. 2
( 7)


O1 Not-O2
Therefore T Therefore Not-T or Not-A

left commits the formal fallacy of af¬rming the consequent and is invalid.
Which is to say that the premises may both be true and the conclusion
false. The argument on the right is valid (which was Popper™s point), but
it shows that T need not be rejected “ on the evidence. More generally,
the idea, so critical to the empiricist understanding of science, that it
was possible to develop a logic of con¬rmation which had the power of
mathematical logic, seems now to have been abandoned.8
Plainly theories need to be judged by their ¬t with observed and exper-
imental data “ even while we acknowledge the always present problem
of assessing the signi¬cance of evidence and the ever present possibil-
ity of a fudge factor “ but a third overlooked factor relevant to theory
acceptance is the explanatory role of the theory. That is, it is a critical
feature of good theory that the representation be a convincing descrip-
tion of real, but perhaps unobservable processes which would explain
not merely observations that are readily available, but observations made
available by controlled experiment “ where this is possible. As Hempel
ultimately came to see, versus instrumentalist (anti-realist) conceptions
of theoretical terms, theories function to provide explanations “ or better
understanding in the sense that chapter 1 argued. But in addition to this,
as Michael Friedman and others have insisted, a fundamental criterion for
building a consensus in the scienti¬c community is explanatory uni¬cation.
The kinetic theory of gases, for example, gives us an understanding of
8 Some of the early key papers in this now largely forgotten debate may be found in Manicas,
1977: section VI, “Induction.”
Theory, experiment and the metaphysics of Laplace 33

the pattern given in the Boyle-Charles law, but other phenomena, for
example, that gases obey Graham™s law of diffusion, are made intelligible
as well.
Even if the foregoing is granted, it is very often said that the real test of
theory is its predictive capacity. There is a sense in which this is true, but
another in which it is not. It is true as regards the typical experiment, but
despite much mythology to the contrary, predictions are not, in general,
a reliable test of theory. To clarify this, we need ¬rst to introduce the
concept of closure. This discussion leads to a sketch of the implications
of the fact that in the real world, all the countless generative mechanisms
are operating open-systemically. The upshot is radical contingency and,
with it, critical limits on our ability to make predictions.

Experiment and the concept of closure
Following Harr´ (1970), and more recently, Bhaskar (1975), the key fea-
ture of experiments is that the experimenter actively intervenes in the
course of nature. She makes things happen which otherwise would not
have happened. Putting aside what might be called “exploratory exper-
iments,” such as anatomical dissection, the aim of an experiment is to
isolate or make constant all those properties except those one wants to
study. Put roughly, the experimenter has a theory about some generative
mechanism / causal process which, once initiated, has a predictable (in
theory) outcome. Her aim, accordingly, is to trigger the mechanism, but
to preclude anything which would have an effect on the outcome so pre-
dicted. The idea is to show that since no other potential causes are at work
in the experimental situation, only the one being manipulated accounts
for the outcome. This is an extraordinarily dif¬cult thing to bring off and
what needs to be done varies with the problem being addressed: from the
attempt to test the mean-speed theorem by having a ball roll down an
inclined plane so that it was easy to measure distances and times accu-
rately, to Michaelson and Morley™s idea to ¬‚oat their apparatus on a bath
of mercury in order to isolate it from vibrations and other disturbances
which would have affected what they expected to occur, to the hurling of
nuclei of one element against the nuclei of another expecting that they
form the nucleus of as yet unknown elements. To use other language here,
the experimenter seeks closure.
We have closure when (1) we are able to identify all the pertinent
initial conditions; (2) we can either isolate the generative mechanisms that
theory says are implicated in the outcome, or serially keep them constant;
(3) we can be assured that there is constancy of extrinsic conditions. In
this situation, the system is not only deterministic, which entails that
34 A Realist Philosophy of Social Science

