientific finding
can support R, it is N. So, what needs to be argued is that accepting
N provides good reason to accept R. Here, then, is a preliminary
argument that connects natural selection with reliability:
1. Natural selection favors traits that increase an organism’s inclusive fitness.
2. If an organism has cognitive faculties at all, it is more conducive
to inclusive fitness to possess reliable cognitive faculties than to
possess unreliable cognitive faculties.
3. Hence, natural selection favors reliable cognitive faculties over
unreliable cognitive faculties.
4. If cognitive faculties evolved in some species during a
long period of natural selection, then the cognitive faculties
possessed by members of that species are reliable.
5. Human cognitive faculties result from a long period of evolution
by natural selection [N].
6. Hence, human cognitive faculties are reliable [R].
I take it that premises 1 and 5 are basic claims of evolutionary
biology. 3 follows from 1 and 2. 4 follows from 2 and 3. 6 follows
from 3, 4 and 5. So, the controversial part of the argument is 2,
which is the claim that it is more conducive to inclusive fitness
to possess reliable cognitive faculties than to possess unreliable
cognitive faculties. Is this claim reasonable?
philpa25.tex; 2/10/2003; 18:14; p.2
HUMAN COGNITIVE FACULTIES
97
Before we assess 2, we must understand how cognitive faculties
might increase an organism’s inclusive fitness. First, an organism’s
behaviors increase its fitness. A cognitive organism’s behaviors
are caused, in part, by the beliefs held by that organism, and
the beliefs held by an organism are generated by the organism’s
cognitive faculties. These connections are a plausible way to link
an organism’s cognitive faculties with its fitness.
Premise 2 asserts that for cognitive organisms, it is more conducive to inclusive fitness to possess reliable cognitive faculties than
to possess unreliable cognitive faculties. But 2 is true only if reliable cognitive faculties are more likely to increase an organism’s
inclusive fitness than are unreliable cognitive faculties.
Now, if by “reliable” all that is meant is that the cognitive faculty
in question generates beliefs that reliably cause certain kinds of
behaviors, and these behaviors (finding food, avoiding predators)
increases an organism’s inclusive fitness, then 2 is acceptable and
the argument succeeds. But, if reliability means nothing more than
reliably increases an organism’s inclusive fitness, then lots of traits
(both cognitive and non-cognitive) can be reliable with respect to
fitness: the zebra’s stripes are reliable, and so are the webbed feet
of Mallard ducks. Moreover, a cautious cognitive faculty that “over
detects” dangerous predators (frequently generating the false belief
that a predator is nearby) may generate an abundance of false
beliefs, though it may turn out to be adaptive because these false
beliefs increase an organism’s inclusive fitness.
Based on the preliminary argument, then, it is unwarranted to
conclude that human cognitive faculties are reliable in the sense
that they generate mostly true beliefs. And it is with this notion
of reliability that epistemologists are primarily concerned. What
is established is that human cognitive faculties reliably cause
behaviors that increase inclusive fitness. This suggests that the
argument trades on an ambiguity in the notion of reliability.
III. TWO SENSES OF RELIABILITY
The notion of reliability has two distinct senses. First, there is the
biological sense in which a trait is reliable because it increases
an organism’s fitness. Call this “fitness-reliability.” Such “fitness-
philpa25.tex; 2/10/2003; 18:14; p.3
98
JAMES SAGE
reliable” traits can be either cognitive or non-cognitive. There is
another sense of reliability that epistemologists typically employ.
Call this “truth-reliability.” A cognitive faculty is “truth-reliable”
just in case it reliably generates true beliefs. So, fitness-reliability
is a property of a trait that reliably contributes to fitness; truthreliability is a property of a cognitive faculty that reliably generates
true beliefs. We can now afford more precision regarding the claim
that human cognitive faculties are reliable:
RF
RT
Human cognitive faculties are fitness-reliable.
Human cognitive faculties are truth-reliable.
Any epistemologist persuaded by the preliminary argument has
failed to distinguish these two senses of reliability. The argument
establishes RF but fails to establish RT , which is what the evolutionary reliabilist needs to established. This distinction requires
modification of K:
K*
S knows that p only if S’s true belief that p is generated
by truth-reliable cognitive faculties
If we are to have reason to believe that we can meet our epistemic
goals, we must have reason to believe RT . The evolutionary reliabilist must now argue for RT by appeal to N. How might this
argument for RT proceed?
