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.

Unhewn demonstrations

Unhewn demonstrations
Andrew Collier

Compell the Reasoner to Demonstrate with unhewn

Let the Indefinite be explored, and let every Man be
By his own Works. Let all Indefinites be thrown
into Demonstrations,
To be pounded to dust & melted in the Furnaces of

He who would do good to another must do it in
Minute Particulars:

General Good is the plea of the scoundrel, hypocrite
& flatterer.

For Art & Science cannot exist but in minutely
organized Particulars
And not in generalizing Demonstrations of the Rational Power.

William Blake, ʻJerusalemʼ1

This article is intended as a contribution to critical realist philosophy, not a criticism of it, but my
starting point is a paradox about the critical realist
corpus, and my conclusion a rather surprising practical consequence of it. Indeed the conclusion involves
incorporating into a scientific realist position some
views which are normally associated with romantic
or Green critiques of science, though there is nothing
essentially anti-scientific about them.2
The motivation of critical realist work has mainly
been the rectification of method in the human sciences.

Roy Bhaskar in particular is explicit about his desire
that critical realism shall do this underlabouring not
only for the work of science but for the work of human
emancipation. This could hardly be claimed if critical
realism limited itself to theorizing the natural sciences,
and indeed it has been in the human sciences that
critical realism has had most impact. Yet the central
and most fundamental argument of critical realism has
been an argument from the possibility and necessity of
experiment in science – and there are no experiments
in the human sciences.

The central argument to which I refer goes as
follows. In experiments, we make nature do what it
would otherwise not have done. We do so in order to
find out how nature produces the effects that it does


Radical Philosophy 81 (Jan/Feb 1997)

spontaneously, when we are not making experiments.

How can such experiments yield such knowledge,
rather than just the knowledge of what happens in the
experiments themselves? And why is it necessary to
force nature in this way, rather than just observe what
nature would do without our interference?

Bhaskarʼs answer, which forms the basis of
critical realism, is that an experiment isolates one
mechanism of nature from the others. Under normal
(non-experimental) conditions, the course of events is
co-determined by a number of mechanisms working
together. By preventing some of these mechanisms
from working, or keeping their operation constant, or
measuring and discounting their operation, an experiment isolates one mechanism, and shows what it is.

We may assume that when no experiment is going
on, the same mechanism works, but in conjunction
with others.

An experiment, in other words, abstracts from
certain mechanisms to identify others. It does not
just do so in thought, but makes the abstraction real.

It gives rise therefore to abstract laws – laws which
would predict how something behaves other things
being equal, but do not predict how anything will
behave in the real world where other things never are
equal. Other things are only equal when we artificially
make them equal – and that is just what an experiment is. Experiments give rise to what may be called
the abstract sciences, since they are each about one
set of laws which we have discovered by abstraction.

They are not about particular entities of one sort or
another. Physics is not specially about the physical
world; chemistry is equally about the physical world.

Chemistry is not about ʻchemicalsʼ in the sense that the
chemical industry produces chemicals. It is about the
chemical aspect of the whole physical world, including
living organisms, for instance.

Bhaskarʼs argument confirms and explains the
importance of experiments for the abstract sciences,
but it also shows that the laws defining those sciences
are not actualized; nothing is more fundamental to the
physical sciences than the law of inertia, which states

that bodies tend to remain at rest or in uniform motion
in a given direction. But nothing in the universe has
ever remained at rest or in uniform motion in a given
direction. This is no skin off the physicistsʼ noses
though, for although a cricket ball does not exemplify
inertia, we need the law of inertia alongside the law of
gravity and the laws governing air resistance in order
to explain its flight.

Now some natural sciences – for instance geography, meteorology, medicine – are about particular
entities. They may be called the concrete sciences.

When making the distinction between abstract and
concrete sciences, Husserl says that abstract sciences
ʻare nomological in so far as their unifying principle,
as well as their essential aim of research, is a lawʼ,
whereas in concrete sciences ʻone connects all the
truths whose content relates to one and the same
object, or to one and the same empirical genusʼ. He
tells us that ʻthe abstract or nomological sciences are
the genuine, basic sciences, from whose theoretical
stock the concrete sciences must derive all that theoretical element by which they are made sciences.ʼ3
It is only by getting the relation between the
abstract and the concrete sciences right that we will
be able to understand the position of the human sciences. The abstract sciences – for example, physics
and chemistry – are self-standing in the sense that
they can justify themselves experimentally, and so
donʼt need to rely on the concrete sciences in which
they are applied in order to vindicate their claims.

