ABSTRACT

    In this paper I argue that aim-oriented empiricism (AOE),

a conception of natural science that I have defended at some

length elsewhere, is a kind of synthesis of the views of

Popper, Kuhn and Lakatos, but is also an improvement over the

views of all three.  Whereas Popper's falsificationism

protects metaphysical assumptions implicitly made by science

from criticism, AOE exposes all such assumptions to sustained

criticism, and furthermore focuses criticism on those

assumptions most likely to need revision if science is to make

progress.  Even though AOE is, in this way, more Popperian

than Popper, it is also, in some respects, more like the views

of Kuhn and Lakatos than falsificationism is.  AOE is able,

however, to solve problems which Kuhn's and Lakatos's views

cannot solve.

POPPER, KUHN, LAKATOS AND AIM-ORIENTED EMPIRICISM

1  Introduction                                       

2  Karl Popper                                       

3  Refutation of Bare Falsificationism                

4  Refutation of Dressed Falsificationism            

5  From Falsificationism to Aim-Oriented Empiricism    

6  Aim-Oriented Empiricism an Improvement over                

  Falsificationism                                    

7  Thomas Kuhn                                         

8  Imre Lakatos                                        

References                                             

Notes                                                  

1 Introduction

    In this paper I argue that aim-oriented empiricism (AOE),

a conception of natural science that I have spelled out and

defended at some length elsewhere,[1] is a kind of synthesis

of the views of Popper, Kuhn and Lakatos, but is also an

improvement over the views of all three.

 

    I begin with Karl Popper, and argue that AOE can be seen

to emerge as a result of modifying Popper's

falsificationism[2] to remove defects inherent in that

position.  AOE does not, however, break with the spirit of

Popper's work; far from committing the Popperian sin of

"justificationism", AOE is even more Popperian than Popper, in

that it is a conception of science which exposes more to

effective criticism than falsificationism does.

Falsificationism, in comparison, shields substantial,

influential and problematic scientific assumptions from

criticism within science.  Whereas falsificationism fails to

solve what may be called the "methodological" problem of

induction, AOE successfully solves the problem.  And,

associated with that success, AOE also solves the problem of

what it means to assert of a physical theory that it is

"simple", "explanatory" or "unified", a problem which

falsificationism fails to solve.

    The conception of science expounded by Thomas Kuhn in his

The Structure of Scientific Revolutions (1970) shares

important elements with Popper's falsificationism.  The big

difference is that whereas Kuhn holds that "normal science" is

an important, healthy and entirely rational (indeed, the most

rational) part of science, Popper regards normal science as

"dogmatic", the result of bad education and "indoctrination",

something that is "a danger to science and, indeed, to our

civilization" (Popper, 1970, 53).  It is the apparent

persistent dogmatism of normal science  -  the persistent

retention of the current paradigm in the teeth of ostensible

empirical refutations  -  that is so irrational, so

unscientific, when viewed from a falsificationist perspective.

AOE, however, though subjecting scientific assumptions to even

greater critical scrutiny than Popper's falsificationism,

turns out to have features which are, in some respects, closer

to Kuhn than to Popper.  For, according to AOE, substantial

and influential metaphysical assumptions are persistently

accepted as a part of scientific knowledge in a way which

seems much closer to the way paradigms are accepted, according

to Kuhn, during normal science, than to the way falsifiable

theories are to be treated in science, according to Popper.

AOE depicts science as, quite properly, proceeding in a way

that is reminiscent, in important respects, of Kuhn's normal

science, something that is anathema to Popper's falsificat-

ionism.  At the same time, AOE is free of some of the serious

defects inherent in Kuhn's conception of science.  Even though

AOE science mimics some aspects of Kuhnian normal science, it

nevertheless entirely lacks the harmful dogmatism of this kind

of science, and avoids problems that arise from Kuhn's

insistence that successive paradigms are "incommensurable".

    Imre Lakatos's "methodology of scientific research

programmes",[3] was invented, specifically, to do justice both

to Popper's insistence on the fundamental importance of

subjecting scientific theories to persistent, ruthless

attempted empirical refutation, and to Kuhn's insistence on

the importance of preserving accepted paradigms from

refutation, scientists, not paradigms, being under test when

ostensible refutations arise.  It is, like AOE, a kind

synthesis of the ideas of Popper and Kuhn.  Just as AOE

incorporates elements of Popper and Kuhn, so too it

incorporates elements of Lakatos's research programme

methodology.  At the same time, AOE is an improvement over

Lakatos's view; it solves problems which Lakatos's view is

unable to solve.  Whereas Lakatos's view provides no means for

the assessment of "hard cores" (Lakatos's "paradigms") other

than by means of the empirical success and failure of the

research programmes to which they give rise, AOE specifies a

way in which "hard cores" (or their equivalent) can be

rationally, but fallibly assessed, independent of the kind of

empirical considerations to which Lakatos is restricted.  This

has important implications for the question of whether or not

there is a rational method of discovery.  It also has

important implications for the strength of scientific method.

For Lakatos, notoriously, scientific method could only decide

which of two competing research programmes was the better long

after the event, when one had proved to be vastly superior,

empirically, to the other.  "The owl of Minerva flies at

dusk", as Lakatos put it, echoing Hegel.  AOE provides a much

more decisive methodology than Lakatos's, one which is able to

deliver verdicts when they are needed, and not long after the

event.

    During the last 30 years, since the publication of

Criticism of the Growth of Knowledge (Lakatos and Musgrave,

1970), which sought to compare and contrast, to assess the

relative merits of, the ideas of Popper and Kuhn, and which

included Lakatos's attempted synthesis, much has, of course,

been published on scientific method.  One thinks, for example,

of Holton (1973), Feyerabend (1975), Glymour (1980), van

Fraassen (1980), Laudan (1984), Watkins (1984), Hooker (1987),

Hull (1988), Howson and Urbach (1993), Kitcher (1993),

Musgrave (1993), Dupr (1995), McAllister (1996), Cartwright

(1999).  It may be thought somewhat antediluvian for me to

compare AOE with the ideas of Popper, Kuhn and Lakatos only,

ignoring more recent developments.  Such comparisons deserve

to be done.  I leave them as intellectual exercises for the

reader.    

2 Karl Popper

    As everyone knows, Popper held that science proceeds by

putting forward empirically falsifiable conjectures which are

then subjected to severe attempts at falsification by means of

observation and experiment.  Scientific theories cannot be

verified by experience, but they can be falsified.  Once a

theory is falsified, scientists have the task of developing a

potentially better theory, even more falsifiable than its

predecessor, at least as ostensibly empirically successful as

its predecessor, and such that it is corroborated where its

predecessor was falsified.  In order to be accepted

(tentatively) as a part of conjectural scientific knowledge a

theory must (at least) be empirically falsifiable.

Non-falsifiable, metaphysical theses are meaningful, and may

influence the direction of scientific research.  There can

even be what Popper has called "metaphysical research

programmes"  -  programmes of research "indispensable for

science, although their character is that of metaphysical or

speculative physics rather than of scientific physics ... more

in the nature of myths, or of dreams, than of science"

(Popper, 1982, 165).  For Popper, metaphysical (that is,

unfalsifiable) theses cannot be a part of (conjectural)

scientific knowledge; such theses cannot help determine what

is accepted and rejected as (conjectural) scientific

knowledge, but they can influence ideas, choice of research

aims and problems, in the context of scientific discovery.

For further details see Popper (1959, 1963, 1982). 

    Popper defended two distinct versions of falsificationism

which, echoing terminology of Maxwell (1998), I shall call

bare and dressed falsificationism.  According to bare

falsificationism, defended in Popper (1959), only empirical

considerations, and such things as the falsifiability of

theories and degrees of falsifiability, decide what is to be

accepted and rejected in science.  According to dressed

falsificationism, a new theory, in order to be acceptable,

"should proceed from some simple, new, and powerful, unifying

idea about some connection or relation (such as gravitational

attraction) between hitherto unconnected things (such as

planets and apples) or facts (such as inertial and

gravitational mass) or new "theoretical entities" (such as

field and particles)" (Popper, 1963, 241).  This "requirement

of simplicity" (as Popper calls it) is in addition to anything

specified in Popper (1959).  In his (1959), Popper does, it is

true, demand of a theory that it should be as simple as

possible, but Popper there identifies degree of simplicity of

a theory with degree of falsifiability.  (There is a second,

related notion, but Popper makes it clear that if the two

clash it is the falsifiability notion, just indicated, which

takes priority: see page 130). Thus, in his (1959), in

requiring of an acceptable theory that it should be as simple

as possible, Popper is demanding no more than that it should

be as falsifiable as possible.  But Popper's "requirement of

simplicity" of his (1963) is wholly in addition to

falsifiability.  A theory of high falsifiability may not

"proceed from some simple, new, and powerful unifying idea",

and vice versa.  We thus have two versions of falsificationism

before us: bare falsificationism of Popper's (1959), and

dressed falsificationism of (1963, chapter 10), with the new

"requirement of simplicity" added onto the (1959) doctrine.

