Upload
abuabdur-razzaqal-misri
View
218
Download
0
Embed Size (px)
Citation preview
7/26/2019 Certainty and Circularity in Evolutionary Taxonomy- by David L. Hull. Evolution, Vol. 21, No. 1 (Mar., 1967)
1/17
Certainty and Circularity in Evolutionary TaxonomyAuthor(s): David L. HullSource: Evolution, Vol. 21, No. 1 (Mar., 1967), pp. 174-189Published by: Society for the Study of EvolutionStable URL: http://www.jstor.org/stable/2406751.
Accessed: 21/07/2014 17:05
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at.http://www.jstor.org/page/info/about/policies/terms.jsp
.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
.
Society for the Study of Evolutionis collaborating with JSTOR to digitize, preserve and extend access to
Evolution.
http://www.jstor.org
This content downloaded from 142.51.1.212 on Mon, 21 Jul 2014 17:05:49 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/action/showPublisher?publisherCode=ssevolhttp://www.jstor.org/stable/2406751?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/2406751?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=ssevol7/26/2019 Certainty and Circularity in Evolutionary Taxonomy- by David L. Hull. Evolution, Vol. 21, No. 1 (Mar., 1967)
2/17
CERTAINTY
AND CIRCULARITY
IN EVOLUTIONARY TAXONOMY
DAVID L. HULL
Department f Philosophy,
University f Wisconsin-Milwaukee,
Milwaukee, Wisconsin 3201
Received
October 11, 1965
After considerable eriod
of uncritical
acceptance,
he
principles
nd procedures
of evolutionary
axonomy
ave been
sub-
jected to
careful crutiny
nd, none too
surprisingly,
ave been
found not
com-
pletely dequateor totally
ree rom ague-
ness and ambiguity. ne of the more eri-
ous
criticisms
urrently
opular
s that
vo-
lutionary easoning
s
inherently
ircular.
For example, obertR.
Sokal and
P.
H.
A.
Sneath 1963) say:
"In recent
years
three comprehensive nalytic
studies
of
systematicprinciples
have
been
pub-
lished
in books
by
Hennig (1950),
Reinane
(1956),
and Simpson (1961)....
"All
three
authors
mentioned
above are
fully
aware of the
dilemma
of
circular reasoning
n-
herent in systematic procedure. They are not
satisfiedwith
solutionsbased on
'groping.'
Simp-
son
(1961)
thinks
that taxonomy
is an
evolu-
tionary
science,
and
he
attempts
to outline
a
series
of
phylogenetic rinciples
on the basis of
which taxonomic evidence
should be
examined
to
yield
evolutionary nterpretations
nd classifi-
cations.
We
shall examine
these
principles
n
de-
tail
in
Chapter
8.
However,
Simpson
nowhere
n
his book
is able to present a logical and
con-
sistentdefense
of
the
circularity
f
reasoning
n-
herent n such
procedures.By calling the process
of classification n art, ratherthan a science,he
definesthe
problem
out of
existence.
"Hennig (1950)
describesthe dilemma
n
even
greaterdetail.
He defendsthe circularity f
rea-
soning by the method
of reciprocal llumination.'
By
this
he means
that some
light
s thrown
from
one
source of
logical
illumination nto a natural
situation
kindling another, brighter ight
in the
latter,which
in turn will throw added illumina-
tion onto the first source. Thus, in a
self-
reinforcing,
ositive feedback
type of analysis,
the
relationships
nder
study are eventually
lari-
fied. Hennig feels that phylogenetic elationships
are
the entity
of
systematics
whose
parts
consist
of
morphological, ecological,
physiological, and
zoogeographic
similarities. Each of these
parts
mirrors
phylogenetic elationships,
which are
to
be investigated
by the method of reciprocal
il-
lumination.
But we cannot
see how the principle
of reciDrocal llumination iffers rom the much-
condemned vertical
construction
of
hypothesis
upon hypothesis.
"Remane (1956),
in
spite
of his
fundamentally
phylogenetic
orientation,has also realized that
phylogenetic easoning cannot serve
as the basis
for
erecting
natural
system.
He
similarly ejects
affinitybased
on
a
few
characters)
as
the
basis
of
a natural
classification.
He
considers
hat
while
both of these approaches enteron occasion into
the
techniques practiced by 'good systematists,'
the exclusive
application
of
only one of
them is
likely to lead to misclassification.
Affinity r re-
semblancewhen based on one or a
few characters
can
lead the systematist stray,
Remane claims,
as
he
would be
too easily deceived by
chance con-
vergences
resulting from poor sampling of the
characters.
Remane attempts escape from the
circulus
vitiosus by basing his taxonomyon non-
phylogenetic
riteria of homology."
The charge of vicious circularity as
been commonthroughout he
history f
taxonomy.
For example,Darwin (1859)
accused
his
predecessors nd
contempo-
raries of
arguing n a circle
when they
claimed
that mportant rgansnevervary
and
decided
whichorganswere
mportant
by which
did
not vary. Sachs
(1890) re-
peated the
charge nd directedt specifi-
cally at
Linnaeus. It was only
matter f
time ntil heevolutionistshemselves ere
accused of
reasoning
n
vicious
circles.
Thompson 1952) seemsto
have been the
first o do so. The charge,
however, as
been
made
mostoften
nd
most
forcefully
by
A.
J.
Cain.
Initially
t
was
directed
against Cuvier;
Darwin was
expressly
exempted
1959a).
Cuvier
lassified
iving
organisms
y
means
of
properties
rdered
according
o
their
presumed
hysiological
importance.Cain allegesthat Cuvierde-
cided
which
roperties
ere
physiologically
important
y observingwhich
characters
were constant
n a
classification
lready
constructed
ccording
o overall
imilarity,
making
phenetic lassification
undamen-
EVOLUTION 21: 174-189. March, 1967
174
This content downloaded from 142.51.1.212 on Mon, 21 Jul 2014 17:05:49 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp7/26/2019 Certainty and Circularity in Evolutionary Taxonomy- by David L. Hull. Evolution, Vol. 21, No. 1 (Mar., 1967)
3/17
CERTAINTY
AND CIRCULARITY
IN TAXONOMY
175
tal.
Darwin,on
the
other
hand, proposed
that iving
rganisms
e classified
ymeans
ofproperties
rderedccording
o
their re-
sumedphyletic mportance. n principle
which
roperties
re
phyletically
mportant
and, hence, ndicative
f
evolutionary
e-
scent
can be
discovered
ndependently
f
the
constancy
f
characters
n
any pre-
established
classification.
Cain
(1962a)
summarizes
is views s
follows:
"By equating
taxonomic with
ancestral
rela-
tionship, arwin,
ike Cuvier, dopted
two criteria
of the importance
of
characters.
On the
one
hand, he said that those characterscommon
to
large groups
(i.e.
natural
groups,
although
he
did
not
say
so
explicitly)
are
more
important
than
those common
to groups
containing ittle
diversity.
This is
always
right
if
only
natural
groups
are to be
made. But
he also said
(a) that
those least likely to
have
been modified
n rela-
tion to
particular
modes of life will
be more im-
portant
in showing ancestral
affinity,
nd (b)
that they can be recognized
because
they
will be
the
most constant
ones within
a
natural
group.
But this
assumes that natural groups
are always
phyletic
or, in
other terms, that
convergence
s
never so great as to obscure or outweigh an-
cestral
resemblance ven
in
poorly
known
groups.
He did
not commit he earlier
error
of arguing
n
a circle. His
principles
f
evolutionary
mportance
were not derived
from
a
pre-existing
axonomy,
but from the results of
artificial selection
and
from
he
study
of
heredity, ariation
nd
ecology.