whatever happens is caused, but that contingency has altogether been
eliminated. The intrinsic structures (the generative mechanisms) of the system
ensure that for each set of antecedent conditions only one result is possible.
Experiment presupposes our ability arti¬cially to establish closure or to
take advantage of partial closures where we can ¬nd them (Conley 2001).
A successful experiment is a highly potent test of a theory exactly because
if conditions of closure are even approximately satis¬ed, the predictions
of the theory are tested. It is in this sense that it is said that a good theory
can predict. As before, logically, if the experiment fails to produce the
predicted outcome, it may be because some assumptions of the theory are
false, or because some other factor has entered, unnoticed, into producing
the actual outcome.
This situation is very different from predicting outcomes in open sys-
tems. In nature, there are no closed systems “ even if, as in astronomy
one has, without experiment, what amounts to closure. Indeed, it is this
fact which makes plausible the idea that theory can be formalized mathe-
matically and that outcomes, accordingly, will be explained as simply the
product of mathematical calculations.
This is easily illustrated. In analyzing the dynamics of the solar system,
we assume, not unreasonably, that we have identi¬ed all the masses, that
all the relevant causal mechanisms governing the movement of all the
masses which have been identi¬ed are known (there are really only two),
and that no large masses, not already identi¬ed, will subsequently become
part of the system. Our dynamical description is, as it were, unchanging.
Indeed, linear equations effectively model the system. The problem of
identifying the location of any body anytime then becomes strictly com-
putational. Thus, we can predict the exact location of the moon at any
instant into the in¬nite future. But to do this we assume, not unreason-
ably, that no huge mass will come ¬‚ying into our solar system. If it were
to do so, all our predictions would fail. The system, closed to that point,
would be opened. All our calculations would be wrong.
Alternatively, when we think of classical physics, perhaps we think
of a projectile whose path is beautifully described by the formula for
a parabola, y = ax2 + bx + y. But we don™t think of a falling leaf or
a boulder crashing down a mountain, splintering and leaving its parts
strewn down the side of the mountain. Yet these phenomena are also, in
principle, describable by the same physics. We can predict the positions
of planets and projectiles with considerable exactitude; we cannot do this
with leaves and boulders. Why not?
The falling leaf is still subject to the laws of motion, but it might go
anywhere exactly because we cannot specify the initial conditions and
there are all kinds of things in the system “ the erratic air mass through
Theory, experiment and the metaphysics of Laplace 35

which it falls, a bicycle rider speeding by “ which will affect its downward
trajectory. The system remains open. (We can construct an experiment,
however: we can create a vacuum in a closed chamber, and so on.)
Getting ahead of ourselves, we can here contrast the behavior of clock-
work soldiers and real people whose behavior is manifestly open-systemic.
As Bhaskar says:

Clockwork soldiers and robots do not more nearly observe the laws of mechanics
than real people. Rather, their peculiarity stems from the fact that if wound up
and left alone their intrinsic structure ensures that for each set of antecedent
conditions only one result is possible. But outside the domain of closure the laws
of mechanics are, as Anscombe has put it, ˜rather like the rules of chess; the play
is seldom determined, though nobody breaks the rules™. (1975: 110)

In our world, most events “ birth, growth, rain, ¬res, earthquakes, depres-
sions, revolutions “ are produced by a complex nexus of causes of many
different kinds, conjunctively at work. Thus, as Bhaskar notes, “the pred-
icates ˜natural,™ ˜social,™ ˜human,™ ˜physical,™ ˜chemical,™ ˜aerodynamical,™
˜biological,™ ˜economic,™ etc. ought not to be regarded as differentiating
different kinds of events, but as differentiating different kinds of mecha-
nisms” (1975: 119). Indeed, a good measure of the extraordinary success
of the disciplines of the abstract physical sciences is due to the fact that
inquirers have been able to ignore concrete complexity and, via abstrac-
tion from the real concrete, they have been able to theorize physical,
chemical and bio-chemical mechanisms as if they were operating with-
out interference. This involves a reductionist strategy in the sense that
as Atkins puts it: “they prefer to disentangle the awesome complexity of
the world, examining it piece by piece, and build it up again, with deeper
understanding” (2003: 2). Here not only is experiment critical but the
capacity to deal with the real concrete in terms of strata “ the physical,
chemical and biological “ has been a critical feature of the successes of
the physical sciences.
To go back to our earlier example, we can think of the periodic table
as abstractly summarizing the chemical possibilities for all the elements,
what causal properties they have, what molecules are possible (and impos-
sible) and what causal properties they must have qua chemical, even
when they are functioning as they normally are, in open systems. Simi-
larly, one can understand mechanical outcomes in terms of the generative
mechanisms of physics, and so for biology, which provides us with theo-
ries of biological mechanisms. But since in the world, they are operating
open-systemically, this knowledge, powerful as it is, is not suf¬cient to
either explain or predict any concrete outcome “ even the dissolving of a
particular spoonful of salt in the water glass in my hand. Pertinently, if
36 A Realist Philosophy of Social Science

