The following text has been automatically reproduced by an Optical Character Recognition (OCR) algorithm. It may not have been checked over by human eyes. For matters of precision please consult the original pdf.

Is Bhaskar’s Realism Realistic?

Is Bhaskar’s Realism

A/an Chalmers
In this paper I begin by briefly outlining what I consider to be
the key features of Roy Bhaskar’s realist account of laws of nature. l I regard his account to be the best available. However, I
do not think his case can be established as conclusively as is
suggested in much of his writing. In my second section I consider the status of Bhaskar’s argument, jettison an unnecessary
component and stress the provisional character of what
remains. In the third and longest section I criticise Bhaskar’s
realism for its limited scope. I indicate how it might be usefully
augmented by drawing on some recent work by Nancy
Cartwright and !an Hacking. I end by indicating why I consider
the limited focus of Bhaskar’s realism has serious repercussions for the programme he has undertaken in his published
books, and, in particular, for the extension of his realism into
the social sciences.

1. Bhaskar’s realism
Bhaskar bases his realist account of laws of nature on two very
general features of physics and its practice. One of these concerns experimentation. When a theory is put to the test of experiment’ the regularities that constitute a significant result, the
mode of deflection of an electron beam in a magnetic field, the
negative result of the Michelson-Morley experiment and so on,
do not, in general, occur of their own accord. They occur only
under the special conditions of a controlled experiment, conditions that have to be striven for in experimental activity. The
second feature of the practice of physics referred to by Bhaskar
is the fact that the laws supported in experimental activity are
presumed to apply, and are applied, outside of experimental
situations. Theories of electromagnetism are employed in the
construction of radios and are assumed to apply both when the
radio functions and fails to function. Further, the employment
of, for example, the laws of mechanics and electromagnetism
in cosmology or geophysics presupposes that those laws are, or
were, applicable to the world prior to there being any humans,
let alone any physics or experimental activity.

These general observations lead to the conclusion that a
wide class of accounts of laws of nature, namely, those that
construe them as regular or constant conjunctions of events
along something like Humean lines, are inadequate. According
to those accounts, laws are to be interpreted as regularities of
the kind ‘event of type A is invariably accompanied or followed by event of type B’. IT this characterisation of laws is
taken literally then laws of physics do not qualify. Autumn
leaves rarely, if ever, fall to the ground in accordance with the
law of fall. On the other hand, if the characterisation is
qualified by asserting that the constant conjunction only occurs


if some condition C is satisfied, then the application of a law is
restricted to those, mainly experimental, situations where condition C is indeed satisfied and the constant conjunction obtains. It leaves the question ‘What governs the world outside of
experimental situations?’ completely unanswered and leaves
us, for example, with no explanation whatsoever of why an
autumn leaf usually ends up on the ground.

Bhaskar accommodates experiment and the application of
science outside of experimental situations by transcending the
empiricist ontology implicit in the constant conjunctions account and invoking the generative mechanisms that lie behind
and give rise to the flux of events. Generative mechanisms are
depicted as the kinds of things that there are in nature and their
mode of behaviour, the latter being best understood as tendencies. Fundamental laws of nature, according to this realist view,
describe how the entities comprising generative mechanisms
tend to behave. They describe, for example, the inertial and
gravitational tendencies of masses and the tend~ncy of charged
bodies and electro-magnetic fields to behave in accordance
with the Maxwell/Lorentz theory. Such tendencies do not in
general lead to regularities at the level of events because they
will typically be juxtaposed in complicated ways with other
mechanisms. A falling leaf will be subject to inertial, gravitational, hydrodynamic, thermal and other mechanisms. It is the
gravitational tendency that is responsible for the leaf falling to
the ground when it does, but to identify that tendency and to
specify the law governing it, it is necessary to practically intervene to remove the perturbing influence of other mechanisms.

It is in this kind of way that Bhaskar’s realist account of laws
renders both experimental activity and the application of laws
outside of experimental situations intelligible.

