Brit. J. Phil. Sci. 42

(1991), 273-284Printed

in Great B

ritain

RE

VIE

W A

RT

ICL

E

Refin

ed F

alsificationism

Meets th

e

Challenge from

the Relativist

Philosophy of S

cience

GU

NN

AR

AN

DE

RS

SO

N [1988]: K

ritik andW

issenschaftsgeschichte.K

uhns,

Lakatos' und,

FeyerabendsK

ritik -des

Kritischen

Ratio

nalism

us. x

+218 p

p.

Hard

back

DM

98. IS

BN

3-1

6-9

45308-4

.

1T

ourd'H

orizon or the Ecology

of the Philosophy of S

cience

2 Do th

eC

ase Studiesw

hichthe 'N

ew P

hilosophy of Science'

hasculled from

the

History of

Science show that the

Falsificationist

Methodology

isU

nrealistic?

3O

bjectives,Structure

and Results

of theV

arious Chapters

of the Volum

e

4 Stocktaking

,

I TO

UR

D'H

OR

IZO

N O

R T

HE

EC

OL

OG

Y O

F T

HE

PH

ILO

SO

PH

Y O

F

SC

IEN

CE

In o

ur cen

tury

, the p

hilo

sophy o

f science h

as been

oversh

adow

ed b

y tw

o

tow

ering fig

ures: P

opper an

d W

ittgen

stein, b

oth

Vien

nese em

igran

ts, who

hav

e beco

me su

bjects to

the Q

ueen

(cf., e.g., R

adnitzk

y [1

987a] E

ntre

Wittgenstein et

Popper ... ).

The d

iscussio

n h

as been

structu

red b

y tw

o g

reat

controversies: from the 30s P

opper versus logical positivism. (or falsificatio-

nism

versus

verificationism/probabilism

), and from the 60s 'the new

philoso-

phy of science' versus Critical R

ationalism. (E

xemplary contributions to these

two co

ntro

versies can

be fo

und, e.g

., in th

e two co

llections R

adnitzk

y an

d

Andersson

(eds.)

[1978],

Pro

gre

ss ... Scie

nce,

and [1

979], S

tructu

re .

Scien

ce.)W

ittgen

stein's T

ractatus h

as been

the id

ee directrice o

f the V

ienna

Circle an

d its su

ccessor, L

ogical E

mpiricism

. The cy

nosu

re of 'th

e new

philo

sophy o

f science' is

Wittg

enstein

's later philo

sophy as p

resented

, in

particular, in his Philosophical Investigations. If you apply it to the philosophy of

science, y

ou w

ill view

'norm

ative' m

ethodolo

gy as a co

unterp

art of id

eal

languag

e philo

sophy, an

d h

ence reg

ard an

y m

ethodolo

gical p

rescriptio

ns as

unrealistic and claim that philosophy of science cannot do m

ore than describe

the p

ractice of scien

ce, preferab

ly in

terms o

f case studies cu

lled fro

m th

e

274

Gerard

Radnitzky

histo

ry o

f science. Y

ou w

ill be p

laced o

n th

e road

to relativ

ism (K

uhn,

Fey

eraben

d, H

ubner, an

d o

thers). A

n in

teresting v

ariant o

f relativism

is

Stegm

tiller's attempt, w

ith the help of the formal m

ethods developed by Sneed,

to g

ive K

uhn's resu

lts a more ex

act form

(Sneed

, Steg

muller, M

oulin

es,

Mostarin)-w

hat Feyerabend has called 'the S

needification of science'. Today,

dep

artmen

ts of p

hilo

sophy o

f science are d

om

inated

by p

hilo

sophers w

ho

either sy

mpath

ize with

logical em

piricism

or w

ith `th

e new

philo

sophy o

f

science'.C

riticalR

atio

nalism

has a

n o

utsid

er p

ositio

n, b

ut e

njo

ys th

e

sympathy of m

any scientists (Bartley [1989], U

nfathomed K

nowledge ...; see

also B

artley [1

987b], 'P

hilo

sophy o

f bio

logy.. . ', an

d M

unz [1

987], ' ... th

e

mirror of.R

orty').

In ad

ditio

n to

Wittg

enstein

's work

two b

ooks h

ave b

een o

f param

ount

importan

ce for th

e discu

ssion: P

opper's

magnum

opus

of 1934, Die L

ogik der

Forsc

hung,

and Kuhn's volum

e of 1962, The S

tructureof Scientific R

evolutions.

Popper's b

ook w

ent. alm

ost u

nnoticed

until th

e English

editio

n ap

peared

in

19 59. It made P

opper so-to-speak famous overnight. K

uhn is a Ludw

ik Fleck

rediv

ivus-if y

ou co

mbin

e Fleck

with

the later W

ittgen

stein y

ou g

et Kuhn's

position. Fleck's book of 19 3 5 rem

ained unnoticed and was rediscovered only

recently

(cf., e.g., A

ndersso

n [1

984a], '.. F

leck's. co

ncep

tion ... '; see also

Jarvie's ex

plan

ation o

f Kuhn's su

ccess in th

e philo

sophy estab

lishm

ent in

Jarvie [1

988], ' ... K

uhn as id

eolo

gue ... ', an

d B

artley [1

989],

Unfath

om

ed

Know

ledge ... ).