But there are no such self-standing abstract sciences
in the human world. Instead, there are abstract parts
of sciences whose whole connection with reality is at
the concrete level.

Husserlʼs reference to the ʻgenuineʼ sciences as
abstract reflects (surprisingly for him) what might
be called common sense in a positivistic culture; but
another apparently contradictory idea goes along with
this as part of positivistic common sense: that only the
concrete is real. It is quite widely taken for granted that
only the abstract sciences are real sciences, but that
only the concrete world really exists. We are then at a
loss to explain how the abstract sciences map onto the
concrete world. Critical realism denies both poles of
this contradiction. The mechanisms corresponding to
the laws of the abstract sciences are also real, hence:

Domain of

Domain of









Domain of


Note that this diagram4 is not saying that mechanisms
are somehow more real than events or experiences, as
is sometimes alleged, but just that they are also real.

But the concrete sciences are real sciences too,
or at any rate they are an essential part of the body
of knowledge outside of which the abstract sciences
would make no sense. For while the concrete sciences
can certainly draw on the knowledge yielded by the
abstract sciences, they can also do quite a few things
that the abstract sciences canʼt, and without which
things being done all science would lose its point.

In the first place, they have in their practical dealings with their objects a source of knowledge independent of the abstract sciences. Second, they can carry
out a depth analysis of concrete beings, which abstract
sciences cannot. And third, they can draw practical
conclusions from their knowledge. I shall return to
these points with reference to the human sciences.

Let me illustrate this with reference to a practical
discipline, which, while it would not normally be
described as a science, can be more or less scientifically done, and which brings out many of the strengths
and weaknesses of the concrete sciences. I refer to the
teaching of singing, particularly operatic singing.5
I have said that concrete sciences have two sources
of knowledge: what they borrow from the abstract
sciences and what they pick up from practical experience. They also have two tasks which abstract sciences
cannot undertake, namely depth analysis and practical
conclusions. They are tied to the practice in which
they are applied in a way that the abstract sciences are
not, since the latter have their own internal practice in
experiment. I have elsewhere called practical concrete
sciences ʻepistemoidsʼ. Some are quite science-like in
their rigour and explanatory power; others, like the
one chosen here, are much more problematic. But this
is a matter of degree, not of kind.

First of all, a singing teacher is or has been a singer
herself. She knows what it feels like. Here is some
knowledge ʻin a practical stateʼ, as Althusser would
say, which however she must put into words if she
is to teach it. I am not denying that some non-verbal
communication may take place (for instance, feeling
each otherʼs bodies to see what happens when some
specific vocal or respiratory instruction is carried out),
but the teaching would not get far without the medium
of language. But ʻputting it into wordsʼ is a genuine
problem in communicating practical knowledge of this
sort. It is often done by asking the pupil to imagine
him- or herself doing something that he or she could
not literally do. For instance: ʻpretend your voice is
coming out of the back of your headʼ; ʻsing from the
top of your head and make the sound come down
your noseʼ; or ʻdraw the sound inʼ. Taken literally, of

Radical Philosophy 81 (Jan/Feb 1997)


course, these instructions would be nonsense. But they
succeed in telling the singer how to make it feel, and
hence what to do.

However, some instructions sound plausible enough
to be quite easily taken as literally true, as descriptions
of what can take place in the singerʼs body, though a
physiologist could show that they are not. For instance:

ʻuse your sinusesʼ (i.e. make your sinuses contribute
to the sound by their vibrations). Now apparently the
vibration of the sinuses cannot appreciably contribute
to the sound. However, if the singer sings in such a
way that she can feel her sinuses vibrating, she will
be singing in the right way. Sometimes the description of how it feels is the opposite of what is physiologically the case. Thus a singer may be advised to
ʻmake more spaceʼ when singing high notes. But the
action which feels like making more space actually
makes less space in the singerʼs throat. This does not
vitiate the instruction as a direction in teaching, since
it is understood and the required effect produced.