The first option (a) keeps the preliminary argument above, but
then argues for a connection between RT and RF . The second option
(b) offers a new argument that utilizes truth-reliability throughout
the argument.
The first option (a) argues for a connection between RF and RT
via one of two strategies. The first strategy tries to deny the distinction between fitness-reliability and truth-reliability. This strategy
is implausible, however, since numerous non-cognitive traits are
fitness-reliable. Because non-cognitive traits do not generate outputs
that carry truth values (i.e., beliefs), it follows that non-cognitive
traits cannot be truth-reliable. Since non-cognitive traits can be
fitness-reliable, the distinction between fitness-reliability and truthreliability remains intact. So the first strategy under option (a)
fails.
The second strategy under option (a) argues that the following
conditional is true: if a cognitive faculty is fitness-reliable then it
philpa25.tex; 2/10/2003; 18:14; p.4
HUMAN COGNITIVE FACULTIES
99
is truth-reliable.1 It is unreasonable for us to accept this conditional because we have good reason to believe that some fitnessreliable cognitive faculties are not truth-reliable: for example, highly
cautious belief-forming strategies can generate adaptive yet false
beliefs. Because fitness-reliable cognitive faculties can fail to be
truth-reliable, the conditional fails. So the second strategy under
option (a) fails. And thus option (a) fails altogether.
The second option (b) offers a new argument, similar to 1 thru 6
above. This new argument must establish RT , but can appeal only
to the connection between natural selection and truth-reliability.
In what follows, I shall explore, and reject, this reconstructed
argument.
IV. CONNECTING TRUTH-RELIABILITY WITH NATURAL
SELECTION
Before turning to the reconstructed argument, let us first identify the
general intuitive appeal behind attempts to provide reason for RT
based on N. The intuition is articulated (though not endorsed) by
Feldman (1988, p. 218) in what he calls a “temping argument”:
[I]f a being has beliefs at all, it is better (that is, more conducive to survival) for
it to have true beliefs than false beliefs. True beliefs about where one’s food is
are more helpful for finding food, and surviving, than are false beliefs. Similarly,
true beliefs about where one’s predators are and how to escape them are more
survival enhancing than false beliefs about these matters. So, natural selection is
likely to select for believers that have mostly true beliefs. The best way, perhaps
the only way, for believers to have mostly true beliefs is for them to have reliable
belief-forming mechanisms or strategies. Reliable mechanisms or strategies are
ones that lead mostly to true beliefs. Hence, natural selection will select believers
that have reliable belief-forming mechanisms.2
Perhaps this “tempting argument” is what Quine had in mind
when he claimed that “creatures inveterately wrong in their
inductions have a pathetic but praiseworthy tendency to die out
before reproducing their kind” (1969, p. 126). We find Daniel
Dennett asserting that “Natural selection guarantees that most of an
organism’s beliefs will be true, most of its strategies rational” (1987,
p. 75). And before rejecting evolutionary theories of content, Jerry
Fodor claimed that “Darwinian selection guarantees that organisms
philpa25.tex; 2/10/2003; 18:14; p.5
100
JAMES SAGE
either know the elements of logic or become posthumous” (1981,
p. 121).
The intuition that these philosophers share is this: if organisms
had largely false beliefs about their world, then they would fail to
navigate their world successfully, and hence they would be unlikely
to survive and reproduce. Competing organisms, who hold mostly
true beliefs about their environment, will have a selective advantage
over those who hold false beliefs. And over time, natural selection
will eliminate the less fit, leaving just those organisms who manage
to generate mostly true beliefs. Our human ancestors, it is thought,
were among the survivors of this selection process. THEY were true
believers. WE are their descendants.
So much for intuitions. Here is a reconstructed argument
connecting natural selection (N) with truth-reliability (RT ):
B1.
B2.
B3.
B4.
B5.
B6.
B7.
B8.
Natural selection favors traits that increase an organism’s
inclusive fitness.
If an organism has beliefs at all, it is more conducive to
inclusive fitness to possess true beliefs than false beliefs.