A com

mon characteristic of the critics of C

ritical Rationalism

is that,

following W

ittgenstein's later philosophy, they adopt what P

opper has called

'the m

yth

of th

e framew

ork

'and interpret the history of science as sequences of

'incomm

ensurable' styles, traditions, or 'paradigms', in w

hich no criteria are

available that overarch traditions or 'paradigms'; and this in turn suggests

that rational appraisal of such traditions or paradigms is im

possible. Popper's

Critical R

ationalism, on the other hand, m

aintains that traditionalism and the

demand for rational scrutiny can be com

bined (cf., e.g., Andersson (ed.) [ 19 84],

Rationality ... ). T

he relativists and the so-called 'sociology of knowledge'

claim that the critical tradition is im

possible. W. W

. Bartley, III has argued that

their criticism applies to w

hat has been called 'comprehensive rationalism

'. but

not to comprehensively critical rationalism

or 'pancritical rationalism' (see,

e.g.,B

artley [1987a], 'A refutation ...'; for a criticism

of the 'sociology of

knowledge' see P

art III of Radnitzky and B

artley (eds.) [1987], and Jarvie

[1988], ' ... sociological turn ... '). Relativism

and the 'sociology of know-

ledge' has far-reaching implications outside the field of the philosophy of

science. How

ever, the strength of the arguments upon w

hich their adherents

base their claims resides on their w

ork in the philosophy of science. If it could

be shown that their philosophy of science is untenable, this should have

repercussions in many other fields.

Refined

Falsificationisni

Meets R

elativistP

hilosophy of ScienceC

hallenge275

2 D

O T

HE

CA

SE

ST

UD

IES

WH

ICH

TH

E 'N

EW

PH

ILO

SO

PH

Y O

F S

CIE

NC

E'

HA

S C

UL

LE

D F

RO

M T

HE

HIS

TO

RY

OF

SC

IEN

CE

SH

OW

TH

AT

TH

E

FA

LS

IFIC

AT

ION

IST

ME

TH

OD

OL

OG

Y IS

UN

RE

AL

IST

IC?

Gunnar A

ndersso

n (U

niv

ersity o

f Um

ea, Sw

eden

) subm

its an an

swer to

this

question in his latest book. Before the case studies used by K

uhn, Lakatos and

Feyerabend can be analyzed the logical structure of falsification argum

ents has

to b

e clarified an

d so

me co

nco

mitan

t meth

odolo

gical p

roblem

s hav

e to b

e

solved. The author exam

ines the most im

portant of the 'classical'. case studies

which

the critics o

f Popperian

meth

odolo

gy h

ave su

bm

itted, an

d h

e show

s

that n

one o

f them

with

stands a critical scru

tiny. In

the p

rocess o

f appraisin

g

the arguments of the critics it turns out that the P

opperian methodology needs

to b

e impro

ved

in certain

central areas. B

y in

troducin

g 'p

ancritical ratio

na-

lism',

W. W

. Bartley

, III has im

pro

ved

the ep

istemolo

gical fram

ework

of

Critical

Ratio

nalism

(Bartley

[1962/8

6], R

etreat to..., B

artley [1

987b]

'Philo

sophy o

f bio

logy...). T

his a

chie

vem

ent is o

f inte

rest m

ain

ly to

epistemology and to philosophy in general. A

ndersson's book provides the first

major im

provement of falsificationist

methodology

since th

e 30s. F

or th

e first

time, falsification argum

ents are provided with

a nietalogicalbasis. T

he author

also shows: (a) how

unproblematic

test statements can

be deduced f rom problem

atic

ones;(b) thatem

pirical testing concernstheoretical

systems

as wholes; and (c) how

Popper's

viewofauxiliary hypotheses

andad-hoc hypotheses

can be improved.

As a

result o

f his in

vestig

ations K

uhn's criticism

of falsificatio

nist m

ethodolo

gy

collapses and the so-called Incomm

ensurability Thesis turns out to be false.

3 O

BJE

CT

IVE

S, S

TR

UC

TU

RE

AN

D R

ES

UL

TS

OF

TH

E V

AR

IOU

S

CH

AP

TE

RS

OF

TH

E V

OL

UM

E

The first ch

apter o

utlin

es and su

mm

arizes the criticism

that K

uhn, L

akato

s,

and Feyerabend have brought forw

ard against falsificationist methodology, a

criticism based on case studies culled from

the history of science.

The

second

chapter clarifies the logical structure of falsification arguments. In

ord

er'to ap

praise th

e conclu

sions th

at the critics h

ave d

rawn fro

m th

eir case

histories, the logical structure of explanation, prediction and falsification has

to b

e elucid

ated an

d th

e logical relatio

nsh

ips b

etween

them

exam

ined

. In h

is

The L

ogicof

ScientificD

iscovery

,P

opper o

nly

deals w

ith tw

o sp

ecial cases of

falsification arguments: (1) the falsification of a theoretical system

(including

the statem

ents o

f the in

itial conditio

ns)- w

ith th

e help

of th

e neg

ation o

f a

prognosis; (2) the falsification of an isolated hypothesis with the help of a 'basic