lacking in interest, the salt has to get into the water and the condition of the
water must be appropriate.
The foregoing has enormous implication for a human science, to be
considered in the chapters that follow. Here we can only notice the con-
sequences regarding experiment in the human sciences, for not only is
there no way to seek even relative closures, but intervening to make things
happen which would not have happened otherwise will likely be immoral.
But there is one piece of un¬nished business.

Explanation and prediction are not symmetrical
Of considerable pertinence to the problem of understanding and expla-
nation in the human sciences is the idea that explanation and prediction
are symmetrical. This idea must be heartily rejected. As already noted,
one often encounters the idea that a good theory makes good predictions.
But where this idea is appropriate, it does not mean that some naturally
occurring event is thereby predicted. Rather, it concerns the powerful
idea, important to accepting a theory as true, that on the basis of the the-
ory, we are able to test our theory and sometimes make new discoveries.
There are many instances of this. An easy one to describe is the ¬lling in
of the periodic table. We noted that lacking any knowledge of the struc-
ture of atoms and their dynamics, Mendeleev had to compile his table
empirically. Still, by interpolating between known properties of neigh-
boring elements in his table, he was able to ¬ll in some of the gaps in that
table. (Atkins notes that he also “predicted elements that do not in fact
exist.”) But the powerful quantum theory developed by Neils Bohr and
Erwin Schrodinger allowed chemists to infer the existence of elements
which experiment then proved to exist. Mendeleev began with 61 known
elements. We now know that there are some 110 elements.
But this is very different to arguing that we can judge a theory by
its ability to predict events in open systems. Thus, as Milton Friedman
argued: “theory is to be judged by the predictive power for the class of
phenomena which it is intended to ˜explain™” (1968: 512). As he says:

the relevant question to ask about the ˜assumptions™ of a theory is not whether they
are descriptively ˜realistic™, for they never are, but whether they are suf¬ciently
good approximations for the purpose at hand. And this question can be answered
only by seeing whether the theory works, which means whether or not it yields
suf¬ciently accurate predictions. (Friedman, 1968: 517)

Again, if prediction means that given our knowledge of chemical mecha-
nisms, there should be an unknown element between two already identi-
¬ed elements, then proof of its existence is a powerful test of the theory.
Theory, experiment and the metaphysics of Laplace 37

But if it means that theory will allow us to predict any and all chemical
outcomes in the world, then while theory gives us an understanding of
powerful constraints on what can happen, there are nonetheless limitless
possibilities regarding what will happen. The salt in my hand may never
dissolve, or be a party to the rusting of the can, to the seasoning of my
steak, and so on.
There is indeed a paradox here: we don™t need current theory to make
very good predictions regarding many chemical outcomes. We noted ear-
lier that to explain that a particular quantity of salt dissolves in water, we
need to understand that salt is water-soluble (but we need not understand
the mechanism which explains this) and (non-trivially) we need to know
that it was put in water. That is, we often can and do offer conditional
predictions: if X, then Y will occur. These are, indeed, the bread and
butter of ordinary life, and as long as we are speaking of the countless
generalizations available to us “ all known independently of the discov-
eries of science “ we are generally not disappointed. But not only is this
hardly the prediction and control so often taken to be the test of scienti¬c
theory, but as noted, we are also very often disappointed, either because
in an open-systemic world, the conditions of the antecedent were not
satis¬ed, or they did not constitute a set of suf¬cient conditions. Were it
otherwise, of course, we would all get rich on the stock market and there
would be no divorces.