Bhaskar’s realism certainly constitutes a decisive refutation
of much orthodox philosophy of science. As we have seen, it
shows the constant conjunction view of laws to be untenable. It
poses problems for the hypothetico-deductive model of explanation. Events can rarely be explained by deducing them
from laws together with initial conditions. At best, this can be
done only for ideal experimental situations. Bhaskar’s account
serves to identify the shortcomings of those anti-realist
positions that construe the ‘surplus element’ in fundamental
laws, over and above mere constant conjunctions, as some
model or idea supplied by us and serving to render phenomena
intelligible to us. This is incompatible with the application of
laws, in cosmology and geophysics for example, to the world
as it existed prior to science and even before there were
humans for science to be intelligible to. Bhaskar’s position is
realist in the strong sense that the laws of nature that we attempt to identify and fonnulate in science are attempted

descriptions of real mechanisms assumed to be in action in the
world irrespective of our knowledge of them.

2. The status of Bhaskar’s argument for his realism
With reservations that will be made clear later, I regard Bhaskar’s realist account of laws of nature to be the best account
available. Nevertheless, I do not think his position has been established as securely as some of his own presentations suggest
In this section I spell out what I consider to be the limitations
of Bhaskar’s argument and the corresponding vulnerability of
the realist theory that he, confidently, and I, tentatively and
reservedly, embrace.

Let me first dispose of an argument which Bhaskar uses to
bolster his case which is fallacious and which he does not need.

I will refer to it as Bhaskar’s ontological argument It runs as
follows. In an experiment an experimenter is the causal agent
of a sequence of events but not of the law that that sequence
enables her to identify. Consequently there is an ontological
distinction between laws and sequences of events. This argument has plausiblity only if we interpret some of Bhaskar’s
remarks in an undesirably anti-realist way. In what sense does
an experimenter ’cause’ the sequence of events generated in a
controlled experiment? She causes them in the sense that she
assembles the appropriate experimental arrangement. But what
happens when she has done so is dependent on the way the
world is. The generative mechanisms at work cause the sequence of events, not the experimenter. The ontological argument works only if the experimenter is taken to be the cause of
the sequence of events, as opposed to the experimental set-up,
an idealist assumption quite out of keeping with Bhaskar’s

Let us return to the sounder aspects of Bhaskar’s case and
assess its status. Bhaskar asks the question ‘What must the
world be like for science to be possible?’ and presents his
realist theory as the answer. More specifically, he notes that
scientific theories are tested in closed experimental situations
and applied outside of those situations, shows that this is incompatible with a wide range of traditional conceptions of
science, and offers a realist account of science that renders
those aspects of science perfectly intelligible. It does not follow
from this that the world must be as described by Bhaskar’s
theory.2 The status of his argument is weaker than that. The
possibility of rival accounts of science equally able to render
experimental activity etc. intelligible cannot be ruled out This
by no means trivialises Bhaskar’s position. As a matter of fact I
know of no rival theory of science that can accommodate certain features of experiment as it occurs in science as well as
Bhaskar’s, which is why I currently accept it Nevertheless,
from this logical point of view, Bhaskar’s position has not been
established with the rigour that some of his remarks suggest it

A further dimension to the inconclusiveness of Bhaskar’s
position can be introduced in the following reflection. Suppose
a medieval scientist, working within some version of Aristotelian theory, had asked Bhaskar’s question ‘what must the
world be like for science to be possible?’ and had attempted to
answer it in a Bhaskarian way. I suggest a most plausible part
of his answer would have been ‘the world must be a finite,
harmonious whole with a centre’. The whole of Aristotelian
science, with its distinctions between forced and natural
motions, terrestrial and celestial regions and so on, presupposed the absolute space provided by such a cosmos. This is
the conclusion that a medieval version of Roy Bhaskar would
have arrived at We now know, of course, that that conclusion

was fallacious. In retrospect we can see that Aristotelian
physics did presuppose a cosmos. We can also understand why
that approach would have limited success in spite of the fact
that it involved false assertions and has been replaced by a
more successful theory and practice. It is not difficult to imagine our current science suffering a similar fate. The practices
that Bhaskar correctly identifies as important components of
contemporary science may be transcended by more successful
practices. In that case a future historian might acknowledge
that Bhaskar had correctly identified a world view presupposed
by the science of his day but one that proved to be of limited
applicability and inadequately geared to coming to grips with
the world as it really is.