statement'. P

opper has not given a satisfactory answer to the question of w

hy

certain falsificatio

n arg

um

ents are v

alid.. B

efore th

at questio

n can

be

answ

ered, th

e. logical relatio

nsh

ips h

ave to

be clarified

betw

een ex

plan

ation

ik

276

Gerard

Radnitzky

and falsification on the one hand and explanation and prediction on the otherhand. T

he author shows that, if predictive argum

ents are valid, certain types offalsificato

ry arg

um

ents are v

alid to

o, an

d v

ice versa. T

he fo

rmal p

roofs fo

rthese m

etalogical equivalences are gathered in an Appendix. It turns out that

the two special types dealt w

ith by Popper are not the only valid ones, that the

idea of a falsificatory argument can be generalized-that m

any other forms are

valid

too. In

ord

er to ap

praise th

e criticism ag

ainst falsificatio

nism

bro

ught

forw

ard b

y K

uhn, L

akato

s, Fey

eraben

d an

d o

thers, it is im

perativ

e to sh

ow

how

theo

retical system

s that co

nsist o

f several g

eneral h

ypoth

eses can b

efalsified.

After all, w

hat is at stake in the history of science and the practice ofscience are very rarely isolated hypotheses, norm

ally theoretical systems are at

the c

en

ter o

f inte

rest. It tu

rns o

ut th

at th

e th

esis th

at e

xp

lan

ato

ry a

nd

pred

ictive arg

um

ents h

ave th

e same lo

gical fo

rm d

oes n

ot h

old

in g

eneral. It

only holds for explanation and for the deduction of unconditional predictions,

bu

t it do

es n

ot h

old

for e

xp

lan

atio

n a

nd

the d

ed

uctio

n o

f co

nd

ition

al

pred

ictions. E

xplan

ations an

d arg

um

ents h

avin

g a co

nditio

nal p

redictio

n as

conclu

sion are m

etalogically

equiv

alent. T

he au

thor ex

tends th

e analy

sis of

the logical structure of explanation (Popper, H

empel, O

ppenheim, S

tegmuller,

etc.) to th

e analy

sis of th

e logical stru

cture o

f pred

ictions (co

nditio

nal an

dunco

nditio

nal), an

d th

e analy

sis of th

e structu

re of falsificatio

n arg

um

ents.

The

thirdchapter is devoted to T

homas K

uhn. It is shown that his criticism

of

falsificationism hinges on the-position he takes vis-n-vis to tw

o methodological

problems: (1)

What are

theim

plications of the theory dependence of experience?.

and (2

)W

hatis

the 'rational' reaction of researchers to a falsification?F

rom

differen

t starting

po

ints, P

op

per, K

uh

n, N

. R. H

anso

n, an

d o

thers, h

ave

reached

the co

nclu

sion th

at the so

-called b

asic statemen

ts or test statem

ents

are theo

ry-d

epen

den

t, fallible, an

d h

ence rev

isable. S

o far so

good. K

uhn

overstates the case when he claim

s that scientists who have adopted radically

differen

t back

gro

und assu

mptio

ns o

r parad

igm

s even

tually

'live in

differen

tw

orld

s', and th

us arriv

e at differen

t test statemen

ts, with

the resu

lt that th

e,th

eories th

ey d

evelo

p are in

com

men

surab

le.W

ith resp

ect to h

is positio

n v

is-h-v

is the seco

nd m

ethodolo

gical p

roblem

(the research

er's response to

a falsification) K

uhn k

eeps in

the co

ntex

t of

justificationist philosophy. He believes that falsification is som

ething definiteand irreversible. T

his mistaken view

of falsification underlies his claim that in

'normal science' a falsification does not lead to the rejection of the theoretical

system falsified but rather to attem

pts to modify that system

. Kuhn then argues

that this strategy imm

unizes theoretical systems against falsifications. H

avingplay

ed d

ow

n th

e role o

f falsification in

scientific research

Kuhn th

inks th

atP

opperian

(norm

ative) m

ethodolo

gy is u

nrealistic. C

hap

ter 3 is d

evoted

tooutlin

ing K

uhn's p

ositio

n. A

ndersso

n sh

ow

s that K

uhnian

key

concep

ts like

'puzzle', 'an

om

aly', 'co

unter ex

ample', 'p

aradig

m', etc., are am

big

uous an

d

Refined

Palsificationisni.Meets R

elativistP

hilosophy oJ'Scierice

Challenge

277

unclear an

d also

that K

uhn's m

ethodolo

gical arg

um

ents are m

arred b

y

irrelevant psychological analyses: The decisive criticism

of Kuhn's position is

postponed to Chapters 6 and 7.

The

fourth

chap

ter is dev

oted

toM

ireL

akatos's attempt to rem

ould Popper's

positio

n in

such

a way

that it acco

mm

odates th

ose p

arts of K

uhn's criticism

that L

akato

s consid

ered co

rrect. It is an attem

pt to

save th

e rationality

of

science from relativism

-in spite of Kuhn. A

s is well know

n, Lakatos exam

ines

vario

us p

roposals h

ow

to d

eal with

the p

roblem

of th

e theo

ry d

epen

den

ce of

experien

ce. After h

avin

g rejected

the p

ositiv

ist pro

posal th

at certain b

asic

statements can be proven to be true, L

akatos first interprets Popper as a naive

falsificationist, who thinks that basic statem

ents are made unfalsifiable by flat,

by a conventional decision. According to T

opper-as a sophisticated falsifica-

tionist-b

asic statemen

ts can b

e criticized b

y d

educin

g fu

rther co

nseq

uen

ces

from

them

. Lak

atos criticizes th

is correct in

sight b

y assertin

g th

at basic

statements are not checked in the w

ay Popper claim

s that they are. He claim

s

that they are checked with the help of 'interpretative theories'. (A

llegedly, such

an `in

terpretativ

e theo

ry' fu

nctio

ns to

allot tru

th v

alues to

test statemen

ts.)