The world is not Laplacean
Pierre Simon Laplace (1749“1827) was a brilliant mathematician who left
us with the powerful idea that a theory of n-variables with n-equations
would make all science computational. Indeed, we can think of the uni-
verse as one gigantic closed system. Remarkably, it is just this assumption
which lingers in the background of what has been called “regularity deter-
minism,” the idea that “the world is so constituted that there are descrip-
tions such that for every event the simple formula, ˜whenever this, then
that™ applies” (Bhaskar 1975: 69). And of course, it is just this assump-
tion which is promoted by the empiricist “search for laws” as the goal of
explanatory science.
But if what happens in the universe is the product of the particular
conjunction of initiated generative mechanisms, and the con¬gurations
of these changes with time, there will be no such description “ and there
will be contingency and plenty of it. This means that after something has
happened, we are often able to explain it “ it was caused, but we could not
have predicted it “ sometimes without even a modest measure of probabil-
ity. This is typical of many of the events which interest us most: a war; the
38 A Realist Philosophy of Social Science

fall of the Berlin Wall; a powerful upswing in the economy; an extended
drought; an earthquake; a hotel ¬re; a fatal stroke; the emergence of a
new virus.
The assumption of regularity determinism encourages two counter-
productive regressions. If the predicted outcome fails to occur, one
searches for “the missing variable,” either to continue to enlarge the sys-
tem to include new “factors,” or to reduce elements to their presumed
atomistic components. There is no rational limit to how far one might go.
In the ¬rst, perhaps more typical case, the system continues to include
variables until it includes everything. In the second case, since there are no
conditions intrinsic to the system, the reduction proceeds until it includes
nothing. The aspirations of Wilfredo Pareto, economist-cum-sociologist,
illustrate this beautifully. For him:
In order thoroughly to grasp the form of a society in every detail, it would be
necessary ¬rst to know what all the very numerous elements are, and then to know
how they function “ and that in quantitative terms . . . The number of equations
would have to be equal to the number of unknowns and would determine them
exclusively. (1935: vol. 4, paragraph 2072)
For Pareto, working before the age of the supercomputer, the problem
was wholly the practical dif¬culty of solving the linear equations.9 Thus,
the economic system was but “a small fraction of the social system,” but
even then, “in the case of 100 individuals and 700 goods there would
be 70,699 conditions . . . We would have to solve a system of 70,699
Perhaps unsurprisingly, Pareto™s vision was not ignored by social sci-
entists. For example, Talcott Parsons insisted that “not only do theo-
retical propositions stand in logical interrelations to each other so that
they may be said to constitute ˜systems™ but it is in the nature of the
case that theoretical systems should attempt to become ˜logically closed.™
That is, a system starts with a group of interrelated propositions which
involve reference to empirical observations within the logical framework
of the propositions in question.” And indeed, “the simplest way to see the
9 We can employ linear equations when the variables represent a stable, closed dynamical
system. General equilibrium theory is a perfect example; see chapter 6. Non-linear equa-
tions are used to model “chaotic systems.” Edward Lorenz (1996) developed three non-
linear equations for the analysis of weather. He programmed these and ran the sequence.
On another run, he stopped the sequence mid-point, but rather than go back to the start,
he typed the mid-point values into the computer and ran the sequence from there. The
two sequences diverged, at ¬rst by a small amount, then increasingly. The computer
stored six digits, but the printout only three. When he began the sequence from the mid-
point, there was a very small difference in the input values of the variable and these were
ampli¬ed as the sequence ran. See the discussion of meteorology, below.
Theory, experiment and the metaphysics of Laplace 39