The problem for Bhaskar that this example brings out has
been indicated by Ted Benton in a review of Bhaskar’s The
Possibility of Naturalism. 3 Bhaskar asks what the world must
be like for science to be possible, but it is clear that mere possibility is much too weak a notion for his purpose. After all,
there is a sense in which astrology is possible. It is widely practised. But I do not think that Bhaskar would be impressed by
support for a realist theory based on the compatibility of that
theory with the practices of astrology. The most obvious way to
strengthen Bhaskar’s position is to replace ‘possible’ by something like ‘highly successful’. There is surely some sense in
which modem science is highly successful however difficult it
might be to characterise that notion of success precisely. Consequently, a realist view that can make sense of basic features
of modem science certainly has something strongly in its

favour. But once we raise the question ‘successful compared to
what?’ the inconclusiveness of Bhaskar’s argument becomes
evident. The possibility that modem science might be replaced
by something more successful cannot be ruled out, and were
this to come about then his form of argument would require
that his realist theory be revised. It may be the case that some
features of modem quantum mechanics are already incompatible with Bhaskar’s realism, although I remain to be convinced. This discussion brings out the extent to which Bhaskar’s theory lacks normative force. If science should productively develop in a way that clashes with Bhaskar’s realism,
then so much the worse for Bhaskar’s realism.

I think a useful comparison can be made between the status
of Bhaskar’s position, as I have construed it, and Popper’s position on methodological rules.4 Popper is, I think correctly, antipathetic to metaphysical principles such as the principle of the
uniformity of nature, which some see as being presupposed by
science, on the grounds that there is no adequate way of es19

tablishing or arguing for such principles. His strategy is to
translate these metaphysical principles into methodological
rules. In place of the principle of the uniformity of nature Popper offers a methodological rule that enjoins scientists to seek
exceptionless, universal laws of nature. He recognises that following such rules is not guaranteed to lead to success, although
we might well be encouraged by their past success. Popper
himself does not adequately discuss the conditions for the acceptance of rejection of methodological rules, but he might
well have argued for them by showing them to be implicit in
the practice of modem science. Of course, the substantive account of science, offered by Popper and his methodological
rules, and by Bhaskar and his transcendental realism differ
markedly, and I hope I have left no doubt which I prefer.

However, they are similar to the extent that they are both
predicated on the practice of successful physical science and
are both subject to revision should that practice change in fundamental ways. Consequently, while they both may have some
normative force for normal science, they do not for extra-ordinary science (to use Kuhn’s distinction).

3. Limitations of Bhaskar’s realism
Two interesting and challenging books on realism in science
that have appeared recently are Nancy Cartwright’s How the
Laws of Physics Lie and lan Hacking’s Representing and Intervening. In this section I exploit that material to develop what I
see as some limitations of Bhaskar’s analysis of physical

Cartwright’s anti-realism has nothing in common with empiricist anti-realism which stems from a reluctance to attribute
reality to unobservable ‘theoretical entities’. Cartwright is
thoroughly realist with respect to causes and is content to include entities such as electrons and electromagnetic fields
amongst the causes of physical phenomena. The anti-realist
aspects of her position are based on certain aspects of the practice of physics which, on her construal, are incompatible with a
realist interpretation of fundamental laws of nature.