This is a
point
worth
emphasizing. In
many
elementary
extbooks of
biology,
classification s
treated
as one
of
the
lines
of
evidence
for
evolu-
tion.
Darwin
did not treat
it
thus;
he discussed
it quite
late in the
Origin'
as consonant
with
the
theoryof evolution,and explicable as a conse-
quence
of
it.
He
never regarded
it
as
primary
evidence
for
evolution,
and
his
caution
was cer-
tainlv
iustified."11
But
later Cain
brings
the
charge
of
vicious
circularity
ull circle.
In
principle
phyletic
development
an
be
discerned
without ecourseo the constancyf prop-
erties
n
a phenetic
lassification
ut
too
often
n
practice
t cannot.
Cain
(1962b)
concludes:
"If there
were any
method
of analysing
ani-
mals so that
the
construction
nd
behavior
of
each
could
be
shown
to follow
from a
few great
principles,
hen ideally
each form
could be ex-
pressed
n terms
of these
principles
n such
a way
as to define
it
and
its properties
completely.
Something
f this
sort
has
happened
n the classi-
ficationof the elementsby their atomic struc-
ture.
Such
a taxonomy
of analyzed
entities
has
been
attempted
many
times. In
Linnaeus's
period
and before, logical
analysis, plus
Aristotelian
physiology,
was
thought
to help
in this.
About
the
beginning
f
the nineteenth
entury,physio-
logical criteria
of what
must be
the
most
im-
portant
characters
were widely
used.
But in
both
periods
there
was
far too
little
information
or
such analytical
taxonomy
to
be attempted,
nd
workerswere in fact
arguing
n a circle
from the
observed constancy
of
certain
characters
n al-
ready recognizednatural groups'to theirphysio-
logical
importance.
Darwin
rightly
ejected
such
attempts,
but
he
tried to
recognize
which
char-
acterswere
more ikely
to remain
constant
during
evolution
and
to use them
as
the best
indicators
of ancestry-and
his criterion
gain,
n
the
bsence
of a
really good
fossil
record,
ould
only
be con-
stancy
within
natural
groups.'
2
1
Darwin
discussed the
hierarchical
arrange-
ment
of taxa quite early n the
Origin
1859: 128)
and
treated
t
as he treated
all evidence
for evo-
lution,
as consonant with his
theory. The group
subordinate to group
relation evident
in nature
was just
one more fact nexplicable
n the
special
creation view but
to be expected
on his view.
Cain is right, owever, hat he did not treat clas-
sification
as primary evidence
for evolution, as
say George J.
Romanes (1892)
was to do
later.
As he
did with
so
many
of the
arguments
put
forth to justify
evolutionary theory,
Fleeming
Jenkin
(1897)
found a
hole in this one.
Initial
success n a rough
hierarchical
lassificationmight
be evidence
for
evolution,
but
just
the
opposite
should be the case
under continued
efforts.
He
argues
cogently
and
in agreement
with
Darwin)
that
evolution
by
gradual
change
entails
diffi-
culty,
not
success,
n classification.
Hence,
if
any-
thing, ontinueddifficultyn classification
hould
be considered
vidence
for
evolution.
2
Cain's
distinction
between
classifications
of
analyzed
and
unanalyzed
entities
reflects
n
ac-
ceptance
of the
Aristotelian
iew
of the
relation
between
mathematics
nd
the empirical
sciences
which
relativity
heory
has
made untenable.
The
reason
that
essential
definitions
n
the
Aristotelian
manner
re
possible
n geometry
s that the
vari-
ous
geometries
re pure
deductive systems
with
no empirical
import.
Such "entities"
as pure
Euclidean
triangles
can
be
completely
analyzed
into
the species
scalene,
isosceles
and
equilateral
because
there
is nothing
to analyze
but
some
axioms
and
definitions.
uch
is not the case
with
any
of
the
entities
n
empirical
cience.
The
clas-
sification
of
the physical
elements
according
to
their
atomic
numbers
s not
in
the
least
like a
taxonomy
of analyzed
entities.
One
end
of the
periodic
table
is
open.
There are
about
a hundred
elements
table enough
to
exist
for
any length
of
This content downloaded from 142.51.1.212 on Mon, 21 Jul 2014 17:05:49 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp7/26/2019 Certainty and Circularity in Evolutionary Taxonomy- by David L. Hull. Evolution, Vol. 21, No. 1 (Mar., 1967)
4/17
176
DAVID
L.
HULL
Sokal
and
Sneath,
nd Cain
have
been
quoted
at some
length
to make
clear
exactly
which
ines
of
reasoning
ommon
in evolutionary
axonomy
re supposed
o
be
viciously
ircular.
According
o Sokal
and
Sneath
no logical
and
consistent
e-
fense
has
been
presented
orsuch
"circu-
larity
n
reasoning."
t
willbe the
purpose
of
this
paper
o provide
ust
such
defense.
By the
very
nature
f the
accusation,
he
defense
will
have
to
be
in terms
f
logic
and logical
distinctions.
n
the first
half
of
the
paper,
t
will
be
explained
what
con-
stitutes logicalfallacy nd why t is un-
desirable
to
reason
fallaciously
nd then
several
inesof
reasoning
hich
have
been
termed
iciously
ircular
willbe
examined
to
see
if
they
ctually
re circular.
t will
be
seen
that
these
ines of
reasoning
e-
generate
nto
vicious
circles
only
if cer-
tain
evolutionary
aws
and principles
re
ignored.
he
ustification
orgnoring
hese
laws
and
principles
s
that
they
are
not
warranted.The secondhalf of thepaper
will
deal
with
heproblem
f
what
n
gen-
eral
makes an
inductive
nference
war-
ranted
nd specifically
hether
hose
in-
ferences
sed
in
evolutionary
econstruc-
tions
re warranted.
LOGICAL
FALLACIES
Since evolutionary
rocedure
s
accused
of being
circular
nd
reasoning
n
vicious
circles s a logical fallacy, t is certainly
worthwhile
o explain
what ogical
fallacies
are
and
why
it
is undesirable
o
commit
them.
The
general
tatement
hat
logical
fallacy
s
any
erroneous
rocess
f
reason-
ing or
arguing
s not veryhelpful.
What
is important
re
thereasons or onsidering
arguments rroneous.
They are
of two
types.
Logicians re mainly
oncerned
ith
formalfallacies,those lines of reasoning
which
re erroneousolely
because
of the
form f
the
propositions
nd arguments,
regardless
of the subject
matter. For
example,
t is fallacious o arguethat
ince
Communists isapprove
of abstract
art,
anyonewho disapproves
f abstract rt is
a Communist.
person eed
knownothing
of
either
ommunismr
abstract rt
to de-
cide that this arguments invalid. The
fallacy f reasoning
n vicious
circlesdoes
not
belong
o this lass of
fallacies.
nstead
it is
an example
of what logicians
all a
materialfallacy.
In diagnosing
material
fallacies othcontentnd
the use to
which
the argument
s
being
put play
central
roles.
An
argument
an
fulfill ll the re-
quirements
f formal
alidity
nd
still
be
fallacious
f
t
fails o
perform
he
task
t is
intendedo perform. ne of the common-
est uses of
arguments
s
to prove
conclu-
sions
th'at
re
in
some way
unknown r
doubtful
or that have
been called
into
question.
A
prerequisite
or
fulfilling
his
purpose
s
that the argument
annot
as-
sume
n
the
premises
what
t proposes o
prove.
An
argument hich
fails to prove
anything
because it somehow
takes for
grantedwhat
it is supposed
to prove
is
calledbegginghe question rpetitio rin-
cipii.
More
subtleversions
f this fallacy
are
often
ingled
out
and
termed
icious
circles.