meaning of the concept of a closed system in this sense is to consider the
example of a system of simultaneous equations. Such a system is deter-
minate, i.e., closed, when there are as many independent equations as
there are independent variables” (1937: 9“10). Parsons™s commitment
to positivist principles is clear enough here. But, as noted, the world is
not Laplacean and celestial mechanics is a poor model for science.
It is clear why we can often explain when we could not have predicted:
time makes the difference. Since the universe is not a closed system,
what happens has consequences regarding what will happen next. These
ideas are best illustrated, perhaps, by considering two historical sciences:
evolutionary biology and meteorology.
Darwin gives us a notion of a science radically unlike the ideal
bequeathed (mistakenly) by classical physics. The key difference is this:
Darwin showed us that, at least with respect to living things, history
matters a great deal (Manicas, 1989c; Rosenberg, 2005). Roughly, he
offered us a way to understand the taxonomic order of living things in
terms of “historical pathway, pure and simple.” That is, species and their
characteristics are explained as adaptations, where, importantly, adap-
tation is a response to the historical sequence of selective demands of
the environment (including other organisms). Darwin thus showed that
there was absolutely no requirement for us to impute some form of design
or intrinsic purpose or meaning to what exists nor, as importantly, that
there was any sort of necessity or inevitability about which species have
perished and which have come to exist. It is important to be clear about
Darwin did not explain the evolution of the species. He provided one
powerful mechanism for explaining this: natural selection. Explaining the
outcomes of natural selection presupposes that we have detailed informa-
tion regarding organisms and the relations of organisms to their environ-
ment. If we had this information from the beginnings of life, we would
have a start in reconstructing the course of evolution. Unfortunately, such
information is not and will not become available. As Richard Burian
writes, “the course of evolution, even on a fairly large scale, is fraught
with the consequences of historical accidents and contingencies” (1989:
160). This does not mean that biological phenomena are either wholly
or partly uncaused. It means rather that, as with any concrete event, the
evolution of a species, like the onslaught of a drought, is the outcome of
a multiplicity of causes in a continually changing con¬guration. To call
something a historical accident is but to say, as above, that it could not
have been predicted, that there was an “incalculability” as regards what
in fact happened. To speak of contingencies is to say only that there is
40 A Realist Philosophy of Social Science

no reason to believe that the world is like the solar system as described
by classical physics, a world where all the masses and their relations are
accounted for and nothing new will happen.
Meteorology, like evolutionary biology, is a historical science and, like
geology, it draws on non-geological laws pertaining to the mechanical
and thermodynamical properties of gases, solids and liquids. Its prob-
lems with predictions are well known, but we can now see clearly why.
Weather is a wonderful example of a chaotic system. Such systems are
“mixing in ¬nite time.” That is, “what is initially known about the sys-
tem becomes probabilistically irrelevant to its future” (Hobbs, 1993:
124). This depends upon noticing that the system is sensitive to ini-
tial conditions, which means that as a function of the accuracy of our
knowledge of these conditions, even under conditions of relative closure,
there will be a range of degrees of freedom as regards the subsequent
This is best illustrated with the example of successive tennis balls hit
into a forest. Two successive balls, hit at nearly identical velocities, can
hit a tree at nearly identical locations. But each time they are de¬‚ected,
their trajectory changes. The very small initial difference results in a dif-
ference in all the subsequent hits. Accordingly, the two balls may end up
in two very different locations. Indeed, as Max Weber long ago pointed
out using a boulder rolling down a rough hill as his example, even if we
assume “ideal conditions of antecedent observation,” while we could cal-
culate “the occurrence and perhaps general angle of the splintering,” we
could not calculate “the number or shape of fragments, the patterns they
formed when they come to rest or a veritable in¬nity of other aspects”
(1975: 122). Even putting aside the lay of the mountain, at each instance
in the downward trajectory, the splintering is itself altering the conditions
of future falling and splintering. Because what is happening has conse-
quences on what then happens, there is, he rightly insisted, an eradicable
“incalculability.” Mathematically, this is a non-linear system. For such
systems, there is in principle unpredictability. Jesse Hobbs applies the
idea to weather.