Cartwright distinguishes between fundamental laws of nature and what she calls phenomenological laws. Cartwright’s
usage of the term ‘phenomenological’ corresponds to that of
physicists rnther than philosophers. Phenomenological laws
lend themselves to fairly direct rest and application, are usually
mathematically formulated but offer or assume no explanation
of or mechanism underlying the mathematical formalism. Examples offered by Cartwright are Airy’s mathematical formulation of Faraday’s magneto-optical effect, which specifies but
does not explain it, the performance of lasers as specified by
their manufacturers and the phenomenology of fundamental
particle interactions, including such things as scattering crosssections. Fundamental laws are general formulations which, by
contrast, do offer or figure in, fundamental explanations of the
phenomena of the world in general and of phenomenological
laws in particular. Newton’s laws, Maxwell’s equations and the
Schrodinger equation are amongst Cartwright’s examples of
fundamental laws.

Some of the reasons why Cartwright regards it as inappropriate to interpret fundamental laws realistically are as follows. She highlights the extensive use of models in physics.

Metals are represented quantum mechanically as a sea of
electrons in a periodic potential, laser materials are represented
as a collection of two-level atoms and the solar system, in
Newtonian theory, is represented as a small number of
regularly shaped bodies. None of these models portray real
world situations accurately. Fundamental laws are applied to


models of reality rather than to reality. A related point made by
Cartwright is that the formulations resulting from the application of fundamental laws to models have to be corrected before
they are practically applicable to real experimental or technological situations. Thus the parameters in the theoretically
derived specification of the performance of a small-signal
amplifier have to be adjusted in the light of empirical
measurements before a realistic description is arrived at. Using
a somewhat technical example from the theory of lasers
Cartwright claims that physicists, whilst they will not tolerate
conflicting causal accounts of one and the same situation, are
content to exploit a number of incompatible theoretical
treatments involving different models. She draws on such examples to argue for a somewhat instrumentalist, Duhemian
construal of fundamental laws, as opposed to her realistic
treatment of causes.

Cartwright’s observations about the practice of physics pose
no real problems for Bhaskar’s realist interpretation of fundamental laws of nature. Using Bhaskar’s terminology, we can
say that Cartwright is led to her anti-realist conclusions because she looks to science for a realistic description at the level
of events. Once our ontology is extended in the way that Bhaskar argues to include powers, tendencies and generative
mechanisms, Cartwright’s case loses its force. I have shown
elsewhere how Bhaskar can be defended in the face of
Cartwright’s arguments.6 I will not repeat the details here.

Rather, I will deal more sympathetically with Cartwright’s
book and try to develop her observations in a way that exposes
limitations in Bhaskar’s position.

Cartwright’s observations about the prnctice of physics are
not inconsistent with Bhaskar’s realism with respect to fundamental laws of nature. I do not retract that claim of my
earlier article. However, to argue that a position is consistent
with the prnctice of physics is not in itself a strong commendation for that position as a useful account of physics, a point
brought home to me by Wal Suchting.7 After all, for all I know,
liberation theology might be consistent with the practice of
physics but it does not tell us anything useful about it. Of
course, Bhaskar’s realism fares better than liberation theology
in this regard. As we have seen, his realism is compatible with
some general features of modem science in a way that many
rival accounts of science are not, and this constitutes a strong
argument for it. However, there are other important aspects of
science about which Bhaskar’s realism has not much to say. I
wish to draw on Cartwright’s discussion, and later on Hacking’s, to illustrate limitations of Bhaskar’s position in this

Let us explore further the extensive use of models in
physics highlighted by Cartwright. The need for idealising
models is not difficult to understand. Whilst from the point of
view of Bhaskar’s realism it can be maintained that fundamental laws of nature apply to the real world in an unqualified way,
it can also be recognised that the behaviour of real world
situations will not in general be fully explicable by or derivable

from those fundamental laws. There are two reasons for this.

Firstly, real situations are typically too complex for a direct application of fundamental laws to be possible. The motions of a
real liquid, the excitation and decay of a molecule, even the
real motions of the planets in the solar system, are too complex
to be precisely characterised by fundamental laws. Secondly,
the behaviour of real world situations is usually determined by
a range of generative mechanisms interacting in complex and
often unknown ways.