An
argument
an
presuppose
r
depend
upon
its
conclusion
n
two
ways.
In
the
simplest
ase
the dependence
s
straight-
forwardlyogical.
One of
the
premises
s
just
a
restatement
f
theconclusion
o
that
anyone
doubtful f
the truth f
the
con-
clusionwouldhave to be equallydoubtful
of
the truth f
the
premises.
For
example,
any proposition
an be deduced
validly
from
tself,
ut we
have
not
thereby
roved
anything.
t
is
formally
orrect
o
con-
clude that
the Pope
is
infallible
when
he
speaks
ex cathedra
rom he
premises
hat
time
under ordinary conditions;
there
are three
and
only
three species
of
triangles
on the
dif-
ferentia
mplied.
Further,
he
superficial
eatness
of
the periodic
table
is
destroyed
y the
existence
of isotopes
and
isomers.
No
such borderline
ases
can
occur in
a
pure
deductive system
ike Eucli-
dean geometry.
The entities
classified
in the
periodic
table have
been more
thoroughly
na-
lyzed
than have
most
of the
entities
lassified
n
the
Linnaean hierarchy,
but there
is not
the
slightest
ope
of ever
completely
nalyzing
either
of them,
n the sense
in
which
the "entities"
of
pure
geometry
can be
completely
analyzed.
This content downloaded from 142.51.1.212 on Mon, 21 Jul 2014 17:05:49 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp7/26/2019 Certainty and Circularity in Evolutionary Taxonomy- by David L. Hull. Evolution, Vol. 21, No. 1 (Mar., 1967)
5/17
CERTAINTY
AND CIRCULARITY
IN TAXONOMY
177
he said so
ex cathedra nd
everythinghat
the
Pope
says ex cathedra s true, ut such
an
argument ould stillbe
fallacious.
The dependence eedn't e
logical,how-
ever. Often the circle s epistemological.
The only
way thatyou couldknow hat he
premises
re true would be
to know that
the
conclusions true. All cases of deduc-
tion from enuine
numerative eneraliza-
tions
(generalizationsrrived t by com-
plete enumeration) re
examples f argu-
ing in
vicious circles. For
example, t is
formally
orrect o argue hat
certain all
in a jar is redbecause ll oftheballs n the
jar are
red, but such an
argument ould
be
viciously ircular
f
the generalization
had been
established only through he
examinationf each of theballs n the ar.
The only
way thatyou couldknow hat ll
of
the
balls
werered wouldbe by knowing
that the
ball in questionwas red. How-
ever,
rguing hat
Tamias
striatus volved
because all species volved s
not an exam-
ple of a viciouscircle. The generalization
was not
arrived t
simply y
enumerative
induction.
ome specieswere
examined
o
establish he
truth
f
the
generalization
o
be
sure
and
many
have been examined
since,
but the
acceptance
f the
generaliza-
tion
rests
primarily
n the
explanatory
power
of
evolutionaryheory.
HOMOLOGY
AND
PHYLOGENY
Thereare three inesof reasoningom-
mon
in
evolutionaryaxonomy
which
are
often
ingled
ut
as circular:
a)
the de-
fining f
"homology"
n
termsfphylogeny
and then
using
characters
laimed
to be
homologous
o
infer
phylogeny, b)
the
basing
of an
evolutionary
lassificationn
a
phenetically
onstructed
lassification,
and
(c)
the
inferring
f
phyletic
escent
from verall
phenetic imilarity.
he
first
two inesofreasoningrederivative f the
third.
They
shall be discussed
n
order.
Sokal
and
Sneath
1963) say
with
respect
to the first:
tor, by
the
common definition
f the term), or
to
decide a priori
which
characters
re important
r
are reliable
guides
to phylogeny,
oon leads
to a
tangle
of circular
arguments
from
which
there s
no escape. Even Simpson (1961), who
strongly
supports a
phylogenetically
ased
taxonomy,
s
aware
of and points
out the circulus
vitiosus
of
this procedure."
Sokal and
Sneath's
eference
o thedefi-
nition f
"homology"
n
termsf
phylogeny
mightmake
t
soundas
if
theywere
rgu-
ing that
uch
a definition
s
circular,
hen
their ntent eems
o be
that
the definition
can lead to
circular
rguments.
s Ghiselin
(1966a and 1966b) has pointedout, the
definition
f "homology"
n terms f phy-
logeny
s not
circular
ecause
"phylogeny"
is notdefined
n terms
fhomology.
okal
and Sneath's
point
s that
one is often
n-
ferred ia
the other
nd that
this
can
lead
to
circular
rguments.3
n
another
place
they
make t
clear
thatthey
havean
argu-
mentn mind
ndthat
he rgument
s sup-
posed
to be epistemologically
ircular.
Theysay:
"We
do
not know
of
any
infallible riteria
for
overall phenetic
convergence
that
may
be
ob-
tained
from
study
of
living
formsof
organisms
alone.
To
detect
convergence,
we have
to
dis-
tinguish
those
features
which
do
accurately
re-
flect
the
phylogeny
romthose
featureswhich
do
not. This,
however,
s
a
question
which can
only
be
answered by
knowing
the
phylogeny
first.
The
problem
therefore
s
insoluble within
this
logical
framework,
nd
one
must have
ndependent
evidence
(not
derived
from
phenetic
relations)
in
order to attack it."
It
is tautological
o
say
that
homologous
resemblances
re
indicative f
common
ine
ofdescent,
ince
by
definition
omologous
resemblances
re those
esemblances
ue to
"Any attempt to decide the phylogeny
n one
set
of
characters, n particular those believed
to
be homologous (derived
from
a
common ances-
3The
difference
etween
the
two
assertions
s
worth
pointing
out, since
there
is
considerable
difference
etween
a definition
nd
an
argument.
The
reasons
for the undesirability
of
circular
definitions
nd circulararguments re, however,
the
same.
If you
don't
know
what
the
word
"ex-
planation"
means,
it
doesn't
help
to be
told
that
an explanation
s something
which
explains
some-
thing.
Similarly,
f you
are
in doubt
of
God's
existence,
he
argument
that
if
God
is
all-good
and
existent,
then
He
exists won't
prove
very
convincing.
This content downloaded from 142.51.1.212 on Mon, 21 Jul 2014 17:05:49 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp7/26/2019 Certainty and Circularity in Evolutionary Taxonomy- by David L. Hull. Evolution, Vol. 21, No. 1 (Mar., 1967)
6/17
178
DAVID
L.
HULL
common
ineof descent.
To be sure,
any
evidence
o the effect
hat a particular
e-
semblance s homologous
ould
necessarily
be evidence
o theeffect
hat
t was due to
common ine of descentand vice versa.
But,the evidence
rom
which hylogeny
s
inferred
s not limited
ust to homologous
resemblances.
or example, wo
axawhich
in
point
of fact are
very closelyrelated
cladistically ould exhibit
striking
is-
similarity.
ny
evidence e.g.,
genetic
vi-
dence)
to the
effect hat this
particular
type f
dissimilarityan
arise
uitequickly
wouldcontributeo thecorrect econstruc-
tion of the
phylogeny
ut would
not
de-
pend
on homologies
t
all.
Fossil
evidence
that
it
did
so
arise
would
be even
more
conclusive.On
the other
hand,
evidence
that the genetic
mechanism ecessary
or
such
dissimilarityo
arise
s
such
that
he
change
could arise
only
very gradually
would
disprove
he
hypothesis
hatthe
two
share
a recent ommon
ncestry ven
in
the absence of fossil evidence. But too
often
the
only evidence
that
is
readily
available
is
phenetic
imilarity,
nd the
first
ine
of reasoning
hich
as been
called
circular
merges
nto
the third.