For example, meteorologists use parameters such as temperature, humidity, pres-
sure, wind direction, and wind velocity to make predictions. This yields systems
with ¬ve or six degrees of freedom multiplied by the number of distinct locations
for which these values are measured or represented “ a level of computational
complexity that already demands the largest supercomputers to manage. Even
so, the resulting “Accu-weather” predictions might as well be called “Unaccu-
weather” predictions because of meteorological chaos . . . But suppose undaunted
meteorologists take the plunge into ever greater levels of precision. Should they
Theory, experiment and the metaphysics of Laplace 41

otherwise succeed, they will run up against what Edward Lorenz calls the “but-
ter¬‚y effect” “ the unanticipated ¬‚apping of a butter¬‚y™s wings in a chaotic air
mass would perturb it enough to throw off all long-range weather predictions
(1993: 124“125).10
While we need not pursue the point here, putting aside the unanticipated
¬‚apping of the wings of the butter¬‚y and the computational problems that
greater precision would bring, Hobbs doubts that taking “the plunge” to
greater levels of precision will improve our predictive abilities. The limit
of precision, as with all chaotic systems, is literally in¬nite. Hence, while
we have determinism “ outcomes are causal products “ there is also in-
principle unpredictability.
These considerations entail that we cannot say that an event had to
happen. To be sure, once something happens, we can always go back in
time, identify the relevant generative mechanisms and causal contingen-
cies and provide an account which explains the event. This will generally
take the form of a narrative which identi¬es the particular collocation of
causes as they developed in time. As Harr´ says:
Temporal concepts allow us to order the in¬‚uences as causes and effects. Each
cause is an in¬‚uence exerted on some mechanism from without, and so itself
produced by some other mechanism; that is, is itself an effect. The stimulus or
stimuli which brought it into being are causes, and to come into existence in a
world of enduring mechanisms must themselves be effects. Effects become causes
of further effects, and causes are the effects of antecedent causes. (1970: 262)
All our explanations of events will be incomplete. The effort to ¬nd the
causes will cease when we have satis¬ed the demand that called for the
In the chapters that follow, we can draw on the account of the foregoing
to examine the problem of explanation and understanding in the human
sciences, and hopefully to generate some useful strategies for responding
to their distinct tasks.
10 The weather forecaster is but minimally interested in explanation, but the meteorologist
is in a position to provide a good understanding of meteorological phenomena, both
before and after the fact. That is, like the physicist, she can offer an account of the
critical generative mechanisms at work in producing meteorological phenomena, for
example, the thermodynamical properties of ocean cooling.
11 This temporal regress of causal explanation has a parallel as regards understanding where
we have a regress of micro-explanation, from “salt is water-soluble,” to an account in
terms of molecules, to atoms to quarks. See Harr´ 1970: chapter 10.
3 Explanation and understanding in
the social sciences

It is often supposed that because the social sciences must deal with people,
social science is either quite impossible, or at best, inevitably incompetent.
Unfortunately, this view is promoted, in quite unintended ways, by many
writers who have a mistaken view of the natural sciences. These writers
suppose that:
1. If science is to be empirical, it must be experimental.
2. The main task of science is prediction.
3. The successful sciences can both explain and predict events (including,
then, the acts of individuals).
4. Nature is uniform in the sense that scienti¬c laws are regularities of
the form, “whenever this, then that.”
5. Theories are “deductive systems.”
6. Scienti¬c observation is theory-neutral.
If we measure the social sciences on any of these grounds, they look
very bad “ even hopeless. But things are not as bad as they seem, since
none of the foregoing propositions is true. In the previous chapters, we
tried to show why. The alternative offered shows that:
1. There are very successful non-experimental sciences.
2. A main task of any science is description and understanding; prediction
plays a minor role.
3. Explaining concrete events is generally neither the interest, nor often
within the competence, of a science.
4. Nature is uniform not in the sense that there are “invariant relations
of resemblance and succession” (regularity determinism), but in the
sense that things have causal powers that allow us to generalize and
have expectations.
5. Theories are almost never deductive systems; rather, they offer a rep-
resentation of causal mechanisms and processes, both observable and