Nevertheless the physiologist should not be entirely
ignored by the singing teacher. His or her findings
do make certain judgements possible about good and
bad practice, since it can be discovered physiologically
that certain singing techniques do harm to the vocal
cords, and so on. Sometimes such information could
have been discovered or guessed at by a good singing
teacher on the basis of experience, but not necessarily.

Physiology can correct singing practices in ways that
phenomenology canʼt – yet it would be absurd to think
that it could replace the phenomenological knowledge
that the singing teacher has. One could never become a
singer or a singing teacher by studying the physiology
of the vocal organs.

Similar considerations apply in psychoanalysis,
whose raw data are entirely phenomenological, but
some of whose abstract parts sound like speculations
about neurology and sexology. Those biological sciences do not vitiate the practice of psychoanalysis
when they appear to conflict with it, yet it would be
obscurantist for psychoanalysis to ignore their findings.

Freudʼs metapsychology has a status much like the
metaphorical physiology used by singing teachers. At
certain points – for instance, the concept of instinct or
drive which Freud explicitly says is a border-concept
between the biological and the psychological sciences
– a tie up with physiology is useful and helps to confirm
or refute the metapsychological theory. At others it
would be inappropriate to take the psychoanalytical
idea physiologically and look for confirmation or
refutation from physiology, since the concept may
be justified phenomenologically, by its clarification
of the patientʼs self-experience. For instance, Freudʼs
distinction between clitoral and vaginal orgasms is


Radical Philosophy 81 (Jan/Feb 1997)

neither contradicted by claims that all female orgasms
are clitorally triggered, nor confirmed by claims about
the G-spot. They are based on the self-experience of
some women whom Freud analysed, and make no
physiological claims.

These points may I think be generalized to the
concrete sciences: their concepts are arrived at partly
by retroduction from practical experience – ʻhow can
we explain the way things seem in practiceʼ – and
partly borrowed from the abstract sciences. By these
two methods, the concrete sciences build up a stock
of abstract concepts of their own. But here we may
note a difference between concrete natural sciences
like meteorology and concrete human sciences like
psychoanalysis. The borrowings are in both cases
from natural sciences since only in these are there
experiments. This means that the human sciences
can learn far less by borrowing than can the natural
concrete sciences. This is what justifies my claim that
the human sciences relate to their objects only through
their concrete parts, even though there is some input
from experimental sciences into human sciences and
disciplines (e.g. from physiology into psychoanalysis
or singing teaching). That input is alien in a way that
the input from physics to meteorology is not.

Now let us look at what one of the only three people
with a credible claim to have founded a human science
says about the relation of abstractness to concreteness.

I refer to Marx (the other two are Freud and Chomsky).

Marx sometimes stresses the importance of abstraction
in science, while at other times he seems to be quite
rude about it. Most of his followers follow either one
or the other of these examples, but it may be that the
two can be reconciled. First, it should be mentioned
that for Marx as for Kant, the term ʻabstractionʼ often
refers to a process rather than a result: abstracting from
something, bracketing it off. Now to one rude remark
about abstraction:

First of all, an abstraction is made from a fact;
then it is declared that the fact is based on the
abstraction. That is how to proceed if you want
to appear German, profound and speculative.

For example: Fact: The cat eats the mouse.

Reflection: Cat = nature, Mouse = nature;
consumption of mouse by cat = self-consumption
of nature.

Philosophical presentation of the fact: the devouring of the mouse by the cat is based upon the
self-consumption of nature.6

Here the point about abstraction is that it gives you
poorer, less specific information than more concrete
language. Wherever abstraction means no more than
leaving something out in order to arrive at a more
general and less specific description, Marx is rude
about it.

But in the section on the method of political
economy in the 1857 introduction, he says that to
start with ʻthe real and concreteʼ, e.g. the population,
commits the same error, since ʻThe population is an
abstraction if I leave out, for example, the classes
of which it is composedʼ.7 We should therefore start
by moving analytically towards ʻever thinner abstractionsʼ, and then put them together again in their due
order to arrive at ʻthe population again, but this time
not as the chaotic conception of a whole, but as a rich
totality of many determinations and relationsʼ.8 We
arrive at this full conception of a concrete entity by
showing the relations between its features which taken
separately would be abstractions, but of which the real
concrete entity is composed: ʻThe concrete is concrete
because it is the union of many determinations, hence
unity of the diverse.ʼ9
Good abstraction consists in specifying the many
interrelated aspects of something, and this is how
a concrete science should proceed; bad abstraction
consists in ignoring the specificity of something, to
subsume it under some more general concept. Thus
a social scientist who wanted to provide knowledge
of Britain in the 1990s should analyse out the many
aspects of that society and show how they are related;
and not abstract from many of its specific features in
order to place it in a statistical population of somewhat
similar societies, and produce statistical data about
them. If social science were an experimental science,
it would be possible to actualize these abstractions
and test them separately – for example, to test what
the effect of exposure of British capital to Continental
competition would be in the absence of trade unions.