Hence, natural selection favors true beliefs over false
beliefs.
The best (perhaps only) way to generate mostly true
beliefs is to possess truth-reliable cognitive faculties.
Hence, natural selection favors truth-reliable cognitive
faculties over cognitive faculties that are not truthreliable.
If cognitive faculties evolved in some species during
a long period of natural selection, then the cognitive
faculties possessed by members of that species are truthreliable.
Human cognitive faculties result from a long period of
evolution by natural selection [N].
Hence, human cognitive faculties are truth-reliable [RT ].
Premises B1 and B7 are basic claims of biological evolution. B3
follows from B1 and B2. B4 is plausible for present purposes.3 B5
follows from B3 and B4. B6 follows from B4 and B5. B8 follows
from B5, B6, and B7. So the crucial step in the argument is B2.
Before assessing B2, it will be valuable to point out a general
objection to this kind of argument: truth-reliable cognitive faculties
philpa25.tex; 2/10/2003; 18:14; p.6
HUMAN COGNITIVE FACULTIES
101
may never have been available in human evolutionary past. If
truth-reliable cognitive faculties were never randomly generated
(say, by mutation or genetic drift), then they could not be retained
by natural selection.4 Just because cognitive faculties constructed
from fiber optics would have been (and would be now) selectively
advantageous, this cannot guarantee that cognitive faculties are
constructed from fiber optics. Cognitive faculties made from fiber
optics had to be available in order to be naturally selected. Similarly,
truth-reliable cognitive faculties may never have been available in
human history, so appealing to the supposed selective advantage
of truth-reliable cognitive faculties cannot guarantee that human
cognitive faculties are truth-reliable.
The availability objection is sufficient to undermine the claim
that natural selection guarantees that cognitive faculties are truthreliable. The availability objection is not fatal for the evolutionary
reliabilist who must show that N provides good reason to accept RT .
A guarantee is not what is sought.5
What is wrong with the reconstructed argument? The point of
contention, for the purposes of this paper, is premise B2. But B2 is
true only if the following assumption is true:
A
True beliefs are more likely to increase an organism’s
inclusive fitness than are false beliefs.
Now, if we accept A, then we have good reason to accept B2,
and the reconstructed argument succeeds (so long as we grant the
other premises). The success of the reconstructed argument, therefore, depends on the plausibility of A. In the next section, I provide
reasons to doubt A.
V. REASONS TO DOUBT A
What counts in favor of accepting A is the intuition articulated
by Feldman (quoted at length above). This intuition invites us to
imagine cases in which having true beliefs about food and water
and safe hiding places will lead to adaptive behaviors. True beliefs,
the intuition suggests, will lead to behaviors that are conducive to
fitness and so favored by natural selection. All this intuition shows,
however, is that there are some situations in which having true
philpa25.tex; 2/10/2003; 18:14; p.7
102
JAMES SAGE
beliefs is conducive to fitness. It does not establish that true beliefs
are more likely to increase an organism’s inclusive fitness than are
false beliefs, which is what the intuition needs to do in order to
support A.
Now, let’s look at how might we provide reason to doubt A.
A number of strategies come to mind. The first strategy identifies
particular false beliefs that happen to be adaptive. Stich (1990,
p. 58) provides an example along these lines: you survive a plane
crash because you had a false belief about your departure time and
therefore missed the plane. However, the fact that there are particular false beliefs that increase an organism’s inclusive fitness does
not seriously call A into question.
Another strategy identifies clusters of false beliefs that increase
an organism’s inclusive fitness. Some such clusters might be religious systems of belief. Because these various religious systems
are incompatible, it follows that many of them are false. Therefore,
individuals who accept these systems hold many false beliefs that,
nevertheless, lead to behaviors that increase an individual’s inclusive
fitness. Some of these beliefs and behaviors might include believing
the following: that hard work and clean living will glorify God, that
having numerous children pleases Allah, that cooperating others is
a form of praising Yahweh, that the use of birth control is prohibited
by God, and so on and so forth. So, inclusive fitness can be increased
by holding clusters of false religious beliefs. The abundance of
adaptive false beliefs gives us reason to doubt that true beliefs are
more likely to increase an organism’s inclusive fitness than are false
beliefs. And this is reason to doubt A.