VICIOUS
CIRCLES
AND SUCCESSIVE
APPROXIMATION
The
second
ine of reasoning
which
has
been
termed
ircular
s
spelled
out
in
some
detail by Sokal and Sneath (1963) as
fnl
MAxIc
fined
on
the basis
of
this charactercomplex, ay
X. It
is
assumed
that taxon A
is
monophyletic
r
a
'natural' taxon.
Thus
every
member
f A
(both
known and
unknown
forms)
s
expected
o
possess
X; conversely, ossessionof the character com-
plex
X
defines
A.
"Henceforth roup A, as definedby X,
assumes
a
degree
of
permanence
nd
reality quite
out
of
keeping with the tentativebasis
on which
it
was
established.
Subsequently
studied
species
are
compared
with
A to
establish their
affinities;
they may be
within
A,
close to
it,
or
far
from t.
It
is
quite possible that a species
not
showing
X
would be
excluded
from
A, although
t
was closer
overall to most of
the
members
of
A than
some
were
to each other. It
may
be said
that
such
problemswould arise only when A was an 'arti-
ficial' group erected on the basis of 'unsuitable'
characters. However, except
in
long-established
taxa
or
those separated by very
wide
gaps from
theirclosest relatives, he effect f the last classi-
fication carried out with a limited number of
characters
s
quite pervasive. The circularreason-
ing arises from the fact that the new characters,
instead of being evaluated on their own merits,
are inevitablyprejudiced by the prior erectionof
taxon
A
on other characters X). Such prejudg-
ment
gnores
he
fact that the existence f A as a
natural
(or 'monophyletic') group defined by
charactercomplexX has been assunmed nd not
demonstrated."
In
all due respect, heprecedings not a
characterizationf thebest n evolutionary
taxonomy ut a parody f the worst. Per-
haps the practice of some evolutionists
sometimes egenerates o this evel,but it
need not and if the principles f evolu-
tionary axonomy re given proper con-
sideration,t shouldnot. Sokal and Sneath
are well awarethat ven themost arefully
formulated rocedures un the dangerof
unimaginativepplication
f
the principles
on
which
hey
re
based have
been
mper-
fectly nderstood.
he
principles f evolu-
tionary axonomyhouldnotbe judged ny
more
by
their
misapplication
han should
those
of
numerical axonomy.There
are
linesof
reasoning
n
evolutionaryaxonomy
which omeclosetobeing ircular. n the
preceding uotation okal
and Sneathhave
extracted
ne
of the
most
important
f
these ines of
reasoning,
ut their
descrip-
tion
tends
to
disguise
rather han reveal
the outlines
f this
potential
ircle.
For
example, ccording
o
their
ccount,
"It may
be advantageous
at this stage
to
out-
line an important
ogical
fallacy underlying
ur-
rent taxonomic
procedure.
It is the
self-reinforc-
ing circular
arguments
used to
establish
cate-
gories
[taxa],
which
on
repeated
application
in-
vest the
latter
with
the
appearance
of
possessing
objective
and
definable
reality.
This
type
of
rea-
soning s,
of course,
not restricted
o
taxonomy-
but it
is no less
fallacious
on that account.
Let
us illustratethis point. An investigators faced
with
a group
of similar species.
He
wishes to
show
relationships
mong
the
members of
the
group
and is looking
for characters
which
will
subdivide
t into
several
mutually
exclusive
taxa.
A search for
characters
eveals
that
within
sub-
group
A certain
characters
ppear
constant,
while
varying
n an uncorrelated
manner n
other sub-
groups.
Hence
a taxon
A is described
and de-
This content downloaded from 142.51.1.212 on Mon, 21 Jul 2014 17:05:49 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp7/26/2019 Certainty and Circularity in Evolutionary Taxonomy- by David L. Hull. Evolution, Vol. 21, No. 1 (Mar., 1967)
7/17
CERTAINTY
AND
CIRCULARITY
IN TAXONOMY
179
afterA
has been
defined
n
terms f
X,
any species
which
has
a
sufficient
umber
of the
properties
n X
is
automatically
placed
in
A
(although
t might
not be
similar n otherrespects o theother pe-
cies
already
n
A) and
any species
which
lacks a
sufficient
umber f these
proper-
ties
is
excluded
from
A
(although t
may
be
similar
n
other
respects
o the
other
species). The
classificationnd reclassifi-
cationwhich
goes
on all the
time
n
evolu-
tionary
axonomy
n
the light
of
the dis-
covery f
previously
nknownpecies
and
additional evidence belies this extreme
position.
However,
what
s
wrong
with
his
procedure,ven
f
evolutionary
axonomists
were
guilty
f
t,
s
not that
self-reinforc-
ing circle
s involved ut
that
no
reinforce-
ment s
involved.
f
anything,
healleged
procedure s
contradictory. n the
one
hand,
evolutionary
axonomists re
sup-
posed
to
reasonthat the
degree
f
covari-
ance
in
X
is
indicative
f phyletic imi-
larity. On the otherhand,they are sup-
posed
to maintain
hat
n
the
ight
f
more
evidence nd
a
slightly ifferentistribu-
tionof
characters, hat this
newcomplex
Y
is not
indicative f phyletic
imilarity.
The
wholepoint
of
evolutionary
rocedure
is that
f
X is
indicative f
phyletic
imi-
larity, henY is
more
ndicative.
Now
in
ordinary
discourse such
self-
reinforcing
roceduresare
often called
"vicious circles" f the consequences re
undesirable. or
example,
n
a
recent ssue
of a
popular
magazine
t
was
reported hat
doctorshad
"found that the
moreover-
weight
he
diabetic
gets, the
more nsulin
there s
in
his
blood.
And
the
morensulin,
the
more
he
tends
o eat and
thus
toreup
more fat in
an
ever-widening
icious cir-
cle."
The
situationmay be
vicious,
ut it
is not
circular
incethediabetic
gets
fatter
and fatter. imilarly,volutionaryaxono-
mists
lassify
nd reclassifyn
an attempt
to
represent
volutionary
escent
with an
ever
ncreasing
ccuracy.4
This
goal
may
be
undesirable.
n this
sense,
he
practice
is
vicious,
ut t
is not
circular
nd,
hence,
not
logically
fallaceous.
The
criticism
t
issue
s whether
r
not
evolutionary
axon-
omists re making logicalerror, ot an
error
n
tactics.
As Sokal
and
Sneath
say, the type
of
reasoning
hich
hey
have
indicated
s
not
restricted
o
taxonomy.
t is inherent
n
any
attempto
obtain
bjective
nowledge,
including
heefforts
f
the
numerical
ax-
onomists.
Abraham
Kaplan
(1964)
has
called
thisspecial
problem
he paradox
of
conceptualizationnd considerst an exis-
tential
ilemma:
"The
proper
concepts
are needed
to
formulate
a
good
theory,
ut we
need a
good
theory
o
ar-
rive at
the
proper
concepts.
Long
before
the
scientific
revolutions
of the twentieth
century,
Jevons
(1892)
remarked
hat almost
every
classi-
fication
which
is proposed
in
the early
stages
of
a science
will
be
found
to break
down
as the
deeper
similarities
f the objects
come
to
be de-
tected.'
Every
taxonomy
s
a
provisional
nd
im-
plicit
theory
or
family
of
theories).
As knowl-
edge of a particular subject-mattergrows, our
conception
of
the
subject-matter
hanges;
as
the
concepts
become
more
fitting,
e
learn
more
and
more.