Andersson show

s that Lakatos's attem

pt to solve the problem of the em

pirical

base by means of introducing the concept of 'interpretative theories' is circular:

basic statem

ents are u

ltimately

appraised

with

the h

elp o

f 'interp

retative

theo

ries', and th

ese theo

ries are then

in tu

rn ap

praised

by m

eans o

f basic

statemen

ts. Play

ing d

ow

n th

e role o

f falsification in

research, L

akato

s

even

tually

arrives at a p

aradoxical m

ixtu

re of co

nven

tionalism

and in

ducti-

vism.

Lakato

s a

ccepts

Kuhn's

thesis

that, in

the c

onte

xt o

f 'norm

al s

cie

nce',

theories are relatively imm

une against falsification, and he attempts to analyze

the mechanism

of such imm

unization strategies. Whenever a theoretical

system faces a falsification, various reactions or (internal) research policies are

possible. You can try to repair the situation either by m

odifying peripheral

parts or by modifying central parts of the theoretical system

concerned.

Lakatos com

es to believe that by making m

odifications in the peripheral parts

(in the 'protective belt'), scientists try to salvage the central parts of the theory;

they

do so

by th

e meth

odolo

gical d

ecision to

mak

e the cen

tral parts

'unfalsifiable'-hence, the robustness of the so-called 'hard core', of a research

program. T

hus, Lakatos concludes that the im

munization strategies that K

uhn

had claimed for. 'norm

al science' are rational after all. By contrast to K

uhn-

who in L

akatos's view has in the last resort fallen back on consensus in the

sense of 'mob psychology'-L

akatos tries to develop objective criteria for the

appraisal of various research strategies. Andersson show

s that these criteria

are problematic, because they involve an

ex an

teappraisal of the 'heuristic

potential' of a research program and thereby involve guesses about the future

performance of the program

in question. Lakatos has not been a "

plain

i

278

Gerard R

adnitzky

how

such

an ex

an

teap

praisal co

uld

work

. Andersso

n co

njectu

res that th

isdifficu

lty m

ay h

ave in

duced

Lak

atos to

aban

don m

ethodolo

gy an

d retract to

the 'ra

tion

al re

con

structio

n' o

f the h

istory

of sc

ience. L

eft w

ithout a

meth

odolo

gy th

at could

supply

criteria of ap

praisal, L

akato

s falls back

on

appraisals contained in the'normative basic judgm

ents' of'the scientific elite'.T

his ag

ain lead

s to a circle: in

ord

er to d

etermin

e who sh

ould

belo

ng to

'the

scientific elite', we need objective criteria for appraising past achievem

ents.

The fifth chapter is devoted to

Pau

l Fey

era

ben

d's p

ositio

n.

Fallibilism

combined

with

a particu

lar interp

retation

of th

e theo

ry d

epen

den

ce of ex

perien

ce

eventually led Feyerabend to the Incom

mensurability T

hesis and to a generalpositio

n th

at he h

imself h

as characterized

as 'epistem

olo

gical an

archism

' or

'dad

aism'. A

ndersso

n ex

amin

es Fey

eraben

d's criticism

of L

akato

s's meth

od-

olo

gy

of scien

tific research p

rog

ram an

d o

f Lak

atos's id

ea of a 'ratio

nal

recon

structio

n' o

f the h

istory

of scien

ce. He th

en an

alyzes F

eyerab

end

'scriticism

. of P

opperian

meth

odolo

gy. F

eyerab

end sh

arpen

s the relativ

isticim

plicatio

ns o

f Kuhn's p

ositio

n. W

ith K

uhn an

d L

akato

s he b

elieves th

at, in'n

orm

al science', th

eories are relativ

ely resistan

t again

st falsification, an

d h

eex

plain

s that alleg

ed fact b

y assertin

g th

at a falsified th

eory

can alw

ays b

edefen

ded

with

the h

elp o

f auxiliary

hypoth

eses or

ad

ho

chypoth

eses. He

regards such a strategy as economical, recom

mendable or at least defensible,

because

it reduces th

e risk

that a

theory

is pre

matu

rely

abandoned.

Abandoning a theory 'too early' could involve high opportunity costs in term

sof scien

tific pro

gress fo

regone-th

e pro

gress w

e mig

ht h

ave ach

ieved

if only

we h

ad g

iven

the falsified

theo

ry m

ore o

pp

ortu

nity

to sh

ow

its wo

rth.