It is, of course, precisely because of the above difficulties
that experimentation, as opposed to mere observation, is necessary in physical science. One can attempt to minimise the
problems I have mentioned by constructing simple arrangements and shielding them from unwanted perturbations, as,
for example, Cavendish did when, in his famous experiment,
he shielded his attracting spheres from draughts, vibrations and
magnetic effects. But I think it is a mistake to believe that all
the problems referred to above can be removed in this kind of
way, even in principle. Experiments involve real metals, real
electrodes, real gasses and so on. The precise structure of such
items is typically unknown, and in any case is not amendable to
treatment by direct application of fundamental laws.

The history of solid state physics provides a useful example
to illustrate my point. From the 1920s onwards solid state
physics emerged as a major branch of physics. The main aim
was to explain the properties of solids in quantum-mechanical
terms. A major component of that effort involved the devising
of appropriate models of metals, crystals and so on, which, on
the one hand, were amenable to theoretical treatment and, on
the other, were able to replicate some of the properties of real
metals, crystals etc. The early treatment of metals, for example,
involved a lattice model and a free electron model, with the latter eventually proving more productive. In 1946, N. F. Mott,
one of the most distinguished solid state physicists, summed up
his approach to the subject thus:

room for an element of anti-realism when we come to consider
the relation between models and the real world. Whilst, as
Ernst McMullin has nicely illustrated,9 models can be systematically ‘de-idealised’, so that they come closer to an accurate description of reality, we never, or very rarely, reach a
stage where the gap between description of model and true
description of reality is closed. Let us consider, for example,
the Wiedemann-Franz law, which relates the thermal and
electrical conductivity of metals, a candidate for what
Cartwright calls a phenomenologicallaw. Bhaskar, illustrating
his realist stance with respect to scientific knowledge, says of
the law that, in a world without science it ‘would continue to
hold although there would be no-one to formulate, experimentally establish or deduce it:lO By contrast, I doubt that any real
metal conforms or ever did conform precisely to the structure
of the model for which, given the truth of fundamental laws,
the Wiedemann-Franz law holds. Contrary to Cartwright,
whilst fundamental laws may be true of reality, it is the
phenomenologicallaws that lie.

A further aspect of Cartwright’s book that can be exploited
to highlight the limits of Bhaskar’s realism is her focus on the
practical, technological aspects of science, the ways in which
scientists and engineers come to grips with the behaviour of
real liquids, real amplifiers, real lasers and so on. This is also
the focus of the second half of Hacking’s book. In what immediately follows I draw heavily on the latter source.

tl ‘.”.

The trouble is that for even the simplest metals the
problem of a large number of electrons interacting
together is so forbiddingly complicated that at present
one can solve it only by making a number of simplifying assumptions, and it is seldom certain that these are
valid. What we can do is to take a simple model of a
phenomenon and to work out its consequences, and to
see how much of the observed facts can be fitted in to
the resulting framework. 8
Roughly speaking, then, in this major branch of physics the
fundamental laws of nature are taken for granted. The main effort is directed towards the construction of models of real
world situations whose behaviour is assumed to be governed
by the fundamental laws, and which can replicate, more or less,
some of the behaviour of real solids. This picture corresponds
quite well with Cartwright’s simulacrum account of science.

Bhaskar says little about such matters and his realism needs to
be augmented in a major way to accommodate them.

Cartwright draws anti-realist conclusions from her
simulacrum account. I do think there are some weak anti-realist
conclusions to be drawn here but they are in direct opposition
to Cartwright’s. Cartwright maintains that the use of models in
science is one reason why a realist interpretation of fundamental laws is inappropriate. I maintain that, once we move from
what Cartwright embraces under the name of actualism, which,
in Bhaskar’s terminology, limits us to an ontology of events, to
a richer ontology involving powers, tendencies and generative
mechanisms, then realism with respect to fundamental laws can
be defended in the face of Cartwright’s case. However, there is

The role of experiment as it is construed by mainstream
philosophy of science is to test theories. This view is clearly
conveyed in the following quotation from Popper’s The Logic

of Scientific Discovery.