Like all
existential
ilemmas
n
science,
of
which
this s
an instance,
he
paradox
is
resolved
by
a
process
of
approximation:
the
better
our
concepts,
he better
the
theory
we
can
formulate
with them,
and
in
turn,
the better
the
concepts
available
for
the next,
improved
theory.
V.
F.
Lenzen (1938)
has
spoken
explicitly
f successive
definition.'
It is only
through
such
successions
that
the
scientist
an
hope ultimately
o
achieve
success."
There
are
several
ossible
ways
to
avoid
the
dilemma
f which
Kaplan
speaks,
none
of which
have
engendered
ny
great
en-
thusiasm mong
cientists.
he
easy
way
s
to
introduce
ome
metaphysical
aculty
such
as
intuitionwhich
s capable
of
di-
rectly
nd
infallibly
pprehending
eality.
This
was
Aristotle's
ay.
At least
onecon-
temporaryaxonomisttill advocates uch
4 The claim
that
classification s to
represent
phylogeny s merely
he claim that there
must
be
some systematic
relationship
between phylogeny
and
some
system
of classification.
The
more
powerful
the
system
of
classification,
he more
extensive
and precise
this
relationship
can
be
made. Unfortunately,
he Linnaean
hierarchy s
not a very
powerfulmeans of classification.
See
Hull (1965) for further
etails.
This content downloaded from 142.51.1.212 on Mon, 21 Jul 2014 17:05:49 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp7/26/2019 Certainty and Circularity in Evolutionary Taxonomy- by David L. Hull. Evolution, Vol. 21, No. 1 (Mar., 1967)
8/17
180
DAVID
L.
HULL
a procedure
Sattler,
963).
Most philoso-
phersand
scientists,
owever,
ave long
since
come
to
the
conclusion
that
this
"solution"ustwon'tdo for cience
n
gen-
eral.
Cain
(1958)
and
Mayr (1959)
have
argued
quite
effectively
hat it
won't
do
for taxonomy
n particular.
(See also,
Hull,
1965.)
Another
lternative
s to deny
that a
classification
s
intended
o
approximate
anything
ut
thatany
classification
stab-
lishedon objective
riteria
s "true"
n its
ownright.
At times, umerical
axonomists
haveseemed obe arguing or hisposition.
At other
times,
however,
hey
maintain
that phenetic
lassification
s
to approxi-
mate something
alled
phenetic
imilarity.
On
this
view,
heevolutionists
nd
numeri-
cal taxonomists
hare
th'e
methodological
problem
f
justifying
he
process
of
ap-
proximation
y
successive
definition,
he
process
alled
"groping"
y
Cain and
"re-
ciprocal
illumination"
y
Hennig.
This
justifications no easy matter,but it
should
be kept
n mind that
theproblem
is one
shared
by
evolutionary
nd
numeri-
cal taxonomists
qually.5
To be
sure, ny
classification
ased upon
a
previous
tentative
lassification
s
in-
evitably
rejudiced
y
it. Any
error
n the
early
classification
might
well
infect
all
laterreclassifications,
utthe
effect
f
the
early
classification
ecreases
s
reclassifi-
cation akesplaceinthe ight fadditional
evidence.
A
gradual
decrease
in
error
might
not
sound
good
enough
for
some
taxonomists,
ut
there
s
no
other
lterna-
tive short
f complete
eclassification
ach
time
a
single
taxonomic
oundary
s
re-
evaluated.
Although
Sokal
and
Sneath
have advocated
some
rather
xtensive
e-
visions
n
taxonomic
rocedure,
venthey
have
notgone
o
far
s
tosuggest
he
com-
plete dissolution f all classificationnd
reclassification
e
novo
each time
single
taxon
is re-evaluated
n
the
basis of
a
single bit of additional
evidence.
Until
such a
drastic tep
s taken, arlyclassifi-
cationswill color
later classifications
e
theyphenetic
r phyletic.
VICIOUS CIRCLES
AND
EMPIRICAL
CERTAINTY
There
remains
he primary
ine
of rea-
soning,
whichhas been termed ircular,
o
defend. Once again
we turn o Sokal
and
Sneath
for
a
vigorous
tatement f
the
criticism:
"The
difficulty
ith the use
of
the
phylogenetic
approach in systematics mergedafter the first
wave
of
enthusiasm
for it
subsided
and
has
re-
mained apparent
to perceptive observors ever
since. We
cannot make
use of phylogeny for
classification, ince
in
the vast majority
of cases
phylogenies
re unknown.
This is one of the state-
ments
most commonlyheard at
meetings f tax-
onomists,yet it
is
most
consistently
gnored. Let
us
restate
t in other words for emphasis.
The
theoretical
principleof descent with
modification
-phylogenetics-is
clearly esponsible
or theexis-
tence and
structure f a natural classification;
we
may even agree with
Tschulok (1922) that the
natural systemcan be considered s proof of the
theory
of
evolution.
However, since
we have
only an
infinitesimal
portion of phylogenetic
history n the fossilrecord,
t is almost
mpossible
to establish
natural taxa
on a phylogenetic
asis.
Conversely, t
is
unsound
to derive a definitive
phylogeny rom a tentative
natural
classification.
We
have
described his fallacy of circularreason-
ing earlier."
[See also
Cain, 1962; Cain and
Harrison,
1960; and Bigelow, 1956
and 1958.]
This
quotation
ontains wo
distinct
riti-
cisms fevolutionaryaxonomy. hefirst,
which
has
already
been
discussed,
oncerns
the erection
f a
phyletic
lassification
n
the
basis
of
a
phenetic
lassification.
he
effect
of
the
original
classification
may
never
be eliminated
as reclassification
takes
place.
Since
reclassification
oes
take
place
and
the
aterclassifications
el-
dom
circleback on
the
original,
his
pro-
cedure
s not
circular.
Definition
y
suc-
cessive pproximationrovideseveralogi-
cal
difficulties,
ut none of
these
nvolves
vicious
ircularity.
he second riticism
s
that
inferringhyletic
elationships
or
a
group
n
the absence
f
a
reallygood
fossil
record
for
that
group
s
unwarranted.
n
most nstances,
o
fossil
ecord
s
available.
I
Sokal
and
Sneath
object
only
to
Hennig's
particular
use of the process
of
successive
defini-
tion,
the process
which
he
calls reciprocal
llumi-
nation,
not
to the process
tself.
This content downloaded from 142.51.1.212 on Mon, 21 Jul 2014 17:05:49 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp7/26/2019 Certainty and Circularity in Evolutionary Taxonomy- by David L. Hull. Evolution, Vol. 21, No. 1 (Mar., 1967)
9/17
CERTAINTY
AND CIRCULARITY
IN TAXONOMY
181
The only
observations hichevolutionists
have to go on are
pheneticproperties
(usually
morphological),nd phenetic
imi-
larity s
exactly hecriterion sed
by clas-
sificationistsnd numericalaxonomistso
construct
heir lassifications.
f an evolu-
tionary
axonomisteginswith
a phenetic
classification,eclassifies
xclusively
n the
basis of phenetic imilarity,
hen he will
end up with phenetic
lassification
nd,
perhaps,
lso a phyletic
lassification,e-
pending
n how goodphenetic
imilaritys
at indicating hyletic
imilarity. f phy-
leticrelationshipsannotbe inferred rom
phenetic similarity
with sufficient
er-
tainty,
then the
line
of reasoning ust
sketched
s unwarranted.
t
is
plainlynot
circular
These
two
criticisms, hough
distinct,
are related. They
are distinct,
ince
it is
one
thing
o arguethat a
line
of
reasoning
is circular; t is another
o arguethat t is
unwarranted.
n
fact,
the two errors re
mutuallyxclusive.No one argumentould
be
both circular nd
unwarranted.