    I now give my argument for holding that neither doctrine

is tenable.  My argument is not that Popper fails to show how

theories can be verified, or rendered probable; nor is my

argument that Popper fails to show how scientific theories can

be falsified, in that falsification requires the verification

of a low-level falsifying hypothesis (which, according to

Popper, is not possible).[4]  There is nothing

"justificationist", in other words, about my criticism.  It

amounts simply to this.  Bare falsificationism fails

dramatically to do justice to the way theories are selected in

science (entirely independently of any question of

verification, justification or falsification).  Dressed

falsificationism does better justice to scientific practice,

but commits science to making substantial, influential and

problematic assumptions that remain implicit, and cannot

adequately be made explicit within science.  Science pursued

in accordance with dressed falsificationism is irrational, in

other words, because it fails to implement the elementary, and

quasi-Popperian, requirement for rationality that "assumptions

that are substantial, influential, problematic and implicit

need to be made explicit, so that they can be critically

assessed and so that alternatives may be put forward and

considered, in the hope that such assumptions can be improved"

(Maxwell, 1998, 21).  Dressed falsificationism fails, in other

words, for good Popperian reasons: it fails to expose

substantial, influential, problematic assumptions to criticism

within science.

3 Refutation of Bare Falsificationism

    Here, then, in a little more detail, is my refutation of

bare falsificationism.  Given any accepted physical theory, at

any stage in the development of physics, however empirically

successful (however highly corroborated)  -  Newtonian theory,

say, or classical electrodynamics, quantum theory, general

relativity, quantum electrodynamics, chromodynamics or the

standard model  -  there will always be endlessly many rival

falsifiable theories that can easily be formulated which will

fit the available data just as well as the accepted theory.

Taking Newtonian theory (NT) as an example of an accepted

theory, here are two examples of rival theories.  NT*:

"Everything occurs as NT asserts, until the first second of

2100, when an inverse cube law of gravitation will abruptly

hold".  NT**: "Everything occurs as NT asserts, except for

systems consisting of gold spheres, each having a mass of

1,000 tons, interacting with each other gravitationally in

outer space, in a vacuum, within a spherical region of 10

miles: for these systems, Newton's law of gravitation is

repulsive, not attractive".  (For further examples and

discussion, see Maxwell, 1998, 47-54).  It is easy to see that

there are infinitely many such rivals to NT, just as

empirically successful (at the moment) as NT.  The predictions

of NT may be represented as points in a multi-dimensional

space, each point corresponding to some specific kind of

system (there being infinitely many points).  NT has only been

verified (corroborated) for a minute region of this space.  In

order to concoct a (grossly ad hoc) rival to NT, just as

empirically successful as NT, all we need do is identify some

region in this space that includes no prediction of NT that

has been verified, and then modify the laws of NT arbitrarily,

for just that identified region. 

    The crucial question now is this: on what basis does bare

falsificationism reject all these falsifiable but unfalsified

rival theories?  According to bare falsificationism, T2 is to

be accepted in preference to T1 if T1 has been falsified, T2

has greater empirical content (is more falsifiable) than T1, T2

successfully predicts all that T1 successfully predicts, T2

successfully predicts the phenomena that falsified T1, and T2

successfully predicts new phenomena not predicted by T1 (see

Popper, 1959, 81-84 and elsewhere).  Given NT, it is a simple

matter to concoct rival theories, of the above type, that

satisfy the above bare falsificationist requirements for being

more acceptable than NT.  Most accepted physical theories

yield empirical predictions that clash with experiments, and

thus are ostensibly falsified.  We can always concoct new

theories, in the way just indicated, doctored to yield the

"correct" predictions.  We can add on independently testable

auxiliary postulates, thus ensuring that the new theory has

greater empirical content than the old one.  And no doubt this

excess content will be corroborated.  For details of how this

can be done see Maxwell (1998), 52-54.  Such theories are, of

course, grossly ad hoc, grossly "aberrant" as I have called

them; but they satisfy Popper's (1959) requirements for being

better theories than accepted physical theories. 

    It is worth noting that such "better" theories need not

be quite as wildly ad hoc as the ones indicated above;

sometimes such theories are actually put forward in the

scientific literature, and yet are not taken seriously, even

by their authors, let alone by the rest of the scientific

community.  An example is an ad hoc version of NT put forward

by Maurice Levy in 1890, which combined in an ad hoc way two

distinct modifications of Newton's law of gravitation, one

based on the way Weber had proposed Coulomb's law should be

modified, the other based on the way Riemann had proposed

Coulomb's law should be modified: for details see North

(1965).  By 1890, NT had been refuted by observation of the

precession of the perihelion of the orbit of Mercury; attempts

to salvage NT by postulating an additional planet, Vulcan, had

failed.  Levy's theory successfully predicted all the success

of NT, and in addition successfully predicted the observed

orbit of Mercury, just that which refuted NT; in addition, of

course, it made predictions different from NT for further

Sun-Mercury type systems not yet observed.  Despite this,

Levy's theory was not taken seriously for a moment, not even

by Levy himself. How can bare falsificationism recommend

rejection of such ad hoc versions of NT when they satisfy all

the requirements of bare falsificationism for being more

acceptable theories?  No adequate answer is forthcoming, and

it is this which spells the downfall of bare falsificationism

(as Popper may himself have realized when he put forward

dressed falsificationism in his (1963), chapter 10).

    Note, again, that this criticism of Popper has nothing

justificational about it whatsoever: it simply points to the

drastic failure of bare falsificationism to do justice to what

actually goes on in physics.

    It may be objected that ad hoc rivals to NT of the kind

just considered are so silly, so crackpot, that they do not

deserve to be taken seriously within physics.[5]  This is of

course correct.  The crucial point, however, is that bare

falsificationism ought to be able to deliver this verdict, and

this it singularly fails to do.  Bare falsificationism

actually declares of appropriately concocted ad hoc rivals to

NT that these are better, more acceptable than NT.

    But can a criticism of Popper that appeals to such silly,

crackpot theories be taken seriously?  I have two replies to

this question.  First, not all the ad hoc or aberrant variants

are entirely silly.  Levy's theory is perhaps an example.

There are degrees of ad hocness, from the utterly crackpot and

absurd, to a degree of ad hocness, so slight, so questionable,

in comparison, that the issue of whether the theory really is

ad hoc or not may be hotly disputed by physicists themselves.

(Such disputes arise especially during scientific

revolutions.)  This is an important point which will have a

bearing on the argument of the next section.  Second, it is, I

submit, the very silliness of these crackpot theories that

makes the above criticism of Popper so serious.  If bare

falsificationism favoured T1 over T2, while most scientists

favoured T2 over T1, even though admitting that T1 is

nevertheless a good theory, almost as acceptable as T2, bare

falsificationism would not be in such trouble.  What is lethal

for bare falsificationism is that it declares T1 to be better

than T2 in circumstances where scientists themselves (and all

of us) can see that T2 is vastly superior to T1, T1 being

grossly ad hoc, aberrant, wholly crackpot and silly.  Bare

falsificationism favours theories that receive, and deserve,

instant rejection: there could scarcely be a more decisive

falsification of falsificationism than that.

4 Refutation of Dressed Falsificationism

    Having argued that Popper's (1959) bare falsificationism

is untenable, I turn my attention now to Popper's (1963,

chapter 10) doctrine of dressed falsificationism.  As I have

mentioned, this adds onto the (1959) doctrine Popper's new

"requirement of simplicity (Popper, 1963, 241): see section 2

above.