But it is not. Hence ʻin the analysis of economic
forms neither microscopes nor chemical reagents are
of assistance. The power of abstraction must replace
both.ʼ10 We make abstraction in thought – ʻexperiments in thoughtʼ – and cannot do more than this.

But we stay with the concrete reality which we are
analysing and try to tease out as many details as we
can, rather than ignoring details in order to subsume
it under generalizations. The abstract parts of good
social science are speculative explanations of concrete
particulars, which are tested only by their capacity
to explain those concrete particulars. To take them
out of their use in such explanation and use them to
compare different concrete particulars in the attempt
to arrive at ʻstatistical causalityʼ is always a mistake,
since the other determinants of events and differences
are being ignored without their having been rendered
ineffective as in a real experiment. Statistics has an
important place among the descriptive preliminaries
of social science; information about the simultaneous
increase in unemployment and crime may be the starting point of a fruitful social-scientific analysis. But the
analysis itself must focus on concrete particulars, and

retroduce explanations from them. It cannot read off
the explanations from the statistical correlations. But
if statistics has no explanatory role in social science,
neither has any other form of mathematics, since we
cannot measure abstract forces without actualizing
them as is done in an experiment.

To return, then, to the original paradox: the analysis
of experiment has not been useless for the human
sciences, since it shows something very general about
the real world, namely that it is structured and stratified, that the concrete really is a union of many
determinations, and hence that abstraction and analysis
are appropriate methods of developing knowledge of
concrete beings. But it also shows that where experiment is not possible, this analysis and abstraction is
not measurable, and is testable only by its capacity
to explain the minute particulars of concrete entities. It shows, in short, the ontological similarity of
natural and human sciences – they are both analysing
concrete structured wholes and explaining them in
terms of the abstractions arrived at – and also the
methodological dissimilarity of natural and human
sciences: the latter cannot use mathematics and should
look rather to paradigms like Freudʼs analysis of the
Rat Man (or any full psychoanalytic case-history), or
Trotskyʼs History of the Russian Revolution. In short,
social science can exist only ʻas minutely organized
particulars, and not in generalizing demonstrations of
the rational powerʼ.

Now to the question of practical applications. Many
of the concrete sciences are inseparably tied to a practice (for example, medicine to healing, psychoanalysis
to therapy, Marxian theory to working-class politics),
but whether this is so or not, all sciences, abstract
or concrete, deliver their discoveries into a world of
ongoing practices which they often transform in some
way. I want to argue in conclusion that as a matter of
scientific ethics and good public policy, abstract sciences ought never to be allowed to influence practice
directly, but only through the assistance they give to
concrete sciences. This applies not only to natural sciences, where it has important ecological implications,
but also with respect to the abstract and concrete parts
of human sciences.

Abstract sciences can yield practical advice of a
sort; for instance, ʻhere is a way w to make product
x with half the labour that it took before.ʼ This may
justify the conclusion: ʻother things being equal, we
should make x by process w.ʼ But other things are not
equal and in this sort of case that is crucial. Process
w may also deplete a scarce resource or cause pollution or produce sickness in the workers. To establish
whether this is so, we need a concrete study of the
process in context – environmental, human, economic
and so on. That study may yield real practical advice,
without the ʻother things being equalʼ clause. However,

Radical Philosophy 81 (Jan/Feb 1997)


it is a feature of our economic system that the ʻother
things being equalʼ clauses of the practical advice of
abstract sciences are not taken seriously – the advice
is carried out as if it were conclusive, which only the
advice of a concrete scientific study can even approximate to being. This is a feature of modern economic
life because if the ʻother thingsʼ do not relate to
profitability, they are equal to the commercial appliers
of science – all equally indifferent. But it is important
to notice that all advice from abstract science is subject
to an other-things-being-equal clause, just because it
is abstract, having abstracted from some of the other
things that affect the outcome. An economic system
that systematically applies science in an abstract state
is a systematically irresponsible system.