The
two
criticisms re related,however,
ince
those
ines
of reasoningwhich
have been
termed ircular ecome ircular nly
f
cer-
tain other
ines
of reasoning an
be shown
to
be unwarranted. hus,
the
really
m-
portant
riticismf evolutionaryaxonomy
does not concern ircularity
n
classifica-
tion
t
all but the
ustification
f
inferring
phyletic evelopmentromphenetic imi-
larity n the basis
of certain
volutionary
principles
n
the absence
of a
fossil
ecord
for
the
group.
If
a good
fossilrecordfor
the
group
n
question
s necessary efore
its evolutionaryevelopment
an be recon-
structedwith
sufficientertainty,
hen n
a vast
majority
f
the
cases
phylogeny
an-
not be reconstructed
nd
the evolutionary
program
n
taxonomy
ecomesuntenable.
It will be argued hatfossil ecords or ll
or evena
majority
f the
taxa to be classi-
fied re
not necessary.
All
that
s
required
is that
there be
some
reallygood
fossil
sequences
from
which
to derive
principles
concerning
he
trends nd
tendencies f
evolutionary
evelopment
nd
others
which
can be used
later to check
them. Further,
the evolutionary
rinciples
sed to
infer
phyletic
evelopment
eedn't
e completely
general
universal
tatements.
Much less
willhave to do. Nor do theyhave to be
eitherverified
r verifiable
n the early
positivistic
ense.
Partial
confirmation
ill
have
to do.
To demand
more f evolution-
ary
taxonomy
wouldbe to
demand more
than
any
science can
deliver-including
physics.
THE QUEST
FOR
ABSOLUTE
CERTAINTY
Philosophersnd scientists hroughhe
centuries
ave
demanded
more,but
these
demandshave
stemmed
rom failure
o
understand
he nature
f
empirical
cience.
Such "apriorists"
s
Aristotle warded
he
honorific
itle "science"
only
to deduction
from purely
universalgeneralizations
o
purely
universal
generalizations,
ecause
such nferencesecessarily
esulted
n
true
conclusions f
the
premises
were
true.
Aristotle ssured he truth fhis premises
by
claiming
hat
we
intuit
hem
nd
intui-
tion s alwaystrue
Largely
n
reaction
o
the excesses
of Aristotelianism,ome
phi-
losophers
nd
many scientists
doptedan
extremely
mpirical iew
of science
n the
18th and 19th
centuries.
These
extreme
empiricists
r "Baconians"
as
theycalled
themselvesooked uponscience
s an
"in-
ductiveprocess,"
but
they
too demanded
absolutecertaintynd attempted o get
it
by nevergoing
beyond
the evidence.
Scientificaws
were ust empirical
enerali-
zations, ust
summations
f the data.
Any-
thingelse was
mere speculation.
Even
though
Darwin
himself
laimed to have
proceeded
n
the true Baconian
manner,
the main
criticism eveled against
his
theory y
the
biologists
f
his
day
was
that
it was speculation.
He had
gone
beyond
his evidence n a manner nbecoming n
inductive
scientist and an
Englishman.
(See
Ellegard,1957.)
To a philosopher
t sounds
strange o
hear
a neo-Platonist
ike
Richard Owen
criticizing
arwin
for
dle
speculation,
all-
ing
himself
n inductive
cientist
nd
com-
This content downloaded from 142.51.1.212 on Mon, 21 Jul 2014 17:05:49 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp7/26/2019 Certainty and Circularity in Evolutionary Taxonomy- by David L. Hull. Evolution, Vol. 21, No. 1 (Mar., 1967)
10/17
182
DAVID
L.
HULL
plimenting uvier
fornot being
"the man
to draw conclusions eyond
his premises"
(Owen, 1860).
Inductive
nferences
re
just those nferences hichdo go beyond
the evidence t hand and which o
run the
risk of leading
to false conclusions
rom
truepremises.
As Peter Caws (1965)
put
it recently:
"One might
define
deduction
as a process of
inference n which one never goes beyond
the
given facts, nd
in which therefore here
s never
any loss of certainty. Induction, on
the other
hand, does go beyond the given facts,
nd there-
fore runs the risks that deduction does
not. We
shall find these risks at the root of scientific
theory."
Apriorists nd
extreme mpiricists ave
tried
o
obtain
bsolute ertaintylthough
by opposite
means. Neither
of these
ex-
treme positions
is appropriate for the
actual
practiceof any science,
ncluding
taxonomy.
During the
early years of this
century,
extreme
mpiricism
n
biology ecame
qui-
escentbut it burst forth n psychology
under the
name
of behaviorism. he be-
havioristsdemanded"operational
defini-
tions"
of
all the
termsnpsychology,efi-
nitionsof a type
which they
mistakenly
thoughtwere common
n
physics.
Under
careful nd repeated riticism y
philoso-
phersand theirfellow sychologists
e.g.,
Bergmann, 954
and 1956), and
afterbe-
ing abandoned
by the founder f opera-
tionism,. W. Bridgman1954), behavior-
istsbegantomodify heir xtreme
osition.
It has
now become
little more than an
emphasison a particular orm
f experi-
ment
nd a
cry for omeempiricalontent
in
psychological
aws.
The
history
f
oper-
ationism hould be an
object
lesson to
those
taxonomists
ho
are
attempting
o
introduce t
into
taxonomy. f
it is
in-
tended as a demandthat the basic con-
cepts
n
taxonomy
e
operationally
efined
in the
strict ense,
hen t is untenable.
f
on
the other
hand,
all
that
s
intended
s
that
the
definitionsnd
laws
in
taxonomy
be
operational
i.e.,
have some
empirical
consequences
hich
an be
checked),
hen
it is an
admirable
rogram
ut
one
that
s
neither
ew
norstartling.
THE
LANGUAGE
OF
LOGICAL
CRITICISM
In
his
early
critical
writings
. J.
Cain
exhibits
he
same
contradiction
hich
has
always
been
characteristic
f the
writings
of
extreme
mpiricists.
At
times
he
de-
mands
bsolute
ertainty
f
the
nferences
made
by
the evolutionists;
t
other
imes
he seems
to
be
arguing
more
reasonably
that
toomany
of these
nferences
re
not
sufficiently
arranted.
A
confusion
n the
language of logical criticism mphasizes
thisvacillation.
For example,
fter
rguing
against
the
use of
deduction
n
taxonomy
(Cain,
1958),
the only
kind
of
inference
which permits
categorical
demonstration
and
apodeictic
certainty,
ain (1959a)
complains
hat phyletic
nferences
re
not
"apodeictically
ertain"
or "categorically
demonstrated."
urthermore,
fter uoting
two
Aristotelian
hilosophers
n
the
im-
possibilityfapplyingAristotelianogicto
biological
lassification,
ain (1959)
says:
"The
relevance
of these quotations
from
ogi-
cians
for the
whole history
f
biological
taxonomy
from
Aristotle
o
the present
day can
hardly
be
over-estimated.
They epitomize
the
most
im-
portant
change
in taxonomic
theory
hat has
oc-
curred,
namely
the gradual
abandoning
of at-
tempts
to set up
classifications
n a
priori
prin-
ciples
agreeable
to the
rules
of
logic
and
some
particular
theory,
nd
the
partial
substitution
f
an empiricalattitude. This substitutionwas not
complete
when
the
theory
of evolution
arrived
to
provide
a
new
theoretical
approach
to the
problem
of classifying
rganisms,
he
full
impli-
cations
of
which
have
still not
been completely
thought
out.
[Taxonomists
should] proceed
em-
pirically,
imply
finding
out
what
subjects
exist
and what
are
theirattributes,
ot
deducing
them
from
known principles
nd
axioms."