    As long as there is no serious ambiguity as to what

proceeding "from some simple, new, and powerful, unifying

idea" means, it is at once clear that the new doctrine is able

to exclude from science all the empirically successful but ad

hoc, aberrant, crackpot, silly theories, of the kind discussed

above.  They do not proceed "from some simple...unifying

idea", and are to be rejected on that account, whatever their

empirical success may be, even if this empirical success is

greater than accepted scientific theories.

    However, adopting Popper's new "principle of simplicity"

as a basic methodological principle of science has the effect

of permanently excluding from science all ad hoc theories that

fail to satisfy the principle, however empirically successful

such theories might be if considered.  This amounts to

assuming permanently that the universe is such that no ad hoc

theory, that fails to satisfy Popper's principle of

simplicity, is true.  It amounts to accepting, as a permanent

item of scientific knowledge, the substantial metaphysical

thesis that the universe is non-ad hoc, in the sense that no

theory that fails to satisfy Popper's principle of simplicity

is true, however empirically successful it might turn out to

be if considered.  But this, of course, clashes with Popper's

criterion of demarcation: that no unfalsifiable, metaphysical

thesis is to be accepted as a part of scientific knowledge.

If the demarcation principle is upheld, then the metaphysical

thesis just indicated, asserting that the universe is non-ad

hoc, remains implicit in the permanent adoption of Popper's

principle of simplicity as a basic methodological principle of

science.  (And this is the way Popper himself seems to have

conceived the matter: he says of metaphysical research

programmes that they are "often held unconsciously", and "are

implicit in the theories and in the attitudes and judgements

of the scientists": (Popper, 1982, 161).)  But in leaving the

metaphysical thesis of non-ad hocness implicit in the

methodological principle of simplicity, science violates an

elementary requirement for rationality, according to which

"assumptions that are substantial, influential, problematic

and implicit need to be made explicit, so that they can be

critically assessed and so that alternatives may be put

forward and considered, in the hope that such assumptions can

be improved" (Maxwell, 1998, 21).  The non-ad hoc metaphysical

assumption may, after all, be false.  We may need to adopt a

modified version of the assumption.  It may be essential for

the progress of science that this assumption is modified.

Just this turns out to be the case, given certain formulations

of the assumption, as we shall see below.  In leaving the

non-ad hoc metaphysical assumption implicit in the adoption of

the methodological principle of simplicity, dressed

falsificationism protects this substantial, influential and

highly problematic assumption from criticism, from the active

consideration of alternatives.[6]

    Dressed falsificationism fails, in other words, for good

Popperian reasons: it is either inconsistent (in that the

untestable, metaphysical thesis that the universe is non-ad

hoc is held to be a part of conjectural scientific knowledge,

in conflict with the principle of demarcation), or it

irrationally protects an implicit, substantial assumption from

explicit criticism within the intellectual domain of science.

    Here again, it should be noted, there is nothing

justificationist about this criticism of Popper's dressed

falsificationism.  On the contrary, what the argument shows is

that dressed falsificationism protects a substantial,

influential, problematic but implicit assumption from

criticism within science: Popper's doctrine fails for the good

Popperian reason of restricting criticism. 

Popper, Kuhn, Lakatos and Aim-Oriented Empiricism

    It may be objected that adopting Popper's methodological

principle of simplicity does not commit science to making a

substantial metaphysical assumption about the universe  -

namely, that it is such that no falsifiable theory, however

empirically successful, which fails to satisfy the principle,

is true.  But I do not see how such an objection can be valid.

Suppose, instead of adopting Popper's principle, science

adopted the principle: in order to be acceptable, a new

physical theory must postulate that the universe is made up of

atoms.  This methodological principle is upheld in such a way

that even though theories are available which postulate fields

rather than atoms, and which are much more empirically

successful than any atomic theory, nevertheless these rival

field theories are all excluded from science.  Would it not be

clear that science, in adopting and implementing the

methodological principle of atomicity in this way, is making

the assumption that the universe is made up of atoms, whether

this is acknowledged or not?  How can this be denied?  Just

the same holds if science adopts and implements Popper's

methodological principle of simplicity. 

    Popper might have tried to wriggle out of accepting this

conclusion by pointing to the fact that he only declared that

a new theory, in order to be acceptable, "should" proceed from

some simple, unifying idea.  It is desirable, but not

essential, that new theories should satisfy this principle.

The principle is relevant to the context of discovery,

perhaps, but not to the context of acceptance and rejection.

(It is a heuristic principle, not a methodological one.)  But

if Popper's doctrine is interpreted in this way, it

immediately fails to overcome the objections spelled out in

section 3 above.  Either falsificationism adopts Popper's

principle of simplicity as a methodological principle, or it

does not.  If it does, it encounters the objections just

indicated; if it does not, it encounters the objections of

section 3.

5 From Falsificationism to Aim-Oriented Empiricism

    The conclusion to be drawn from the argument so far is

that science is more rational, more intellectually rigorous if

it makes explicit, as a criticizable tenet of (conjectural)

scientific knowledge, that substantial, influential and

problematic metaphysical thesis which is implicit in the way

physics persistently rejects ad hoc theories, however

empirically successful they may be.  At once two important new

problems leap to our attention.  What, precisely, does this

metaphysical thesis assert?  And on what grounds is it to be

(conjecturally) accepted as a part of scientific knowledge?

The conception of science which I uphold as a radical

improvement over Popper's falsificationism, namely

aim-oriented empiricism (AOE), is put forward as the solution

to these two problems.  I now give a brief exposition of AOE,

indicate how it solves the two problems just mentioned,

indicate further how it solves the methodological problem of

induction and the related problem of simplicity, and then

consider possible objections.

    As far as the first of the above two problems is

concerned, a wide range of metaphysical theses are available.

As I indicated in section 3 above, ad hoc theories range from

the utterly crackpot and silly, to theories that are only

somewhat lacking in simplicity or unity.  At one extreme, we

might adopt a metaphysical thesis that excludes only utterly

silly theories; at the other extreme, we might adopt the

thesis that the universe is physically comprehensible in the

sense that it has a unified dynamic structure, some

yet-to-be-discovered unified physical "theory of everything"

being true  -  a thesis that I shall call "physicalism".  We

might even adopt some specific version of physicalism, which

asserts that the underlying physical unity is of a specific

type: it is made up of a unified field perhaps, o, a quantum

field, or empty topologically complex curved space-time, or a

quantum string field.  Other things being equal, the more

specific the thesis (and thus the more it excludes) so the

more likely it is to be false, whereas the more unspecific it

is so the more likely it is to be true.

    As far as the second of the above two problems is

concerned, there are three considerations that we can appeal

to, wholly Popperian in spirit if not in the letter of

Popperian doctrine.

(1) If some metaphysical thesis, M, is implicit in some

scientific methodological practice, then science is more

rigorous if M is made explicit, since this facilitates

criticism of it, the consideration of alternatives.

(2) A metaphysical thesis may be such that its truth is a

necessary condition for it to be possible for us to acquire

knowledge: if so, accepting the thesis can only help, and

cannot undermine, the pursuit of knowledge of truth.

(3) Given two rival metaphysical theses, M1 and M2, it may be

the case that M1 supports an empirical scientific research

programme that has apparently met with far greater empirical

success than any rival empirical research programme based on

M2: in this case we may favour M1 over M2, at least until M2, or

some third thesis, M3, shows signs of supporting an even more

empirically progressive research programme. 

    The arguments of sections 3 and 4 have established that

physics must accept (conjecturally) some kind of metaphysical

thesis of non-ad hocness, if crackpot theories are to be

excluded: it makes sense to adopt that thesis which seems to

be the most fruitful in promoting scientific progress.  (To

say that M1 "supports" an empirically successful research

programme is to say that the programme develops a succession

of theories, each empirically more successful than its

predecessors, in a Popperian sense, and each being closer to

exemplifying, to being a precise, testable instantiation of M1

than its predecessors.)