How

ever, in the context of Feyerabend's 'epistem

ological anarchism', it is not

possib

le to g

ive an

objectiv

e explicatio

n o

f the id

ea of scien

tific pro

gress-

accord

ing to

him

there are n

o o

bjectiv

e or g

eneral criteria. T

he o

nly

thin

gF

eyerab

end

can d

o is-h

ow

he h

imself h

as exp

ressed it-to

use th

e term'scien

tific pro

gress' in

the sen

se in w

hich

'oth

ers use it'. F

eyerab

end

has

abandoned the problem of rational theory preference, because he regards it as

unsolvable.A

ndersson proceeds by analyzing Feyerabend's case histories, in, particular,G

alileo's defense of the Copernican system

by introducing new auxiliary

hypotheses like the hypothesis about the reliability of observations through atelescope and the hypotheses that introduce a new

dynamics. L

ike Kuhn,

Feyerabend uses the case studies to support the Incomm

ensurability Thesis, in

particular, his claim that the P

tolemaic and the C

opernican systems are

optically and dynamically incom

mensurable.

The analysis of Feyerabend's criticism

of Popperian methodology show

s thathis criticism

, exactly like Kuhn's, hinges on the position he takes w

ith respectto the tw

oaforesaid

methodological problem

s:(1) the problem

of the theorydependence of observation, and (2) the problem

of the empirical testing of

Refined

Falsificationisrn

Meets

Relativist

Ph

ilosoph

yof S

cience Challenge

279

theoretical systems.

That m

eansthat the challenge to falsificationism

from the

history of science is basically a discussion of methodological problem

s with the

help of examples culled from

the history of science.T

hus,Andersson now

turnsto the analysis of these tw

o basic methodological problem

s.

The sixth chapter is devoted to the problem

solutions that Popper has offered tothe so-called problem

of the empirical base. A

ndersson shows that P

opper'sform

al requirement that a basic statem

ent should have the form of a 'there-is'-

sentence has the unacceptable consequence that basic statements cannot

contradict each other and therefore are unfalsiflable (and hence also non-scientific). A

ndersson shows that this difficulty can be overcom

e by a slightm

odification of the Popperian requirem

ent, and he proposes thatall

sorts ofsingular sentences that are properly individuated and describe observablephenom

ena qualify as basic statements or as test statem

ents. This proposal is

supported by the metalogical equivalence betw

een explanation and falsifica-tion (w

hich has been demonstrated in the A

ppendix). For the empirical testing

of theories the importance of the reproducibility of the effects described in the

basic statements can scarcely be overrated.

Popper's basic statem

ents follow from

a conjunction of singular statements

that d

escribe th

e anteced

ent co

nditio

ns an

d a n

egated

unco

nditio

nal p

redic-

tion. It is ap

pro

priate to

view

a falsification arg

um

ent as an

argum

ent w

hose

premises consist of the antecedent conditions A

and a negated unconditionalprediction _1P

.T

his in

terp

reta

tion

is pre

fera

ble

to th

e c

usto

mary

on

eacco

rdin

g to

which

the p

remises co

nsist o

f a single b

asic statemen

t, main

ly(b

ut n

ot ex

clusiv

ely) b

ecause th

ereby th

e relationsh

ip b

etween

falsification

and th

e ded

uctio

n o

f pred

ictions is clearly

show

n: 'A

, HI-P

' bein

g m

etalogi-

callyeq

uiv

alent to

'A,,P

I--iH'.

Hence,

An

dersso

n's ex

plicatio

n o

f the

concept of a Falsifying A

rgument is w

ider than that of Popper.

Popper claim

s that a critical discussion of theory-dependent test statements

ispossible, but he has given only som

e hints on how this could be done.

Andersson show

s in detail how unproblem

atic test statements can be derived

from problem

atic ones with the help of auxiliary hypotheses. It is alw

ayspossible from

two theories that describe the sam

e sort of phenomenon but are

allegedly incomm

ensurable to deduce further test statements until one arrives

at test statements that are unproblem

atic in the sense that they are neutral vis-i -v

is the tw

o co

mp

eting

theo

ries. In o

rder to

sub

stantiate th

is claim,

Andersson analyzes som

e of Kuhn's and F

eyerabend's historical case studiesand dem

onstrates in detail how theories, w

hich according to Kuhn and

Feyerabend, are incom

mensurable, can be m

ade comm

ensurable by deducingfurther test statem

ents that are unproblematic. T

hus, he can show that, e.g.,

the Copernican and P

tolemaic theories turn out to be optically and dynam

i-cally com

mensurable, and that the phlogiston theory and the oxygene theory

can b

e com

pared

with

each o

ther. It tran

spires th

at a falsification

ist

280

Gerard

Radnitzky

interpretation of the 'classical' case studies of Kuhn and F

eyerabend is far

superior to the interpretations that Kuhn and F

eyerabend have offered. The

incomm

ensurability thesis evaporates.

The seventh chapter is devoted to the problem

of the modification of theoretical

systems and theories after a falsification. P

opper's anti-conventionalist

methodological proposals are discussed. A

s is well-know

n, Popper recom

-

mends that new

auxiliary hypotheses that are introduced after a falsification

must m

eet the following requirem

ents: (1) Their introduction m

ust not reduce

the empirical content of the theoretical system

, but should increase it; (2) they

should be independently testable, and (3) their introduction should be viewed

as a rebuilding of the theoretical system. B

y contrast, Andersson argues that,

for logical reasons, adding an auxiliary hypothesis to a theoretical system can

never lead to a reduction of the empirical content of that system

, and that

therefore it is impossible to neutralize a falsification by adding new

hypotheses

to the premises of a falsification argum

ent. Popper's first requirem

ent is

superfluous. It is not recomm

endable to require that the newly introduced

auxiliary hypotheses should be independently testable. Such a requirem

ent

would a lim

ine preclude modifications of the theoretical system

that might

prove profitable in terms of new

knowledge. It is sufficient to regard the new

auxiliary hypotheses as a modification of the theoretical system

and to test

them as parts

of the theoreticalsystem

. The falsificationist m

ethodology only

requires that, after a falsification, the theoretical system be m

odified; and this

by no means entails that according to falsificationist m

ethodology a falsified

theoretical system has to be 'rejected' in

totoor 'abandoned'.