The theoretician puts certain definite questions to the
experimenter and the latter, by his experiments, tries to
elicit a decisive answer to their questions, and to no
others. All other questions he tries hard to exclude ….

But it is a mistake to suppose that the experimenter
proceeds in this way ‘in order to lighten the task of the
theoretician’ , or perhaps, in order to furnish the
theoretician with a basis for inductive generalisations.


On the contrary, the theoretician must long before have
done his work, or at least what is the most important
part of his work; he must have formulated questions as
sharply as possible. Thus it is he who shows the experimenter the way…. Theory dominates the experimental work from its initial planning up to the finishing
touches in the laboratory. I I
It is undoubtedly the case that these words describe a vital role
of experiment in science, and it is this role that is well accommodated and skilfully exploited by Bhaskar. But there are other
dimensions to experimentation within science that Hacking
stresses which, whilst not inconsistent with Bhaskar’s realism,
illustrate limitations of it, and perhaps suggest that it is unduly
Much experiment in science involves what Bacon referred
to as ‘twisting the lions tail’, practically intervening with nature
to see how it will behave under unusual circumstances, often in
response to practically or technologically posed problems and
not guided by an explicit theory. Into this category come, for
example, many properties of light discovered and exploited
prior to their explanation by the wave theory, double refraction,
and the phenomenon of polarisation, together with many of the
laws governing that phenomenon discovered by Brewster.

Brewster would have none of the wave theory of light As
Hacking puts it, ‘Brewster was not testing or comparin~
theories at all. He was trying to find out how light behaves.’ I
In a similar vein one could mention the conduction properties
of metals, well known before their quantum mechanical explanation, or knowledge of emission spectra prior to their explanation. One of the most telling examples of practically
efficacious, through not theoretical or fundamental, scientific
knowledge is the series of practical inventions culminating in
the steam engine and the Industrial Revolution. This practical
knowledge owed little or nothing to knowledge of fundamental
laws of nature. As Hacking puts it, experiment sometimes has a
life of its own.

Now, of course, there is a straightforward response that
mainstream philosophers, and Roy Bhaskar, can make to these
observations. Phenomenological laws, practical inventions and
the like can be accepted as some low level, lesser kind of,
scientific knowledge, but it can be insisted that they reach the
level of genuine scientific knowledge only when they have
been explained by a deeper theory involving fundamental laws
of nature. Such a view could be solidly backed by pointing to
the increased power and range of practical, phenomenological
knowledge made possible by the explanatory theory. The im-


proved design and greater range of engines made possible once
the steam engine was understood in terms of the laws of thermodynamics would provide an obvious example to support
such a case.

I do not wish to dispute that there is much that is correct
about that response. However, undue emphasis on the theoretical mode leads to a failure to capture the character of and the
vital role played in science by practically efficacious
knowledge at the experimental, technological level. Whilst it is
true that fundamental theory often shows the way towards practically efficacious knowledge, as with the discovery of radio
waves, for example, it is also the case that that order is frequently reversed, as the examples drawn from Hacking illustrate. What is more, knowledge at the practical,
phenomenological level constitutes an important part of the
preconditions for the emergence of fundamental theory.

Theories do not come out of the air, or straight out of the minds
of geniuses, as Bhaskar’s discussion of the social production of
knowledge makes clear. Whether it is more appropriate, at
some particular conjuncture within a science, to pursue a
theoretical quest for fundamental laws or to employ a more
practically orientated strategy will depend on the contingencies
of the situation. ‘Seek fundamental laws characterising the
generative mechanisms of nature’ would have been poor advice
for the pioneers of the industrial revolution. There was a period
of a quarter of a century (1880-1905) in which
phenomenological thermodynamics was more productive than
the kinetic theory. A search for underlying mechanisms to explain the behaviour of gasses was eventually productive. Corresponding efforts to discover mechanical properties of an
aether to explain electro-magnetic fields proved to be futile.