    Two difficulties arise, however, when one attempts to use

(2) and (3) to select the best available metaphysical thesis

from the infinitely many options available.  As far as (2) is

concerned, any thesis sufficiently substantial to exclude

empirically successful crackpot theories from science is such

that acquisition of knowledge might still be possible even if

the thesis is false.  On the other hand, any thesis such that

its truth is necessary for knowledge to be acquired is much

too insubstantial to exclude crackpot theories.  As far as (3)

is concerned, given any metaphysical thesis, M, that supports

a non-crackpot empirically progressive scientific research

programme, we can  mimic this with a crackpot M* that supports

a crackpot empirically progressive research programme, with a

series of crackpot theories, T1*, T2*, ..., these theories

becoming progressively more and more empirically successful,

and closer and closer to exemplifying M*.

    These two difficulties can be overcome, however, if

physics is construed as adopting a hierarchy of metaphysical

conjectures concerning the comprehensibility and knowability

of the universe, these conjectures becoming more and more

insubstantial as one ascends the hierarchy, more and more

likely to be true: see diagram.  (A crude version is below; for

a more adequate version please request jpg file; or see either Maxwell, 1998, p. 8; or 1999, p. 142; or 2000a, p. 34; or 2001,

p. 42; or 2002a, p. 269; or 2002b, p. 6).  As I have formulated this

--------------------------------------------------------------

          DIAGRAM OF AIM-ORIENTED EMPIRICISM

Level 10   PARTIAL KNOWABILITY

  [Thesis that the universe is such that some knowledge of local                              circumstances can be acquired.]

Level 9    EPISTEMOLOGICAL NON-MALICIOUSNESS

  [Thesis that the universe is such that local knowledge can be used

  to acquire some knowledge of non-local conditions.]                      

Level 8    META-KNOWABILITY

  [Thesis that the universe is such that, relative to existing knowledge, some

    thesis can be discovered which leads to improved methods for the

    improvement of knowledge.

Level 7    ROUGH COMPREHENSIBILITY

  [Thesis that the universe is more or less comprehensible.]

Level 6    NEAR COMPREHENSIBIITY

  [Thesis that the universe is such that the best assumption science can make, at this level, in order to achieve empirical success, is that the universe is perfectly comprehensible.]

Level 5   COMPREHENSIBILITY

  [The thesis that the universe is comprehensible in the sense that                       something exists at all times and places (God, tribe of gods, cosmic purpose, cosmic programme, or unified pattern of physical law), which determines (perhaps probabilistically) all change and diversity, and in terms of which everthing can, in principle, be explained and understood.]

Level 4   PHYSICALISM

  [The thesis that the universe is physically comprehensible.]

Level 3   BEST BLUEPRINT

  [The thesis that the universe is physically comprehensible in some                        more or less specific way.]

Level 2   ACCEPTED FUNDAMENTAL PHYSICAL THEORIES

Level 1   EVIDENCE

-----------------------------------------------------------------

idea, in Maxwell (1998), the top conjecture in the hierarchy

(at level 10) asserts, roughly, that the universe is such that

some (conjectural) knowledge of our local circumstances can be

acquired.  This, and the next conjecture down are, I argue, to

be accepted as permanent items of scientific knowledge, in

accordance with (2), on the grounds, that is, that such

acceptance can only help, and cannot hinder, the search for

factual knowledge whatever the universe may be like.  At level

4 the conjecture to be adopted is, I argue, physicalism.  At

level 5 there is the less precise conjecture that the universe

is comprehensible in some way or other; and as one goes up the

hierarchy to levels 9 and 10, the conjectures become

progressively less and less substantial and precise.  At level

3 there is the best currently available more or less specific

version of physicalism, which I call the current "metaphysical

blueprint".  Examples from the history of physics are: the

universe consists of (a) corpuscles which interact by contact

(b) point-atoms which interact by means of forces (c) a

unified classical field (d) a unified quantum field (e) empty,

curved, topologically complex space-time (f) a unified quantum

string field.  At level 2 are currently accepted fundamental

physical theories, and at level 1 there are empirical data.

Two considerations govern acceptance of metaphysical

conjectures from level 3 to level 8.  Any such conjecture

must, as far as possible (A) exemplify, be a precise version

of, and imply, the next conjecture up in the hierarchy, (B) be

more empirically fruitful than any rival conjecture, in that

it is a part of an empirical research programme that seems to

be more empirically progressive than any rival research

programme, in accordance with (3) above.  Two considerations

also govern acceptance of testable fundamental dynamical

physical theories.  Such a theory must be such that (i) it,

together with all other accepted fundamental physical

theories, exemplifies, or is a special case of, the best

available metaphysical blueprint (at level 3) to a

sufficiently good extent, (ii) it is sufficiently successful

empirically (where empirical success is to be understood,

roughly, in a Popperian sense).

    This hierarchical view of AOE overcomes the two

difficulties, indicated above, roughly as follows.  Only the

top two theses are accepted as a result of an appeal to (2);

theses at levels 3 to 8 are accepted as a result of (a) an

appeal to (3), and (b) compatibility with the top two theses

at levels 10 and 9; this suffices to exclude aberrant rivals

at levels 3 to 8 (which might be construed to support

aberrant, empirically progressive research programmes).  For

further details of how AOE overcomes the two difficulties

indicated above, and for further details of the view itself,

see Maxwell (1998, chapter 5, and elsewhere).

    A basic idea of AOE is to channel or direct criticism so

that it is as fruitful as possible, from the standpoint of

aiding progress in knowledge.  The function of criticism

within science is to promote scientific progress.  When

criticism demonstrably cannot help promote scientific

progress, it becomes irrational (the idea behind (2) above).

In an attempt to make criticism as fruitful as possible, we

need to try to direct it at targets which are the most

fruitful, the most productive, to criticize (from the

standpoint of the growth of knowledge).  This is the basic

idea behind the hierarchy of AOE.  Conjectures at all levels

remain open to criticism.  But, as we ascend the hierarchy,

conjectures are less and less likely to be false; it is less

and less likely that criticism, here, will help promote

scientific knowledge.  The best currently available level 3

conjecture is almost bound to be false: the history of physics

reveals, at this level, as I have indicated above, that a

number of different conjectures have been adopted and rejected

in turn.  Here, criticism, the activity of developing

alternatives (compatible with physicalism) is likely to be

immensely fruitful for progress in theoretical physics.

Indeed, in Maxwell (1998), 78-89, 159-163 and especially

217-223, I argue that this provides physics with a rational,

though fallible and non-mechanical method for the discovery of

new fundamental physical theories, a method invented and

exploited by Einstein in discovering special and general

relativity (Maxwell, 1993, 275-305 ), something which Popper

has argued is not possible: see Popper (1959), 31-32.

Criticizing physicalism, at level 4, may also be fruitful for

physics, but (the conjecture of AOE is) that this is not as

likely to be as fruitful as criticism at level 3.  (Elsewhere

I have suggested alternatives to physicalism.)  And, as we

ascend the hierarchy (so AOE conjectures), criticism becomes

progressively less and less likely to be fruitful.  Against

that, it must be admitted that the higher in the hierarchy we

need to modify our ideas, so the more dramatic the

intellectual revolution that this would bring about. If

physicalism is rejected altogether, and some quite different

version of the level 5 conjecture of comprehensibility is

adopted instead, the whole character of natural science would

change dramatically; physics, as we know it, might even cease

to exist.

    The biggest change, in moving from falsificationism to

AOE, has to do with the role of metaphysics in science, and

the scope of scientific knowledge.  According to

falsificationism, untestable metaphysical theses may influence

scientific research in the context of discovery, and may even

lead to metaphysical research programmes; they cannot,

however, be a part of scientific knowledge itself.  But

according to AOE, the metaphysical theses at levels 3 to 10

are all a part of current (conjectural) scientific knowledge.

In particular, physicalism is.  According to AOE, it is a part

of current scientific knowledge that the universe is

physically comprehensible  -  certainly not the case granted

falsificationism.

    Another important change has to do with the relationship

between science and the philosophy of science.

Falsificationism places the study of scientific method, the

philosophy of science, outside science itself, in accordance

with Popper's demarcation principle.  AOE, by contrast, makes

scientific method and the philosophy of science an integral

part of science itself.  The activity of tackling problems

inherent in the aims of science, at a variety of levels, and

of developing new possible aims and methods, new possible more

specific or less specific philosophies of science (views about

what the aims and methods of science ought to be) is,

according to AOE, a vital research activity of science itself.

But this is also philosophy of science, being carried on

within the framework of AOE.[7]

    AOE differs in many other important ways from Popper's

falsificationism, whether bare or dressed (see Maxwell, 1998).