Thus, K

uhn's claim that in 'norm

al science' scientists react to a falsification

by modifying the falsified system

but not by abandoning it, is consistent with

falsificationism. T

herefore, many of K

uhn's case studies cannot function as

'counter examples' against falsificationism

. Likew

ise, the strategy recom-

mended by F

eyerabend, i.e., the interplay of the principle of proliferation and

the principle of tenacity, iscom

patiblew

ith falsificationism. T

he conventionalist

objections against falsificationism (w

hich Lakatos accepted), nam

ely the claim

that falsifications can always be outm

anoeuvred merely w

ith the help of adhoc

hypotheses, is false. The only rational w

ay of reacting to a falsification is by

modifying

the theoretical system. O

f course, in any concrete case it is impossible

ex ante to know w

hether modifying the theoretical system

or completely

remaking it is the better strategy. O

nly with the benefit of hindsight, only after

having tested the empirical consequences that w

e have got by applying those

strategies, can we tell w

hich strategy has proved to be more profitable in the

case at hand.F

eyerabend has based his thesis of the criticism-deflecting effect of ad hoc

hypotheses on certain case studies. Andersson discusses in detail G

alileo's

introduction of new auxiliary hypotheses, i.e., the

hypothesisthat

the telescopeis

Refined P

alsificationism M

eetsR

elativistP

hilosophy of Science Challenge

281

areliable instrum

entfo

rastronom

icalobservations,

and his replacing the

Aristotelean dynam

icsby different dynam

icalauxiliary

hypotheses(circular

inertia). Andersson show

s that these auxiliary hypotheses did not function as

devices that deflect criticism or neutralize a falsification. T

hey modified the

theoretical system, and equally im

portantly, they made explicit certain

auxiliary hypotheses which so far had rem

ained implicit-for instance the

hypothesis that astronomical observations w

ith the naked eye are reliable, and

the assumption that the A

ristotelean theory of motion is correct. T

he new

auxiliary hypothesis about the reliability of the telescope can be regarded

either as a part of the theoretical system to be tested or as an independently

testable premise in the falsifying argum

ent. The sam

e holds, mutatis m

utandis,

of the new dynam

ical hypotheses that Galileo used to explain the tow

er

experiment. F

eyerabend regards Galileo's hypotheses as ad

hocin

an

objectionable sense, because Galileo introduced them

after the alleged

falsification of the traditional assumptions about the m

otion of the earth that

the tower experim

ent was supposed to test, and he thinks that G

alileo had

introduced them w

ith a view to deflecting the criticism

against the Copernican

system. H

owever, for a m

ethodological appraisal of a theory it is irrelevant

when an auxiliary hypothesis has been introduced, and it is also irrelevant

what psychological m

otives may have prom

pted the researcher to introduce it.

4 S

TO

CK

TA

KIN

G

The challenge to falsificationist m

ethodology from exam

ples culled from the

history of science has proved a powerful incentive for efforts to solve the tw

o

methodological problem

s that underlie that criticism. T

he theory dependence

of experience and the revisability of basic statements are in harm

ony with

falsificationist methodology. T

hey need not lead to relativism, because it can be

shown how

basic statements can be criticized. T

he so-called problem of

incomm

ensurability can be solved, or better dissolved, by the deduction of

unproblematic test statem

ents from problem

atic ones. The research strategy

thatK

uhn claim

s to b

e typical fo

r 'norm

al science', i.e., reactin

g to

falsifications by attempting to m

odify the theoretical system hit by the

falsification but not totally 'abandoning' it, is in perfect agreement w

ith

falsificationist methodology. Falsificationist m

ethodology only requires that a

falsified theoretical system has to be changed in som

e way. It could not request

that a falsified theoretical system be rejected in the sense that a researcher w

ho

tried to improve such a system

couldeo ipso

be accused of having adopted an

irrational investment strategy-investm

ent of time and effort into a particular

researchproject

(see,e.g.,

Radnitzky

[1987b],' ... "E

conomic

Approach" ... '). T

he introduction of new auxiliary hypotheses m

odifies the

theoretical system concerned, but that introduction need not be interpreted as

a criticism-deflecting device. F

alsificationist methodology cannot give any

282

Gerard

Radriitzky

advice as to the best way of reacting, to a falsification, w

hether a minor

modification of the old theoretical system

or a complete replacem

ent of it by anew

one will lead to scientific progress. T

here is no 'logic' of scientific discoveryin the sense of an algorithm

.In the 20th century, the philosophy of science has produced three im

portant

'schools' o

r styles: p

ositiv

ism,

Wittg

enstein

's later philo

sophy, an

d C

riticalR

ationalism. P

opper and Wittgenstein II are both critics of positivism

. Popper

criticized positivism m

uch earlier than Wittgenstein did. Y

et, Wittgenstein is

stillw

idely seen as the main critic of positivism

.T

he 'n

ew p

hilo

sophy o

f science' criticizes th

e positiv

ist appro

ach to

the

philo

sophy o

f science, an

d it w

rongly

view

s Popper as a fello

w-p

ositiv

ist. It

imputes that P

opper believes 'basic statements' to provide an epistem

ologicalrockbottom

and that falsifications definitely disprove the theory falsified, i.e.,that falsifications function like negative verifications. 'T