Cain terms
he
principles
sedby
theo-
retically
riented
axonomists,
rom
Aris-
totle o theevolutionists,priori nd their
inferences
eductions,
hen
n
point
ffact
the
thrust f
his
objections
o evolutionary
taxonomy
s that the principles
re a
pos-
teriori
and the
nferences
nductions.
Cain
objects
to
evolutionary
rinciples
ecause
they
re
supposedly
priori
nd
then om-
This content downloaded from 142.51.1.212 on Mon, 21 Jul 2014 17:05:49 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp7/26/2019 Certainty and Circularity in Evolutionary Taxonomy- by David L. Hull. Evolution, Vol. 21, No. 1 (Mar., 1967)
11/17
CERTAINTY AND
CIRCULARITY IN
TAXONOMY
183
plains that theydo not have certain
rop-
ertieswhichonly
a
priori
tatements
an
have. He objects to evolutionary
nfer-
ences because they re supposedly educ-
tions nd then omplains hat theydo
not
exhibit type of certainty hat only
de-
ductions an
have. Cain findsAristotelian
logic inadequate for the
purposes of
science,but he turns roundand uses its
basic distinctionso criticize
volutionary
taxonomy.
f
Aristotelianogic won't
do
for he
working axonomist
n
his
everyday
pursuits,
hen t
won't
do for
the taxono-
mistwhenhe steps back to evaluatethe
structure f
his
science.
Terminological
confusions
ike
those
ust
mentioned
er-
vade the taxonomic
iterature,
nd a rea-
sonable decision s to
the ustification
f
evolutionaryeconstructionsependsupon
just those
distinctions
lurred by these
confusions.
The
principles
sed
by evolutionistso
reconstructhylogeny re
not
a
priori
n
the philosophical ense of this avowedly
philosophic xpression.According o ac-
cepted philosophicusage,
the
phrase
a
priori ppliesonlyto our knowledge f
the
truth f the
statement
lthough
t
is often
used
elliptically o refer o
the
statement
itself.
A
statement
s
a
priori rue
f
its
truth an be decided prior to experience.
For
example,
he
tatementhat ither
t is
raining omewhere n the plain
in
Spain
rightnow or it is not is a priori rue. Of
course,
ts
truth annotbe
known rior
o
all
experience,
ince
we
must
know what
the
words
n
the statementmean and
we
must
understand nglish syntax,but
its
truth
an be decidedwithout
eference
o
the
weather. f the truth f a statement
can be decided
priori,
erifications
ir-
relevant.
Empirical
considerations
ust
don'tmatter.
As
a
consequence,
uch
tate-
mentsare necessarily rue. We can be
apodeictically
ertain f
their ruth. The
scientific
rinciples as
distinct
rom
he
logical
and
mathematical
rinciples)
used
by
the evolutionists o reconstruct
hy-
logeny,
ike
all scientific
rinciples,
re not
a
priori
n
this ense.
Hence, they re
not
necessarily rue n
and of
themselves.At
best
they an be
onlyhighly onfirmedr
madehighly
robable.
In addition o thisphilosophic se of the
expression,here
s also a use in
everyday
discourse
ccording o
which a belief is
termed
a
priori
if
the person
who
holds
it refuses
o entertain
ounterevidence.
or
example, he
belief of
many people that
smoking
will
not
increase heir hance
of
contracting ung
cancer is
a
priori
in
this
sense.
The
belief
is
not
a
priori
in
the
philosophic ense,
since evidence s
rele-
vant. The problems thatthepersonre-
fuses
o
acknowledge
he
relevant vidence.
He
is certain hat
his belief s true, uthis
certainty
s
not
justified.
Perhaps
some
evolutionists
ave treated heir
principles
as
a
priori
in
this second
sense. If
they
have, t is
unfortunate. ut
sometimeshe
appearance
of
ignoring
counterevidence
stems
from
misconstrual f the
logical
nature f both
evolutionary
rinciplesnd
the nferences ade fromhem.The critics
of
evolutionary
econstructionsomplain
that even
though every
single
principle
used
by
the
evolutionistss
known ohave
exceptions,
he
volutionists
ontinue o use
them. It might
be noted
in
conjunction
withthis
claim that
every
ingleprinciple
in
Newtonianmechanicss
known o have
exceptions,
ut
physicists
ontinue
o
use
them. (See
Scriven, 1961
and Kaplan,
1964.)
The
misunderstanding
s
expressedmost
clearly
by
R. S.
Bigelow
1959)
when
he
argues
hat
a
hypothesis
o the effect
hat
all
the
balls
in
a
particular
ag
are
white
is
not probably
rueonce t is
known hat
one
ball
in
the
bag
is
red;
the
hypothesis
is
false.
Similarly,
he
hypothesis
hat
phenetic
imilarityorresponds
o
recency
of
common
ncestry
s
not
probably
rue
once it is known hatthere re numerous
exceptions
o
this
rule;
t s
false.
This
ine
of
reasoning
ollows, owever, nly
f
the
statement
elating
henetic
imilarity
nd
recency
f common
ncestrys
taken o be
categorical
n
form.
f
it is
intended
o be
a
tendency
tatement
as
it
certainly
s),
This content downloaded from 142.51.1.212 on Mon, 21 Jul 2014 17:05:49 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp7/26/2019 Certainty and Circularity in Evolutionary Taxonomy- by David L. Hull. Evolution, Vol. 21, No. 1 (Mar., 1967)
12/17
184
DAVID
L.
HULL
then
single
xception
oes
not
falsify
t.
Even
numerous
xceptions
re permissible.
Consequently,
f Bigelow
still wishes
to
registern objection o evolutionary
ea-
soning,
e is
left
n
the
unhappy
position
of
arguing
(as some
philosophers
ave
argued)
that
the
only
principles
roperly
admitted
nto the
pandects
of science
re
true
universal
eneralizations
nd
conclud-
ing
(as
some
philosophers
ave
concluded)
that
since
there
re
almost
no such
gen-
eralizations
nbiology,
iology
s almost
n-
tirely
evoid
of scientific
aws.
The logicbehind uch a lineof reason-
ing s
interesting.
f
scientific
aws
are re-
stricted
o
true universal
generalizations,
then
deduction
o
the particular
ase
is
possible
and
these particular
onclusions
follow
necessarily.
f the conclusion
urns
out
to be false
and
the argument
s
valid,
then t
follows
ecessarily
hat
t
least one
of the premises
s
false.
Hence,
universal
generalizations
re
in
principle
asily
falsi-
fiable.Such s notthecasewith nferences
from
generalizations
hich
are
less
than
universal
n
form.
nferences
o
particular
statements
re
only
more
or less
probable
and
falsification
s not so easy.
They are
falsifiable,ut
a
single
observation
on't
do
it. For
example,
rom
heuniversal
en-
eralization
hat
all parents
with
blue eyes
produce
nly
blue-eyed
hildren,
t can
be
deduced
with
podeictic
ertainty
hat f
a
certain hild s biologicallyheoffspringf
these
parents,
t
will
have
blue eyes.6
But
from
he
statements
hat
blue-eyed
arents
almost lwaysproduce
blue-eyed
hildren,
it cannot
e deduced
hat
fa certain
hild
is biologically
heirs,
t
will
have blue
eyes.
The
inference
s only
inductive
nd
the
conclusion
ollows
rom he premises
with
only
a degree
of certainty
r probability.
As G. G. Simpson 1961) has said:
"Scientists
themselves
frequently
eemed
con-
fused
as to the
degree
and indeed
also the
kinds
of 'certainty'
actually
always
probability)
that
are required
or
are possible
n
science."