Nevertheless the impulse, the intellectual aspirations and

values, behind the hierarchical view of AOE are, as I have

tried to indicate, thoroughly Popperian in character and

spirit.  The whole idea is to turn implicit assumptions into

explicit conjectures in such a way that criticism may be

directed at what most needs to be criticized from the

standpoint of aiding progress in knowledge, so that

conjectures may be developed and adopted that are the most

fruitful in promoting scientific progress, at the same time no

substantial conjecture, implicit or explicit, being held

immune from critical scrutiny.

6  Aim-Oriented Empiricism an Improvement over

Falsificationism

    AOE is also, in a number of ways, a considerable

improvement over Popper's falsificationism. 

1. Consistency.  Bare falsificationism fails dramatically to

do justice to scientific practice, and is an inherently

unworkable methodology, in any case.  (In what follows I shall

mostly ignore bare falsificationism as obviously untenable,

and concentrate on comparing dressed falsificationism and

AOE.)  Dressed falsificationism does better justice to

scientific practice, but at the cost of consistency;

persistent rejection of empirically successful theories that

do not "proceed from some simple...unifying idea" commits

science to accepting a metaphysical thesis of simplicity as a

part of scientific knowledge (though this is not recognized);

this contradicts Popper's demarcation principle.  AOE is free

of such lethal defects.

2. Criticism.  Pursuing physics in accordance with dressed

falsificationism protects the implicit metaphysical thesis of

simplicity from criticism within science itself, just because

this thesis is metaphysical (and therefore not a part of

science) and implicit (and therefore not available for

sustained, explicit critical scrutiny).  AOE, by contrast, is

specifically designed to provide a framework of metaphysical

assumptions and corresponding methodological rules within

which level 3 metaphysical blueprints may be developed, and

critically assessed, within science.

3. Rigour.  Science pursued in accordance with AOE is more

rigorous than science pursued in accordance with

falsificationism.  An elementary, but important requirement

for rigour is that assumptions that are substantial,

influential, problematic and implicit need to be made explicit

so that they can be criticized, and so that alternatives can

be considered.  If the attempt is made to do science in

accordance with falsificationism, bare or dressed, one

substantial, influential and problematic assumption must

remain implicit (as we have just seen), namely the

metaphysical assumption that nature behaves as if simple, no

ad hoc theory being true.  This is implicit in the adoption of

the simplicity methodological principle of dressed

falsificationism.  AOE, by contrast, makes this implicit

assumption explicit, and provides a framework within which

rival versions can be proposed and critically assessed.

4. Simplicity.  Falsificationism fails to say what the

simplicity of a theory is.  Bare falsificationism provides an

account of simplicity in terms of falsifiability, but we have

already seen that this account is untenable.  Popper's 1963

"requirement of simplicity" appeals to a conception of

simplicity or unity that is wholly in addition to

falsifiability, but does not explain what the simplicity or

unity of a theory is.  It fails to explain how the simplicity

of a theory can possibly be methodologically or

epistemologically significant when a simple theory can always

be made complex by a suitable change of terminology, and vice

versa.  Popper himself recognized the inadequacy of his

simplicity requirement when he called it "a bit vague", said

that "it seems difficult to formulate it very clearly", and

acknowledged that it threatened to involve one in infinite

regress (Popper, 1963, 241).  By contrast, AOE solves the

problems of explaining what the simplicity or unity of a

theory is without difficulty.  The totality of fundamental

physical theory, T, is unified to the extent that its content

exemplifies physicalism.  The more the content of T departs

from exemplifying physicalism, the more disunified T is.[8]

Because what matters is content, not form, the way T is

formulated is irrelevant to this way of assessing simplicity

or unity.  Falsificationism cannot avail itself of this way of

assessing unity because it involves acknowledging that

physicalism is a basic tenet of scientific knowledge,

something which falsificationism denies.  Within AOE, there is

a second way in which the unity of T may be assessed: in terms

of the extent to which the content of T exemplifies the best

available level 3 metaphysical blueprint.  This second

conception of simplicity or unity evolves with the evolution

of level 3 ideas.  As we improve our ideas about how the

universe is unified, with the advance of knowledge in

theoretical physics, so non-empirical methods for selecting

theories on the basis of simplicity or unity improve as well.

Thus current symmetry principles of modern physics, such as

Lorentz invariance and gauge invariance, which guide

acceptance of theory, are an advance over simplicity criteria

upheld by Newton.  This account of simplicity can be extended

to individual theories in two ways.  First, we may treat an

individual theory as a candidate theory of everything.

Second, given two individual theories, T1 and T2, and given the

rest of fundamental theory, T, T1 is simpler than T2 iff T + T1

is simpler than T + T2, where the latter is assessed in one or

other of the ways indicated above.[9]

    It may be objected that this proposed solution to the

problem of simplicity is circular: the unity of level 2 theory

is explicated in terms of the unity of level 4 physicalism.

But this objection is not valid.  In order to solve the

problem, it is not necessary to explicate what "simplicity" or

"unity" mean; rather, what needs to be done is to show how

theories can be partially ordered with respect to "simplicity"

or "unity" in a way that does not depend on formulation.  This

is achieved by partially ordering theories in terms of how

well their content exemplifies the content of physicalism, so

that, roughly, the more the content of a theory violates the

symmetries associated with the content of physicalism, the

less unity it has.  As long as physicalism is a meaningful

thesis, and provides a formulation-independent way of

partially ordering theories in the way indicated, this

suffices to solve the problem.  That physicalism embodies

intuitive ideas of "unity" is a bonus.  For a more detailed

rebuttal of this objection, see Maxwell (1998), 118-123.

5. Scientific Method.  Dressed falsificationism acknowledge

(correctly) that two considerations govern selection of theory

in science, namely considerations that have to do with (a)

evidence, and (b) simplicity.  But because it cannot solve the

problem of what simplicity is, dressed falsificationism

cannot, with any precision, specify what methods are involved

when theories are selected on the basis of simplicity.  Nor

can the view do justice to the way in which the methods of

physics evolve with evolving knowledge, especially methods

that assert that acceptable theories must satisfy this or that

symmetry.  In other words, falsificationism fails to solve

what may be called the "methodological" problem of induction,

the problem of specifying, merely, what the methods are that

are employed by science in accepting and rejecting theories

(leaving aside the further problem of justifying these methods

given that the aim is to acquire knowledge).  AOE, by

contrast, solves the problem of simplicity, and thus can

specify precisely what methods are involved when theories are

selected on the basis of simplicity.  Furthermore, AOE can do

justice to evolving criteria of simplicity (as we have just

seen), and hence evolving methods.  According to AOE, the

totality of fundamental physical theory, T, can be assessed

with respect to how well its content exemplifies (i) the

relatively fixed level 4 thesis of physicalism, or (ii) the

evolving, best available level 3 thesis.  Whereas (i)

constitute fixed criteria of simplicity or unity (as long as

physicalism is not modified), (ii) constitute evolving

criteria, criteria of unity that improve with improving

knowledge.

6. Evolving aims-and-methods.  A point, briefly alluded to in

4 and 5 above, deserves further emphasis.  As physics has

evolved, from Newton's time to today, non-empirical methods,

determining what theories will be accepted and rejected, have

evolved as well.  Newton, in his Principia, formulated four

rules of reasoning, three of which are concerned with

simplicity (Newton, 1962, vol 2, 398-400).  Principles that

have been proposed since his day include: invariance with

respect to position, orientation, time, uniform velocity,

charge conjugation, parity, time-reversal; principles of

conservation of mass, momentum, angular momentum, energy,

charge; Lorentz invariance; Mach's principle, the principle of

equivalence; principles of gauge invariance, global and local;

supersymmetry; duality principles; the principle that

different kinds of particle should be reduced to one kind, and

different kinds of force should be reduced to one kind; the

principle that space-time on the one hand, and particles-and-

forces on the other, should be unified.  All of these

principles can be interpreted as methodological rules which

specify requirements theories must meet in order to be

accepted.  They can also be interpreted as physical

principles, making substantial assertions about such things as

space, time, matter, force.  Some, such as conservation of

mass, parity, and charge conjugation, have been shown to be

false; others, such as Mach's principle, have never been

generally accepted; still others, such as supersymmetry,

remain speculative.