he new philosophy of

science' keeps in the context of justificationist philosophy. Wittgenstein's later

philo

sophy starts fro

m 'p

ractice'. In p

ractice, the Id

eal Lan

guag

e does n

ot

function. The starting point of 'the new

philosophy of science' is the practice ofresearch

as it is describ

ed an

d ex

plain

ed in

the h

istory

of scien

ce-hen

ce, the

'challen

ge to

meth

od

olo

gy

from

the h

istory

of scien

ce'. It com

es to th

e

conclu

sion th

at a

study o

f the h

istory

of sc

ience sh

ow

s that C

ritical

Rationalism

gives methodological recom

mendations or prescriptions that are

unrealistic. In Lakatos's view

the history of science falsifies falsificationism.

The m

ain resu

lts of A

ndersso

n's in

vestig

ations are th

e follo

win

g:

(1)T

he challenge to the philosophy of science from the history of science

hinges on methodological considerations, in particular, on tw

o fundamen-

talm

ethodological problems: the problem

of how em

pirically to testtheoretical system

s, and the problem of how

to criticize, how to 'test' basic

statements (test statem

ents).(2)

The criticism

brought forward by 'the new

philosophy of science' is foundto be partly justified. It draw

s attention to weak spots in the m

ethodologyof C

ritical Rationalism

. Popper has only dealt w

ith the empirical testing of

isolated hypotheses-not of theoretical systems, w

hich are what usually is

at stake in the practice of research. One reason for this m

ay be the fact that

an isolated universal statement can be falsified w

ith the help of existentialstatem

ents ('there-is'-statements), and that it is relatively easy to show

how this functions. It is by far m

ore difficult to show how

a theoreticalsystem

as a whole can be falsified. In A

ndersson's book that problem has

been solved by demonstrating the m

etalogical equivalence of explanation,

prediction, and falsification. This upgrading of the m

ethodology of Critical

Rationalism

is completely new

.(3)

By

sho

win

g h

ow

a critical discu

ssion

, an em

pirical testin

g o

f teststatem

ents, can be done, Andersson develops som

e ideas that Popper has

n

Refined Falsificationism

Meets R

elativist Philosophyof Science

Challenge

283

only sketched. He show

s in detail how unproblem

atic test statements can

be deduced from problem

atic ones with 'the help of auxiliary hypotheses.

These statem

ents are u

npro

blem

atic in th

e sense th

at they

can b

e

intersubjectively tested so that it becomes possible that adherents to

different 'paradigms' agree to such an unproblem

atic test statement.

Hence, the so-called incom

mensurability problem

has been solved, or

perhaps more accurately speaking, has been dissolved. T

hen, Andersson

tests his contentions by applying them in a detailed discussion of som

e of

the 'classical' case studies which K

uhn and Feyerabend have submitted in

support of the Incomm

ensurability Thesis.

In summ

ary, Andersson has solved tw

o important m

ethodological prob-

lems. T

his made it possible to m

eet the challenge to the methodology of C

ritical

Rationalism

from the history of science. T

he methodological problem

s have

been solved by processing the methodology of C

ritical Rationalism

. In its

capacity to solve methodological problem

s the revised version of Critical

Rationalism

is far superior to both positivism and W

ittgensteinian relativism.

Refuting the claim

s of relativism w

ith respect to methodology w

ill have

important im

plications for the discussion of relativism in m

oral and political

philosophy. Critical argum

entation has been shown to possess a greater

problem solving capability than propaganda, persuasion, or other non-

rational or irrational methods.

The challenge to falsificationism

from, the history of science has led to

intellectual progress, i.e., to the processing of falsilicationist methodology. L

ike

Popper's classic of 1934 A

ndersson's book will m

ake an impact only w

hen it

has become available in the new

linguafran

ca.It is to be hoped that it w

ill not

have to wait for an E

nglish translation as long as Popper's volum

e.

GER

AR

D R

AD

NITZK

Y

University

of Trier,

West G

erman

y

RE

FE

RE

NC

ES

AN

DE

RS

SO

N, G

. [1984a]: 'Problem

s in Ludw

ik Fleck's conception of science',

Meth

od-

ologyand

Science,

17, p

p. 2

5-3

4; Italian

translatio

n in

Parad

igm

i, 2, p

p. 4

79-5

05.

AN

DE

RSSO

N, G

. (ed.). [1984b]: Rationality in

Science and

Politics. (B

oston Studies in the .

Philosophyof S

cience,V

ol. 79), Dordrecht: R

eidel.

BA

RT

LE

Y, W

. W.,111.[19621:

The

Retreat to C

omm

itment. N

ew Y

ork, NY

: Alfred A

. Knopf;

2nd ed. [1984] with 100 pp. A

ppendices, La S

alle, IL: O

pen Court.