A third strategy identifies fitness-reliable cognitive faculties or
belief-forming processes that systematically generate false beliefs.
The Garcia effect6 is an example of a belief-forming process that
generates many false beliefs that increase an organism’s inclusive
fitness. For example, an organism may hide because it believes
falsely that a predator is nearby. Evolutionarily, it pays to have
cautious belief-forming processes that “over detect” dangerous
predators, especially when false beliefs carry little cost. There are
numerous examples of fitness-reliable processes that systematically
generate false beliefs and, therefore, are not truth-reliable: believing
that all spotted mushrooms are poisonous (because consumption of
philpa25.tex; 2/10/2003; 18:14; p.8
HUMAN COGNITIVE FACULTIES
103
spotted mushrooms was once followed by illness); believing that
all strangers of the same sex are untrustworthy (because a few
strangers of the same sex have been untrustworthy); and so on.7
So, cautious belief-forming processes (say, those based on weak
inductive generalizations) can systematically generate false beliefs,
but still be fitness-reliable. This strategy shows that in some cases
having processes that typically generate false beliefs can increase
inclusive fitness. And this is reason to doubt A.
Beliefs about the colors of objects may provide another example
of adaptive false beliefs that are generated systematically. While it
may be cautious to believe that all spotted mushrooms are poisonous, this may result in the exclusion of an important food source,
and therefore decreased fitness. So, it might be beneficial to distinguish brown mushrooms with white spots from white mushrooms
with brown spots (the former may be associated with sickness;
the latter not). An organism unable to distinguish safe mushrooms
from poisonous mushrooms may avoid all mushrooms, perhaps at a
great cost. An organism with beliefs about the colors of mushrooms
might be able to track additional consistencies, say, linking sickness
with brown mushrooms with white spots. Beliefs about the colored
objects, therefore, would allow an organism to select white mushrooms with brown spots as a food source, and this may increase
inclusive fitness. Some philosophers and physicists, however, claim
that physical objects are not colored at all. If they are correct, then
all of our beliefs, to the effect that objects are colored, turn out to
be false.8 Even if these beliefs turn out to be false, still, as I have
argued, they may increase an organism’s inclusive fitness. In some
cases, therefore, a belief’s ability to increase an organism’s fitness
is independent of its truth value. Again, this strategy suggests that
fitness-reliable processes or strategies can systematically generate
false beliefs. If adaptive false beliefs are generated systematically,
then this provides some reason to think that false beliefs are in fact
likely to increase inclusive fitness. And this is reason to doubt A.
While more precise “detection” of predators and poisonous foods
might generate fewer false beliefs about predators and poisonous
foods, the biological cost of such precision might outweigh the
benefits of generating mostly true beliefs. More precise detection
requires additional biological resources. Those resources could be
philpa25.tex; 2/10/2003; 18:14; p.9
104
JAMES SAGE
used elsewhere to increase inclusive fitness, and so natural selection
might favor cognitive organisms with cautious, less precise beliefforming processes that are not truth-reliable.9 This observation
suggests another way to call A into question.
A fourth strategy, then, identifies the biological cost of having
truth-reliable cognitive faculties. Truth-reliable cognitive faculties
come at a high price: (i) the brain requires oxygen, calories, and
cooling, (ii) calculating detailed inferences (even with minimal data)
requires considerable time and concentration, (iii) accessing information from past experience requires extensive storage capacity and
retrieval pathways, (iv) identifying relevant information requires
multi-level sorting subroutines, (v) ranking desires and goals
requires extensive deliberation and reflection, and (vi) utilizing
“detectors” (and other perceptual inputs) requires precision and
acuity. Each of these factors carries a significant biological cost.
Since biological resources utilized by truth-reliable cognitive
faculties could be used in other ways to increase inclusive fitness
(ways that would confer an immediate benefit to the organism), it
follows that natural selection may favor fitness-reliable cognitive
faculties that are not truth-reliable. And again, this is reason to doubt
A.