    Principles such as these, which can be interpreted either

as physical assertions or as methodological principles, which

are made explicit, developed, revised and, on occasions,

rejected or refuted, are hard to account for within the

framework of falsificationism.  It is especially difficult,

within this framework, to account for principles which (a)

have a quasi a priori role in specifying requirements theories

must satisfy in order to be accepted, but which at the same

time (b) make substantial physical assertions about the nature

of the universe.  AOE, on the other hand, predicts the

existence of such principles, with just the features that have

been indicated.  Accepted principles are components of the

currently accepted level 3 blueprint.  As the accepted

blueprint evolves, these principles, interpreted either as

physical or methodological principles, evolve as well.

Indeed, according to AOE, these principles, and associated

blueprints, do not just evolve; they are improved with

improving theoretical knowledge.  AOE provides a more or less

fixed framework of relatively unproblematic aims-and-methods

(at level 4 or above) within which highly problematic level 3

aims-and-methods[10] may be improved in the light of the

empirical success and failure of rival research programmes

(adopting rival level 3 aims-and-methods).  In other words,

AOE provides a framework within which there can be positive

feedback between improving scientific knowledge and improving

aims-and-methods.  As knowledge improves, knowledge-about-how-

to-improve-knowledge improves as well.  This capacity of

science to adapt itself  -  its methods  -  to what it finds

out about the universe is, according to AOE, the

methodological key to the astonishing progressive success of

science.  Falsificationism, with its fixed aim and fixed

methods, is quite unable to do justice to this positive

feedback, meta-methodological feature of science, this

capacity of science to learn about learning as it proceeds. 

7. Verisimilitude.  The so-called problem of verisimilitude

arises because physics usually proceeds from one false theory

to another, thus rendering obscure what it can mean to say

that science makes progress.  Popper (1963, chapter 10 and

Addenda) tried to solve this problem within the framework of

falsificationism but, as Miller (1974) and Tichy (1974) have

shown, this attempted solution does not work.  Not only does

falsificationism fail to specify properly the methods that

make progress in theoretical physics possible; it fails even

to say what progress in theoretical physics means.

    AOE solves the problem without difficulty.  First, the

fact that physics does proceed from one false theory to

another, far from undermining physicalism, and hence AOE as

well, is just the way theoretical physics must proceed,

granted physicalism.  For, granted physicalism, any theory,

T*, which captures precisely how phenomena evolve in some

restricted domain, must be generalizable to cover all

phenomena.  If T* cannot be so generalized then, granted

physicalism, it cannot be precisely true.  In so far as

physics proceeds by developing theories which apply to

restricted, but successively increasing, domains of phenomena,

it is bound (granted physicalism) to proceed by proposing one

false theory after another.

    Second, AOE solves the problem of what it can mean to say

that theories, T0, ... TN, get successively closer and closer

to the true theory-of-everything, T, as follows.  For this we

require that TN can be "approximately derived" from T (but not

vice versa), TN-1 can be "approximately derived" from TN (but

not vice versa), and so on down to To being "approximately

derivable" from T1 (but not vice versa).

    The key notion of "approximate derivation" can be

indicated by considering a particular example, the

"approximate derivation" of Kepler's law that planets move in

ellipses around the sun (K) from Newtonian theory (NT).

    The "derivation" is done in three steps.  First, NT is

restricted to N body systems interacting by gravitation alone

within some definite volume, no two bodies being closer than

some given distance r.  Second, keeping the mass of one object

constant, we consider the paths followed by the other bodies

as their masses tend to zero.  According to NT, in the limit,

these paths are precisely those specified by K for planets.

In this way we recover the form of K from NT.  Third, we

reinterpret this "derived" version of K so that it is now

taken to apply to systems like that of our solar system.  (It

is of course this third step of reinterpretation that

introduces error: mutual gravitational attraction between

planets, and between planets and the sun, ensure that the

paths of planets, with masses greater than zero, must diverge,

however slightly, from precise Keplerian orbits.)

    Quite generally, we can say that Tr-1 is "approximately

derivable" from Tr if and only if a theory empirically

equivalent to Tr-1 can be extracted from Tr by taking finitely

many steps of the above type, involving (a) restricting the

range of application of a theory, (b) allowing some

combination of variables of a theory to tend to zero, and (c)

reinterpreting a theory so that it applies to a wider range of

phenomena.

    This solution to the problem of what progress in

theoretical physics means requires AOE to be presupposed; it

does not work if falsificationism is presupposed.  This is

because the solution requires one to assume (a) that the

universe is such that a yet-to-be-discovered, true theory of

everything, T, exists, and (b) current theoretical knowledge

can be approximately derived from T.  Both assumptions, (a)

and (b), are justified granted AOE; neither assumption is

justifiable granted falsificationism.[11]

8. Discovery of New Fundamental Theories.  Given

falsificationism, the discovery of new fundamental physical

theories that turn out, subsequently, to meet with great

empirical success, is inexplicable.  (One thinks here of

Newton's discovery of his mechanical theory and theory of

gravitation, Maxwell's discovery of classical

electromagnetism, Einstein's discovery of the special and

general theories of relativity, Bohr's discovery of "old"

quantum theory, Heisenberg's and Schr"dinger's discovery of

"new" quantum theory, Dirac's discovery of the relativistic

quantum theory of the electron and, in more recent times, the

discovery of quantum electrodynamics, the electroweak theory,

quantum chromodynamics, the standard model and string theory.)

Granted that a new theory is required to explain a range of

phenomena, there are, on the face of it, infinitely many

possibilities.  In the absence of rational guidance towards

good conjectures, it would seem to be infinitely improbable

that anyone should, in a finite time, be able to come up with

a theory that successfully predicts new phenomena.  The only

guidance that falsificationism can provide is to think up new

theories that "proceed from some simple, new, and powerful,

unifying idea", in accordance with Popper's 1963 requirement

of simplicity, but this is so vague and ambiguous as to be

almost useless.  Famously, Popper explicitly denied that a

rational method of discovery is possible at all: see Popper

(1959, 31).  But if discovery is not rational, it becomes

miraculous that good new theories are ever discovered.

Scientific progress becomes all but inexplicable.

    AOE, by contrast, provides physics with a rational, if

fallible and non-mechanical, method for the discovery of new

fundamental physical theories.  This method involves modifying

the current best level 3 blueprint so that:

(a) the new blueprint exemplifies physicalism better than its

predecessor;

(b) the new blueprint promises, when made sufficiently precise

to become a testable theory, to unify clashes between

predecessor theories;

(c) the new theory promises to exemplify the new blueprint

better than the predecessor theories exemplify the predecessor

blueprint.

    (a), (b) and (c) provide means for assessing how good an

idea for a new theory is which do not involve empirical

testing (which is brought in once the new theory has been

formulated).  The level 4 thesis of physicalism provides

continuity between the state of knowledge before the discovery

of the new theory, and the state of knowledge after this

discovery.  Modifying the current level 3 blueprint ensures

that the new theory will be incompatible with its

predecessors; it will postulate new kinds of entities, forces,

space-time structure, and will exhibit new symmetries.  In

other words, because of the hierarchical structure of AOE,

there is (across revolutions) both continuity (at level 4) and

discontinuity (at levels 2 and 3), something that is not

possible given falsificationism.  AOE provides physics with

specific non-empirical tasks to perform, specific non-

empirical problems to be solved, and non-empirical methods for

the assessment of ideas for new theories, all of which adds up

to a rational, if fallible, method of discovery.  It all stems

from recognizing that physicalism is a part of current

scientific knowledge.  The discovery of new fundamental

physical theories ceases to be inexplicable.  None of this is

possible granted falsificationism.[12]

    The fact that AOE is able to provide a rational method of

discovery, while falsificationism is not, is due to the

greater rigour of AOE (a point mentioned in 3 above).  AOE has

greater rigour because AOE acknowledges, while

falsificationism denies, metaphysical assumptions implicit in

persistent scientific preference for simple, explanatory

theories.  It is precisely the explicit acknowledgement of

these metaphysical assumptions which makes the rational method

of discovery of AOE possible.