There is a tradition of hostile criticism of science
which makes very similar points to those that I have
been making. Blake in some moods is part of that
tradition. Today it often comes from those within
the Green movement who blame science rather than
its commercial use. There is also a tendency within
ʻpostmodernismʼ to counterpose the practical concrete
knowledge of (for instance) shepherds to the pretensions of the Ministry of Agriculture experts, trained
in abstract science. In so far as these tendencies have
directed attention to the necessity of concrete knowledge and its sole right to guide practice, they are
absolutely right. Yet the fault lies not with abstract
science but with the tendency of its commercial and
military users to apply science in its abstract state,
rather than treating abstract sciences as contributory
disciplines whose results must flow together into the
sea of concrete science before they are in a fit state
to be applied practically.

Since I am claiming that the rationality of the
practical application of science depends on following
the full sequence abstract sciences – concrete science
– practice rather than the abridged sequence abstract
science – practice, I should perhaps say something
about what sort of discipline counts as a concrete
science, as distinct from an abstract science on the
one hand and a practice on the other. I have given
examples of concrete sciences such as meteorology and
medicine, but it is not always easy to identify them.

Whereas abstract sciences are individuated (as physics,
chemistry, biology and the like) by the kinds of laws
that they discover, and practices are individuated (as
agriculture, health care, war and so on) by the aims
that they pursue and the means that they use, concrete
sciences are individuated by their concrete objects,
which may be of greater or lesser scope, so that the
concrete science required for a given bit of application
of science to practice may not have a ready-made
name. To take the instance of pesticides, the abstract
science of chemistry is applied in the practice of
agriculture. The concrete scientific disciplines which


Radical Philosophy 81 (Jan/Feb 1997)

ought to stand between these two would include in
their objects the state of the soil, water and air, and
various ecosystems dependent on that state. It is a
matter of ʻpractical wisdomʼ to determine in any given
case what the scope of the requisite concrete science
should be.

The application of abstract science is irrational
and the application of concrete science is rational.

But concrete science can never give you what abstract
science appears to give: a method of calculating which
course of action is rational. For the discovery that w
is the most economical way of producing x can be
quantified: it costs half what process v had cost. But as
soon as a multiplicity of incommensurable reasons for
using or not using w is delivered by the concrete study
of w-in-context, we have to make a decision between
incommensurable alternatives. Calculation becomes
useless. But that is not an objection to basing action
on concrete science, for calculation between the actual
alternatives was always impossible anyway. It was only
because many of the incommensurable values were left
out of account or assigned arbitrary commensurable
values that the issue looked calculable in the first
place. Deciding on the basis of an abstract science
just is leaving some things out of account – that is
what abstract means. If we have to decide between a
motorway and an ancient wood of special scientific
interest, there is nothing we can rationally do but sit
down and think ʻwould life be better with a wood and
no motorway or with a motorway and no wood?ʼ At
which point I can perhaps conclude as I started with
a quote from Blake:

Improvement makes strait roads; but the crooked
roads without Improvement are the roads of Genius.11

1. William Blake, Complete Writings, Oxford University
Press, London, 1966, p. 687.

2. An earlier version of this paper was read to the Seminar
on Critical Realism organized in the Faculty of Economics and Politics at the University of Cambridge.

3. Edmund Husserl, Logical Investigations, Vol. 1,
Routledge & Kegan Paul, London, 1970, pp. 230–31.

4. Roy Bhaskar, A Realist Theory of Science, Harvester
Wheatsheaf, Hemel Hempstead, 1978, p. 13.

5. I am grateful to Heather Collier for information in this
part of the article.

6. Karl Marx and Frederick Engels, The German Ideology,
edited by Pascal, International Publishers, New York, pp.


7. Karl Marx, Grundrisse, Penguin, Harmondsworth, 1973,
p. 100.

8. Ibid.

9. Ibid., p. 101.

10. Karl Marx, Capital, Vol. 1, Penguin, Harmondsworth,
1976, p. 90.

11. Complete Writings, p. 152.

Download the PDFBuy the latest issue