BA

RT

LE

Y, W

. W., III. [1987a]: 'A

refutation of the alleged comprehensively critical

rationalism' in R

adnitzky and Bartley (eds.), pp. 313-42.

BA

RT

LE

Y, W

. W., III. [1987b]: 'P

hilosophy of biology versus philosophy of physics'. in

Radnitzky and B

artley (eds.), pp. 7-46.

-

-

284

Gerard

Radnitzky

BA

RTL

EY

, W. W

., 111. [1989]:U

nfathon ►e'dK

nowledge,

Unm

easuredW

ealth. On U

niversitiesand the

Wealth

ofN

ations. La S

alle, IL: O

pen Court,

forthcoming.

FLE

CK

, L. [1979]: G

enesis andD

evelopment of a

ScientificF

act,in T

haddeus J. Trenn and

Robert K

. Merton (eds.) translation by F

red Bradley. C

hicago: University of C

hicagoP

ress; Germ

an original [1935], Entstehung

andE

ntwicklung

einerw

issenschaftlichenT

atsache.E

infahrungin die L

ehre vomD

enkstil andD

enkkoilektiv.B

asel: Benno

Sch

wab

e; reprin

ted [1

98

0], F

rank

furt am

Main

: Su

hrk

amp

.FE

YE

RA

BE

ND

, P. [1975]: Against M

ethod. Outline

ofan

Anachistic T

heory of Know

ledge.L

ondon: New

Left B

ooks.JA

RV

IE, I. C

. [1988]: 'Explanation, reduction and the sociological turn in the philosophy

of science or Kuhn as ideologue for M

erton's theory of science', in Radnitzky (ed.),

pp

. 29

9-3

20

.K

UH

N, T

. S. [1962]:

The S

tructure of Scientific R

evolutions.C

hicag

o, IL

: Ch

icago

University P

ress.L

AK

AT

OS, I. and M

USG

RA

VE

, A. (eds.) [1970]: C

riticismand the

Grow

thof K

nowledge.

London: C

ambridge U

niversity Press.

MU

NZ

, P. [1987]: 'P

hilosophy and the Mirror of R

orty', in Radnitzky and B

artley (eds.),p

p. 3

45

-99

.PO

PPER

, K. [1934]:D

ie Logik der F

orschung. Wien: S

pringer; 7. verb. Aufl. T

ubingen: J. C.

B.

Mo

hr [1

98

2].

PO

PP

ER

, K. [1959]:

The L

ogic of Scientific D

iscovery.L

on

do

n: H

utch

inso

n; 1

1th

ed.

[19

82

]: Germ

an o

rigin

al 19

34

(cop

yrig

ht 1

93

5) L

og

ikder F

orschung.W

ien:S

pringer.R

AD

NIT

ZK

Y, G

. [1987a]: Entre W

ittgenstein etPopper.

Detours vers

la Decouverte: L

e Vrai,

leF

aux,I'H

ypothese.P

aris: Vrin.

RA

DN

ITZ

KY

, G. [1987b]: 'T

he "economic" approach to the philosophy of science', T

heB

ritish Journalfor the Philosophy of

Science 38, pp.159-79.

RA

DN

ITZ

KY

, G. (ed.) [1988]:

Centripetal F

orces in the Sciences,

Vol. 2. N

ew Y

ork, NY

:P

aragon House P

ublishers.R

AD

NIT

ZK

Y, G

. and AN

DE

RSSO

N, G

. (eds.) [1978]:P

rogress andR

ationality inScience.

(Boston S

tudiesin

the Philosophy of S

cience,V

ol 5

8), D

ord

recht: R

eidel; also

inItalian

, Germ

an, an

d S

pan

ish tran

slation

.R

AD

NIT

ZK

Y, G

. and AN

DE

RSSO

N, G

. (eds.). [1979]:T

he Structure

andD

evelopment of

Science. (Boston

Studies in the P

hilosophyof Science,

vol. 59), Dordrecht: R

eidel; alsoin

Italian, G

erman

, and

Sp

anish

translatio

n.

RA

DN

ITZKY

, G. and B

AR

TLEY

, W. W

., III(eds.)

[1987]:E

volutionary Epistem

ology, Theory

ofRationality, and

the Sociology of Know

ledge.L

a Salle, IL

: Open C

ourt.STEG

MU

LL

ER, W

. [1973]:P

robleme

and Resultate

der Wissenschaftstheorie

andanalytischen

Philosophie. 2.2.:

Theorienstrukturen and T

heoriendynarnik.B

erlin: Springer.

WIT

TG

EN

ST

EIN

, L. [1921]: 'L

ogisch-philosophische Abhandlung',

Annalen der

Naturphi-

losophie.(Hg.

Ostw

ald)W

ien,14, pp. 195-262; bilingual edition: T

ractatusL

ogico-P

hilosophicus. London: K

egan Paul, F

rench and Trubner [1922]: new

translationby D

. Pears and B

. McG

uinnes [1961].W

ITT

GE

NST

EIN

, L. [1953]: Philosophische

Untersuchungen/P

hilosophicalInvestigations,

bilingual edition by G. E

. M. A

nscombe and R

. Rhees (eds.). O

xford: Basil B

lackwell.

ti,