9. Diversity of Scientific Method.  One striking feature of

natural science, often commented on, is that different

branches of the natural sciences have somewhat different

methods.  Experimental and observational methods, and methods

or principles employed in constructing and assessing theories,

vary as one moves from theoretical to phenomenological

physics, from physics to chemistry, from astronomy to biology,

from geology to ethology.  Falsificationism can hardly do

justice to this striking diversity of method within the

natural sciences.  Popper, indeed, tends to argue that there

is unity of method, not only in natural science, but across

the whole of science, including social science as well: see

Popper (1961).  AOE, by contrast, predicts diversity of method

throughout natural science, overlaid by unity of method at a

meta-methodological level.  AOE can do justice to the

diversity of methods to be found in diverse sciences, without

underlying unity and rationality being sacrificed.

    It is important to appreciate, first, that different

branches of the natural sciences are not isolated from one

another: they form an interconnected whole, from theoretical

physics to molecular biology, neurology and the study of

animal behaviour.  Different branches of natural science, even

different branches of a single science such as physics,

chemistry or biology, have, at some level of specificity,

different aims, and hence different methods.  But at some

level of generality all these branches of natural science have

a common aim, and therefore common methods: to improve

knowledge and understanding of the natural world.  All (more

or less explicitly) put AOE into practice, but because

different scientific specialities have different specific

aims, at the lower end of the hierarchy of methods different

specialities have somewhat different methods, even though some

more general methods are common to all the sciences.

Furthermore, all natural sciences apart from theoretical

physics presuppose and use results from other scientific

specialities, as when chemistry presupposes atomic theory and

quantum theory, and biology presupposes chemistry.  The

results of one science become a part of the presuppositions of

another, implicit in the aims of the other science (equivalent

to the level 3 blueprint of physics, or the level 4 thesis of

physicalism).  This further enhances unity throughout

diversity, and helps explain the need for diversity of method.

    But in order to exhibit the rationality of the diversity

of method in natural science, apparent in the evolution of

methods of a single science, and apparent as one moves, at a

given time, from one scientific speciality to another, it is

essential to adopt the meta-methodological, hierarchical

standpoint of AOE, which alone enables one to depict

methodological unity (high up in the hierarchy) throughout

methodological diversity (low down in the hierarchy).

Falsificationism, lacking this hierarchical structure, cannot

begin to do justice to this key feature of scientific method,

diversity at one level, unity at another; nor can it begin to

do justice to the rational need for this feature of scientific

method.

    There is a further, important point.  Any new conception

of science which improves our understanding of science ought

to enable us to improve scientific practice.  It would be very

odd if our ability to do science well were wholly divorced

from our understanding of what we are doing.  A test for a new

theory of scientific method ought to be, then, that it

improves scientific practice, and does not merely accurately

depict current practice.  AOE passes this test.  In providing

a framework for the articulation and scrutiny of level 3

metaphysical blueprints, as an integral part of science

itself, thus providing a rational means for the development of

new non-empirical methods, new symmetry principles, and new

theories, AOE advocates, in effect, that current practice in

theoretical physics be modified.  AOE makes explicit what is

at present only implicit.  And more generally, in depicting

scientific method in a hierarchical, meta-methodological

fashion, AOE has implications for method throughout the

natural sciences, and not just for theoretical physics.

    In case it should seem miraculous that science has made

progress without AOE being generally understood and accepted,

I should add that good science has always put something close

to AOE into practice in an implicit, somewhat covert way, and

it is this which has made progress possible.

7 Thomas Kuhn

    As I remarked in section 1 above, the main difference

between Kuhn's picture of science and Popper's is that,

whereas Kuhn stresses that, within normal science, paradigms

are dogmatically protected from refutation, from criticism,

Popper holds that theories must always be subjected to severe

attempted refutation.  AOE is even more Popperian than

Popper's falsificationism, in that AOE exposes to criticism

assumptions that falsificationism denies, and thus shields

from criticism.  One might think, therefore, that AOE would

differ even more from Kuhn's picture of science than

falsificationism does.

    It is therefore rather surprising that exactly the

opposite is the case.  In some important respects, AOE is

closer to Kuhn than to Popper.

    The picture of science that emerges from Kuhn (1962) may

be summarized like this.  There are three stages to consider. 

   First, there is a pre-scientific stage: the discipline is

split into a number of competing schools of thought which give

different answers to fundamental questions.  There is debate

about fundamental questions between the schools, but no

overall progress, and no science. 

    Second, the ideas of one such school begin to meet with

empirical success; these ideas become a "paradigm", and the

pre-scientific school becomes normal science (competing

schools withering away).  Within normal science, no attempt is

made to refute the paradigm (roughly, the basic theory of the

science); indeed, the paradigm may be accepted even though

there are well known apparent refutations.  When the paradigm

fails to predict some phenomenon, it is not the paradigm, but

the skill of the scientist, that is put to the test.  The task

of the normal scientist is to solve puzzles, rather than

problems.  The paradigm specifies what is to count as a

solution, specifies what methods are to be employed in order

to obtain the solution, and guarantees that the solution

exists: these are all characteristics of puzzles rather than

open-ended problems.  The task is gradually to extend the

range of application of the paradigm to new phenomena,

textbook successes being taken as models of how to proceed.

Methods devolve from paradigms.

    Third, the paradigm begins to accumulate serious failures

of prediction; these resist all attempts at resolution, and

some scientists lose faith in the capacity of the paradigm to

overcome these "anomalies".  A new paradigm is proposed, which

does resolve these recalcitrant anomalies, but which may not,

initially, successfully predict all that the old paradigm

predicted.  Empirical considerations do not declare that the

new paradigm is, unequivocally, better than the old.  Normal

science gives way to a period of revolutionary science.

Scientists again debate fundamentals, arguments for and

against the rival paradigms often presupposing what they seek

to establish.  Rationality breaks down.  If the revolution is

successful, the new paradigm wins out, and becomes the basis

for a new phase of normal science.  Many old scientists do not

accept the new paradigm; they die holding onto their

convictions.

    Kuhn argues that the dogmatic attitude inherent in normal

science is necessary if science is to make progress.  Applying

a paradigm to new phenomena, or to old phenomena with

increasing accuracy, is often extremely difficult.  If every

failure was interpreted as a failure of the paradigm, rather

than of the scientist, paradigms would be rejected before

their full range of successful application had been

discovered.  By refusing to reject a paradigm until the limits

of its successes have been reached, scientists put themselves

into a much better position to develop and apply a new

paradigm.  For reasons such as these, normal science, despite

being ostensibly designed to discover only the expected, is

actually uniquely effective in disclosing novelty.  Popper

(1970), in criticizing Kuhn on normal science, ignored these

arguments in support of the necessity of normal science for

scientific progress.

    AOE holds that much scientific work ought indeed to

resemble Kuhn's normal science, in part for reasons just

indicated.  But there are even more important considerations.

According to AOE, and in sharp contrast with falsificationism,

theoretical physics accepts a level 3 metaphysical blueprint,

which exercises a powerful constraint on what kind of new

theories physicists can try to develop, consider or accept.

The blueprint has a role reminiscent, in some respects, of

Kuhn's paradigm, and theoretical physics, working within the

constraints of the blueprint, its non-empirical methods set by

the blueprint, has some features of Kuhn's normal science.

    Furthermore, according to AOE, other branches of natural

science less fundamental than theoretical physics invariably

presuppose relevant parts of more fundamental branches.  Thus

chemistry presupposes relevant parts of atomic theory and

quantum theory; biology relevant parts of chemistry; astronomy

relevant parts of physics.  Such presuppositions of a science

have a role, for that science, that is analogous to the role

that the current level 3 blueprint, or the level 4 thesis of

physicalism, has for theoretical physics.  The presuppositions

act as a powerful constraint on theorizing within the science.

They set non-empirical methods for that science.  Such

presuppositions have a role, in other words, which is similar,

in important respects, to Kuhn's paradigms.  Viewed from an

AOE perspective, one can readily see how and why much of

science is Kuhnian puzzle-solving rather than Popperian

problem-solving.

    There are also, it must be emphasized, major differences

between Kuhn and AOE.  The chief difference is that, according

to AOE, science has a paradigm for paradigms  -  to put it in

Kuhnian terms.  In order to be acceptable, level 3 blueprints

must exemplify the level 4 thesis of physicalism (which in

turn must exemplify the level 5 thesis of comprehensibility

and so on, up to level 10).  This means that, as long as

physicalism continues to be accepted as the best available

level 4 thesis for science, metaphysical blueprints can be

assessed in a quasi non-empirical way, in terms of how well

they accord with physicalism.  Natural science is, according

to AOE, one sustained, gigantic chunk of normal science, with

physicalism as its paradigm.  In this respect, AOE is more

Kuhnian than Kuhn (in addition to being more Popperian than

Popper!).  