The view of science emphasised in Bhaskar’s realism, a search
for fundamental laws characterising the generative mechanisms
of nature, whilst sometimes appropriate, often misses the mark.

What is more, if his portrayal of science were ased as a basis
for advising scientists it would often yield bad advice.

Another aspect of Hacking’s analysis of science is worthy
of mention before I pursue this line of criticism a little further.

Hacking, by focussing his attention on what is done in experimental science is able to develop a robust brand of realism
that differs from Bhaskar’s. He argues for the reality of
electrons and the entities observed through high-powered
microscopes, for example, in a low-level way that focusses on
the practical procedures involved. He argues, convincingly, that
experiments that involve the spraying of electrons in a controlled way to produce causal effects in targets presuppose the existence of electrons whilst he argues, equally convincingly, that
strong evidence for the reality of microscopic structures is
provided when a number of microscopes utilising quite different physical principles reveal identical structures. Hacking
wishes to confine his realism to a realism with respect to
theoretical entities, defending his position with reference to experimental activity using examples like the ones I have mentioned. He is reluctant to interpret fundamental laws realistically, and signals his agreement with Cartwright in that regard.

I can find nothing in Hacking’s book that provides a convincing
reason to abandon Bhaskar’s realist interpretation of fundamental laws of nature. However, I suggest that the addition to
Bhaskar’s realism of some of Hacking’s contributions would
enrich the former and give it a healthy, down to earth, practical
dimensions which it might be said to lack.

Perhaps this account of the limitations of Bhaskar’s realism
could be summed up thus. Whilst Bhaskar has given a satisfactory realist account of fundamental laws of nature, there is
much more to science than establishing such laws.

4. Implications for Bhaskar’s programme
My case for there being serious omissions in Bhaskar’s realist
account of science does not challenge his realist rendering of
laws of nature. My considerations concerning models, drawing
on Cartwright, and those concerning experimentation and
theoretical entities, drawing on Hacking, all point to ways in
which Bhaskar’s theory can be usefully augmented, rather than
to ways in which it is incorrect This suggests a way of construing my case in a way that is charitable to Bhaskar. He has given
us a realist account of fundamental laws of nature that is superior to others and free from telling objections. If his account
can be usefully augmented by adding to it characterisations of
other aspects of science drawn from Cartwright, Hacking or
whomever, all well and good. One person cannot be expected
to do everything.

Given what I understand to be the nature of Bhaskar’s
project, I am not sure that such a charitable reading of my critique is appropriate. Bhaskar uses his realist account of science
as a starting point for proceeding to an analysis of the human
sciences and thence to general considerations concerning the
nature of ideology and possible routes to human emancipation.

If my assessment of the partial character of his analysis of
physical science is correct, and he does emphasise the theoretical as opposed to more practically orientated modes, then there
is the danger that this emphasis will persist through the
remainder of his programme. Indeed, I believe that it does. I
believe that his passage into the social and political domain
takes place at a level of theory that his arguments are unable to
sustain and removes him too far from the domain of practically
efficacious intervention.

My criticism of Bhaskar’s realism can be put with a slightly
different emphasis which will help point to a further difficulty
with his progression to the social sciences. Bhaskar takes certain features of scientific practice as the starting point for his
‘transcendental deduction’ of a realist theory of natural
science. 13 But his choice of features is selective. His view of
realism does justice to those features of science that he selects,
but not to other, equally important, features of contemporary
science. The general problem raised here is the all-important
one of the criteria to be employed in deciding what is to be included in the premises of the argument.