VI. CONCLUSION
I have provided various reasons to doubt A. Because there is reason
to doubt A, B2 remains unsupported. Without support for B2, we
have no reason to accept the evolutionary reliabilist’s reconstructed
argument linking natural selection, N, with the truth-reliability of
human cognitive faculties, RT . Therefore, the reconstructed argument fails to show how N provides good reason to accept RT . And
without good reason to accept RT , we have no reason to think that
humans meet the conditions for knowledge mentioned in K*. Thus
we have no reason to accept evolutionary reliabilism. In particular,
we have no reason to think that evolutionary considerations support
the claim that human cognitive faculties are truth-reliable. And
without such reason, the evolutionary reliabilist fails to meet the
skeptical challenge that knowledge is possible, or that knowledge is
a common achievement.10
philpa25.tex; 2/10/2003; 18:14; p.10
HUMAN COGNITIVE FACULTIES
105
NOTES
1 My thanks to Tom Reed for pointing out that the evolutionary reliabilist may
claim that this formulation of the conditional is too strong. Rather, what could
be asserted is: if a cognitive faculty is fitness-reliable then it is likely to be
truth-reliable. For reasons discussed below, regarding assumption A, I think this
probabilistic rendering remains problematic.
2 It is worth noting that Feldman does not, in the end, agree with this line of
thought. See Clarke (1996) for similar assertions and statements of intuitions.
3 I think it is contentious, but I won’t challenge it here. See Feldman (1988).
4 For similar points, see Sober (1981), Lycan (1988), and Stein (1996).
5 I think the availability objection can still create difficulties for the evolutionary
reliabilist. I pursue such issues elsewhere.
6 See Garcia et al. (1972), Stein (1996, pp. 190–197), and Stich (1990, pp. 61–
63).
7 These are examples of using weak induction to form beliefs. While weak induction might not be approved by logicians or Bayesians as a basis for forming beliefs
(because, in part, such a method generates lots of false beliefs), it is not at all clear
that the systematic generation of cautious false beliefs based on weak induction is
necessarily maladaptive. Natural selection is not necessarily guided by the norms
of Bayesian probability.
8 See Hall (1996).
9 In some cases, these processes may also be “fast and frugal.” See Gigerenzer
et al. (1999) for a discussion of fast and frugal cognitive heuristics. While
Gigerenzer uses “fast and frugal” to identify cognitive heuristics that are nevertheless truth-reliable (they generate truths despite taking shortcuts), I remain open
to the view that such cognitive heuristics are not necessarily truth-reliable, though
they are fitness-reliable.
10 I would like to thank Aaron Holland, Ram Neta, Lex Newman, and Tom Reed
for comments on earlier drafts of this paper.
REFERENCES
Clarke, M. (1996): ‘Natural Selection and Indexical Representation’, in M.
Marion and R.S. Cohen (eds.), Quebec Studies in the Philosophy of Science
II.
Dennett, D. (1987): The Intentional Stance, Cambridge, MA: MIT Press.
Feldman, R. (1988): ‘Rationality, Reliability and Natural Selection’, Philosophy
of Science 55.
Fodor, J. (1981): ‘Three Cheers for Propositional Attitudes’, in Representations,
Cambridge, MA: MIT Press.
Garcia, J. et al. (1972): ‘Biological Constraints on Conditioning’, in A. Black
and W. Prokasy (eds.), Classical Conditioning, Hillsdale, NJ: L. Erlbaum
Associates.
philpa25.tex; 2/10/2003; 18:14; p.11
106
JAMES SAGE
Gigerenzer, G., Todd, P.M. and ABC Research Group (1999): Simple Heuristics
That Make Us Smart, New York: Oxford University Press.
Hall, R. (1996): ‘The Evolution of Color Vision Without Colors’, Philosophy of
Science (Proceedings) 63, S125–S133.
Lycan, W. (1988): Judgment and Justification, Cambridge: Cambridge University
Press.
Quine, W.V.O. (1969): Ontological Relativity and Other Essays, New York:
Columbia University Press.
Sober, E. (1981): ‘The Evolution of Rationality’, Synthese 46, 95–120.
Stein, E. (1996): Without Good Reason, Oxford University Press.
Stich, S. (1990): The Fragmentation of Reason, Cambridge, MA: MIT Press.
Department of Philosophy
University of Wisconsin-Stevens Point
Stevens Point, WI 54481
USA
E-mail: [email protected]
philpa25.tex; 2/10/2003; 18:14; p.12