    Like falsificationism, Kuhn's picture of science is

hardly tenable.  In the first place, it does not fit

scientific practice very well.  Normal science undoubtedly

exists, as even Popper recognized; it may well be that most

scientific activity has the character of Kuhn's normal

science.  But even when a discipline seems most like normal

science, almost always there are a few scientists actively

engaged in developing alternatives to the reigning paradigm.

And on occasions, it is from the work of these few that a new

paradigm, and a new phase of normal science springs, often in

a way that is quite different from Kuhn's account.  It is not

obvious that accumulation of anomalies, resulting in a crisis

in biology, led to Darwin's theory of evolution.  Quantum

theory did not emerge, initially, from a crisis in classical

physics.  Planck's work around 1900 on black body radiation

engendered the quantum revolution.  It is true that classical

physics, applied to a so-called black body emitting

electromagnetic radiation, made a drastically incorrect

prediction, but no one, not even Planck, thought that this

posed a serious problem for classical physics.  The fallacious

prediction of classical physics was dubbed "the ultra-violet

catastrophe"; but this phrase was coined by Ehrenfest, after

the quantum revolution was under way, around 1912, as

propaganda for the new theory.  It was Einstein who first

recognized that Planck's work spelled the downfall of

classical physics; but general recognition of this only came

later, probably with Bohr's quantum theory of the atom, around

1913.  Again, Einstein's general theory of relativity emerged,

not because Newton's theory had accumulated anomalies and was

in a state of crisis, but because it contradicted special

relativity.  Einstein sought a theory of gravitation

compatible with special relativity, and it was this that led

him to general relativity.  These three revolutions, resulting

in Darwinian theory, quantum theory and general relativity,

are among the biggest and most important in the history of

science; and yet they do not fit Kuhn's pattern.

    Failure to fit scientific practice in detail does not,

however, provide decisive grounds for rejecting a normative

account of scientific method.  One can always reply that the

account specifies how science ought to proceed, not how it has

in fact proceeded.  Much more serious are the objections of

principle to Kuhn's account.  Kuhn, like Popper, provides no

account of the creation of new paradigms.  There are no non-

empirical criteria that new paradigms have to satisfy in order

to deserve consideration.  It is not clear, in other words,

what rules out grotesquely ad hoc paradigms.  Kuhn does

suggest that considerations of simplicity persist in science

through revolutions (see, for example, Kuhn, 1970, 155), but

there is no account of what simplicity is.  Kuhn is emphatic

that no sense can be made of the idea that there is progress

in knowledge across revolutions, the new paradigm being

better, closer to the truth, than the old one (see Kuhn, 1970,

chapter XIII).  But this is a disaster for Kuhn's whole view.

Why engage in normal science if the end result is the

rejection of all that has been achieved, all the progress in

knowledge of that period of normal science being sacrificed

when the science adopts a new paradigm?  Kuhn's arguments for

the progressive character of normal science, indicated above,

are all defeated.

    Perhaps the most serious objection to Kuhn's picture of

science is the obvious basic unintelligence of its

prescriptions for scientific research.  Suppose we have the

task of crossing on foot difficult terrain, containing

ravines, cliffs, rivers, swamps, thickets.  Kuhn's view,

applied to this task, would be as follows.  After debate about

which route to follow (pre-science), one particular route is

chosen and then followed with head down, no further

consideration being given to changing the route (normal

science).  Eventually, this leads to an impasse: one comes

face to face with an unclimbable cliff, or finds oneself waist

deep in a swamp, and in danger of drowning (crisis).  Finding

oneself in these dire circumstances, a new route is taken (new

paradigm), and again, with head down, this new route is

blindly followed (normal science) until, again, one finds

oneself unable to proceed, about to drown in a river, or

tumble into a ravine.

    This is clearly a stupid way to proceed.  It would be

rather more intelligent if, as one tackles immediate problems

of wading through this stream, climbing down this scree

(puzzle-solving of normal science), one looks ahead, whenever

possible, and reconsiders, in the light of the terrain that

has been crossed, what adjustments one needs to make to the

route one has opted to follow.  Exactly the same point holds

for science.  There can be division of labour.  Even if a

majority of scientists tackle the multitude of puzzles that go

to make up normal scientific research, taking the current

theory, or paradigm, for granted, there ought also to be some

scientists who are concerned to look ahead, consider more

fundamental problems, explore alternatives to the current

paradigm.  In this way new paradigms may be developed before

science plunges deep into crisis.  And just this does go on in

scientific practice, as I have already indicated in the brief

discussion of the work of Darwin and Einstein (and somewhat

less convincingly, Planck).  Another example of a new,

revolutionary theory or paradigm being proposed in the absence

of crisis is Wegener's advocacy of the movement of continents,

anticipating the plate tectonic revolution by decades.

Science is, in practice, more intelligent than Kuhn allows.

    In sharp contrast to Kuhn, AOE does not merely stress the

importance of "looking ahead", of trying to develop new

theories, new paradigms before science has plunged into

crisis; even more important, AOE provides a framework for

theoretical physics (and therefore, in a sense, for the whole

of natural science) within which ideas for fundamental new

theories may be developed and assessed.

    According to Kuhn, successful revolutions mark radical

discontinuities in the advancement of science, to the extent,

indeed, that old and new paradigms are "incommensurable" (i.e.

so different that they cannot be compared).  This Kuhnian view

is most likely to be correct when applied to revolutions in

fundamental theoretical physics, where radical discontinuity

seems most marked.  But it is precisely here that Kuhn's claim

turns out to be seriously inadequate.  All revolutions in

theoretical physics, despite their diversity in other

respects, reveal one common theme: they are all gigantic steps

in unification.  Thus Newton unifies Kepler and Galileo.

Maxwell's theory of the electromagnetic field unifies

electricity, magnetism and optics.  Darwinian theory brings a

kind of unity to the whole of biology.  Quantum theory unifies

chemistry, properties of matter, and ultimately, with the

development of quantum electrodynamics, electromagnetic

phenomena.  General relativity unifies special relativity,

gravitation and the structure of space-time.  Quantum

electroweak theory (partially) unifies the electromagnetic and

weak forces.  The so-called standard model (partially) unifies

all known phenomena apart from gravitation.  String theory, or

M theory, if successful, will unify all phenomena.  The very

phenomenon that Kuhn holds to mark discontinuity, namely

revolution, actually also reveals continuity  -  continuity of

the search for, and the successful discovery of, underlying

theoretical unity.

    This aspect of natural science, to which Kuhn fails

entirely to do justice, is especially emphasized by AOE.

According to AOE, revolutions in theoretical physics mark

discontinuity at the level of theory, at level 2, and even

discontinuity at level 3, but continuity at level 4.

Physicalism, which asserts that underlying dynamic unity

exists in nature, persists through revolutions  -  or, at

least, has persisted through all revolutions in physics since

Galileo.  In order to make rational sense of natural science,

we need to interpret the whole enterprise as seeking to turn

physicalism, the assertion of underlying dynamic unity in

nature, into a precise, unified, testable, physical "theory of

everything".  That, in a sentence, is what AOE asserts.

Physicalism, according to AOE, despite its metaphysical

(untestable) character, is the most secure item of theoretical

knowledge in science; it is the most fruitful idea that

science has come up with, at that level in the hierarchy of

assumptions.

    Because of its recognition that, despite the

discontinuity of revolutions at levels 2 and 3, there is the

continuity of the persistence of physicalism at level 4 (and

of other theses at levels higher up in the hierarchy), AOE is

able to resolve problems concerning the discovery and

assessment of paradigms which Kuhn's view is quite unable to

solve.  Both fundamental physical theories, and level 3

blueprints, can be partially ordered with respect to how well

they exemplify physicalism, entirely independent of ordinary

empirical assessment.  Assessing progress through revolution

poses no problem for AOE.  As we have seen, AOE solves the

problem of verisimilitude.  more....

Nicholas Maxwell