Whilst in the physical sciences, there may be a problem
concerning which aspects of science to focus on, at least there
is an uncontentiously successful science to turn to. When we
turn to the social domain, there is no uncontentious social
science whose practices we can draw on for premises for an argument for realism. This difficulty is acknowledged by Bhaskar. 14 His response is to attempt to isolate ‘more or less universally recognised features of substantive social life itself, which
do not beg the issue at the outset in favour of one type of social
science rather than another’ .15 The problem here goes beyond
the serious one of how to select amongst possible candidates
for such ‘universally recognised features’. The very fact that
some features of social life appear uncontentious and are
universally accepted is strong evidence that they are merely
pre-scientific reflections of the ideology of the day. Bhaskar
himself makes the point, in another context, when talking of
the birth of a science.

Probably the most significant type of event in the history
of any science is that in which it comes to define – or
rather redefine – its object of enquiry… Typically this
process will necessitate some scientists breaking free …

of the ’tissue of tenacious truisms’ currently congealed
in their field. 16

It would seem that Bhaskar recognises the danger that his account of realism in the social sciences will turn out to be based
on a tissue of tenacious truisms. In actual fact, the features of
humans and society that he does select as the basis for his
theory are such ‘as to make it possible to pull a version of historical materialism out of the philosophical hat’ as an antipodean group of reviewers unkindly putit. 17 Whilst I am sympathetic to the general drift of many of the conclusions Bhaskar
arrives at in this way, I believe that his attempt to establish
them by philosophical argument is flawed. I do not see how
philosophy can be seen as having the resources to arrive at
such substantive conclusions. In my view, philosophy can play
an important critical role, exposing false philosophical and
ideological intrusions into science and social and political life
and I regard Bhaskar’s realist account of physical science as
providing a useful weapon in that struggle. However, when it
comes to a positive role for philosophy, I regard Bhaskar’s
project as misconceived, and believe we must recognise the
priority of substantive science and politics over philosophy.

When looking for positive guidance from philosophy we must
rest content with some vague generalisations about the need to
be specific.

This is the text of a paper read at a conference on ‘Realism in the
Human Sciences’ at the University of Sussex in September 1987.

Roy Bhaskar, A Realist Theory 0/ Science, Sussex, Harvester,

Ope cit., p. 27.

Roy Bhaskar, The Possibility 0/ Naturalism, Sussex, Harvester,
1979, reviewed by Ted Benton, ‘Realism and .social theory’,
Radical Philosophy 27, 1981, pp. 13-21.


K. R. Popper, The Logic 0/ Scientific Discovery, London,
Hutchinson, 1968, p. 253.

A Realist Theory 0/ Science, pp. 217-18, fails to grasp this aspect
of Popper’s position.

AIan Chalmers, ‘Bhaskar, Cartwright and Realism in Physics’,
Methodology and Science 20, 1987, pp. 77-96.

W. Suchting, ‘Realism in physics: Comments on a paper by AIan
Chalmers’, forthcoming in Methodology and Science.

N. F. Mott, ‘Atomic physics and the strength of materials’,
Journal 0/ the Institute 0/ Metals, 72, 1946, p. 371. I owe this
source to S. T. Leith and Paul K. Hoch, ‘Formation of a research
school: Theoretical physics at Bristol 1930-54’, British Journal
for the History o/Science, 19, 1986, pp. 19-44.

E. Mullin, ‘Galilean idealization’, Studies in History and
Philosophy o/Science, 16, 1985, pp. 2A7-73.

lOA Realist Theory 0/ Science, p. 22.

11 The Logic o/Scientific Discovery, p. 107.

12 I. Hacking, Representing and Intervening, Cambridge,
Cambridge University Press, 1983, p. 157.

13 Of course, the considerations I raise in section 2 indicate that the
term ‘deduction’ is much too strong here.

14 The Possibility o/Naturalism, p.17.

15 Ibid., p. 18.

16 Roy Bhaskar, Scientific Realism and Human Emancipation, New
Left Books, London, 1986, p. 104.

17 R. AIbury, G. Payne and W. Suchting, ‘Naturalism and the
Human Sciences’, &onomy and Society, 10, 1981, pp. 367-79.


Download the PDFBuy the latest issue