The moral
f the
preceding
iscussion
s
that
f
such
empiricists
s Cain
are taken
at their
word,
thenthey
do not hold
an
empirical
view
of empirical
certainty.
Scientists ave a right o say thatthey re
certain f
the
truth f
a particular
tate-
ment ven
though
hey
mayfind
ut
later
that
the statement
as
false.
Any
use of
"certainty"
which
makes t
impossible
or
a scientist o be justifiably
ertain
f the
truth
of a statement
which
s later
dis-
covered
o
be false
s a view
of
certainty
inappropriate
o empirical
cience.
It
would eem
unlikely
hat he
founders
of numericalaxonomy, movement hich
relies
so heavily
on
the use of statistics,
should
hold
a view
of
inference
otally
t
variancewith
he
foundations
f statistics.
Yet,
at least
in
his
early
writings,
ain
seems
to do just
this.
In his later
works,
Cain clearly
eveals n
empirical
otion f
certainty,
or example,
n
Cain
and Har-
rison 1960).
But
in
thisvery
paper,
the
authors tate with
respect
o
the
recogni-
tionofconvergence,But wehaveno way
of estimating
hese probabilities;
ven
if
we
did,
we should obtain
only
probabili-
ties.
.
."
It
may
well
be
true that
pres-
ently
there
s no
way
to estimate
hese
probabilities,
ut
to
say
that
even
if
we
did,
we
shouldobtain
onlyprobabilities
s
to
label
oneself
an
extreme
empiricist.
Probabilities
re
all
that
scientists
ver
have
to go on, probabilities
hich'
ome-
times re so highthatthey an be termed
certainties.
Perhaps
statements
uch as
those cited
are
only
due
to careless
expression;
per-
haps
they
reveal a deep-seated
ntipathy
to
the
kind
nd
degree
f
certaintyossible
in
empirical
cience,
n antipathy
which
6
It should
be noted
that as
is
usually
the case,
this
universal
generalization
has exceptions
and,
hence,
s not
strictly
rue. One
way
to eliminate
the
obvious
exceptions
s
to add the phrase
"ex-
cept
in cases
of
mutations,"
but if
"mutation"
s
thendefined s it once was as any variationfrom
the regularities
f inheritance,
hen
the law de-
generates
o a
tautology
nd
is no
longer
an em-
pirical
aw.
In
order o retain
ts status
as an em-
pirical
aw, the
generalization
n question
must be
such
that
it is at least
logically possible
for there
to be
exceptions.
If
in point
of
fact, there
are
none,
then
t
is
true.
This content downloaded from 142.51.1.212 on Mon, 21 Jul 2014 17:05:49 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp7/26/2019 Certainty and Circularity in Evolutionary Taxonomy- by David L. Hull. Evolution, Vol. 21, No. 1 (Mar., 1967)
13/17
CERTAINTY
AND
CIRCULARITY IN TAXONOMY
185
has been traditionally
haracteristicf the
apriorists.Considering
he
mportant
on-
tribution hich ain has
madeto taxonomy
by pointing ut the dilatory ffect f the
a
priori utlook
n
taxonomy,ne hopes t
is the former. owever,whenthepoint
of
contentions
precisely
hose ssues which
the verbal confusion
lur, the results re
fatal.
EMPIRICAL CERTAINTY
AND EMPIRICAL
LAWS
With
such
distinctions
n
mind,
we are
now n a position oevaluate heevolution-
ary
laws and
principleswhich are at the
heart of the chargeof vicious circularity.
When
thecritics erm he aws and
princi-
ples of evolutionary
heory priori,
what
they usually
have
in
mind s
that evolu-
tionists ormulate
hemfrom he
study
of
certain
groups
of
organisms
n
less than
total evidence
for those
groups
nd then
extend hem o new
groups
or
which
on-
firmatorybservationsave notbeen made
and
may
notevenbe
possible.Hence,
with
respect o these attergroups, he applica-
tionof the
principless
prior
o
experience.
But
in
this sense all scientific rinciples
are
a
priori. They go
beyond
he
evidence
at hand.
They
wouldbe useless s
princi-
ples
if
theydidn't. This
is the sense of a
priori used when
weightings iven to a
character
efore he
taxon has
been
con-
structed re termed priori nd those s-
signed fter hetaxon
has been constructed
a
posteriori.
t is
also the
senseof
a
priori
used
by William
Coleman (1964)
in
his
book on
Cuvierwhenhesays, The natural
system
an be
prepared
n
two
ways:
1.
a
posteriori, y
direct
observation;
2. a
priori, y
the
principle
f the subordina-
tion of characters."
How inappropriate
his
particular sage
is in criticizingcience an be seenin the
followingxample for
the
sake
of
brevity
only
one
example
will be
given). Kepler
enunciated
his
laws
of
planetary
motion
whichwere
to
apply
to all
the
planets
n
the
solar
system
n
the
basis
of
ess than
handfulof
observationsmade
by Tycho
Brahe on a single planet,
Mars.
Later
astronomers
heckedhis laws for
other
planets n
the solar system nd for
other
positions f Mars, but they lso extended
them o cover
ll planetsrevolving
round
all stars.
We have little
hope of eververi-
fying r evenconfirming
hishypothesis
y
observation
ormore han an infinitesimal
portion
f the
star systems
n the
universe,
and
yet
no one wouldwant
to call
the
n-
ference unwarranted
r the hypothesis
meaningless.
Evolutionaryaws have
been devised
n
thebasis of innumerablebservationsut
observations
which have
been made pri-
marily
n
restrictedreasof the plant
and
animalkingdoms. ertain
ortions
f
phy-
logeny re
as inaccessible s
the farthest
corners f the universe.
These laws have
been formulated n
less than total
evi-
dence, but what is more
they have
been
found
o
be
not
completely
ccurate
ven
for hosegroups romwhich heywere
de-
rived. Even so, evolutionists xtrapolate
to other
groups
for
which
ittle
or
no evi-
dence s available.
Phenetic imilarity
as
been
found o
be a fairly ood
ndicator f
evolutionary
elationships
n
certain roups.
Cases
are
known, owever,
n
which
his
s
not true.
Even so, evolutionists
se phe-
netic similarity o infer hylogeny
or all
taxa.
We have ittlehope
of eververifying
or
even confirminghe
correspondencey
observation ormore han an infinitesimal
portion f
taxa thathave
evolved, nd yet
the
inference
s
not unwarranted,
or
the
assertion
meaningless.
volutionists
now
that
phyletic
imilarity
s not directly ro-
portional
o
phenetic
imilarity
n
all
cases,
but
it is
good enough
for their
purposes.
Astronomers
new hatKepler's aws
were
not completely ccurate,
but they
were
close
enough
or heir
urposes.
What us-
tifies evolutionarynferences s exactly
what
justifies
astronomical nferences-
theory.
The
inductiveeap which strono-
mers
made
concerning
lanetarymotion
s
much
reaterhan hose
volutionists
ould
ever
hink
f making-and for very
good
reason. Evolutionary
heory, ven com-
This content downloaded from 142.51.1.212 on Mon, 21 Jul 2014 17:05:49 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp7/26/2019 Certainty and Circularity in Evolutionary Taxonomy- by David L. Hull. Evolution, Vol. 21, No. 1 (Mar., 1967)
14/17
186
DAVID
L. HULL
binedwith
modern
enetic
heory,
oes
not
haveanything
ike the
power
f
Newtonian
theory.But
the ustification
n
both
cases
remains
he same.
The crux of
the
dispute
between
the
evolutionistsnd the
empiricists
s the
kind and
degree
f certainty
ecessary
o
justify
cientific
nferences.
Are
the
in-
ductive
nferences ade
by
evolutionists
n
reconstructing
hylog