Explanation and understanding in the social sciences 43

6. Finally, it is quite impossible to “observe” anything independently of
some conceptual frame of reference, but this does not undermine the
quest for a true representation of reality.
We will put some of these ideas to work in the present chapter. We
shall not argue, however, that there are no important differences between
inquiry in the human sciences and inquiry in the natural sciences. Unfor-
tunately, these differences are not, in general, properly understood. These
misunderstandings are, usually, part of the more general misunderstand-
ing about science generally.
There are two very large differences to be considered. The ¬rst is the
obvious one that studying people is not the same as studying “things.”
The second follows on this. Unlike the objects of study in natural science,
the objects of study in social science “ institutions, social structures, social
relations “ do not exist independently of us. They are, as we shall explain,
real but concept- and activity-dependent. We begin with an account of

Explaining human powers
Persons are organisms, but they are also social beings. We need to see what
this means, and we need to be careful here. Both in ordinary conversation
and in social science, we tend to speak not of persons, but of individuals.
We do this because by “individuals” we tend to mean persons “ individuals
with a host of capacities which they employ in interaction. These are,
in the jargon, “socialized” individuals. Understanding these capacities
is a necessary ¬rst step. We can then draw what are some important
conclusions bearing on explaining the actions of persons.
The view of causality that we have sketched helps enormously in clar-
ifying what is at issue and we think also to dispel an illusory problem:
the bearing of biology on human action. We can begin with the fact that
(excepting for identical twins) no two human genotypes are the same
and that from the moment of conception the developmental process is
epigenetic. That is, everything that happens is a complex transactional
interplay of causes and processes through time.
It may be useful here to have a workable de¬nition of “genotype,” “phe-
notype” and “epigenesis.” The genotype is the “internally coded, inher-
itable information” carried by all living organisms. The phenotype is the
observable physical features of an organism and includes anything that is
part of the observable structure, function or behavior of a living organism.
Another way to speak of epigenesis is to say that the phenotype is the non-
additive causal product of gene“gene transactions, gene“environment
transactions and environment“environment transactions. As regards
44 A Realist Philosophy of Social Science

phenotypical outcomes, including even most genetic disorders, nearly
all are epigenetic causal products, a point of considerable importance.
Each of us begins (at conception) as but two cells “ the genome which
establishes the genotype. It is often said, wrongly, that the “genetic code”
is a “program,” as if whatever we come to be was fully “determined”
by the string of DNA. But as the biologist Paul Weiss says, in the ¬rst
place, the genome “is, and always has been a captive of an ordered envi-
ronment,” and “while the genome contributes to the speci¬c properties
of that environment in mutual interactions with it . . . it is only by virtue
of the primordial frame of organization of the cytoplasm of an egg that an
individual can maintain . . . the unity of overall design.”1 For the overall
design, for example, Felix catus or Homo sapiens, not any DNA chain will
do, of course, but an immensely complex “ and contingent “ interplay
of causal processes determines the speci¬c character of the organism.
That is, DNA contains all the information necessary to build and sus-
tain an organism, but it needs a living organism in an environment. And
the building and sustaining of it involves a marvelously complex causal
From the point of view of biology, an organism is an ordered complex
of orderly complex systems. Biochemistry starts from the level of atoms
and molecules and works upward through the larger and more complex
molecules to complicated systems, organelles, cells, tissues, organs, sys-
tems and ¬nally to the organism itself. Activities within systems may have,
as the outcome of their causal transactions, properties at higher levels.
These are properly termed emergent properties. For example, proteins
are capable of at least eight major activities of which the amino acids from
which they are polymerized are not capable. Complete information about
all the atomic positions of an unknown protein does not allow us to infer
even that the protein is an enzyme, still less, what in a speci¬c system
its particular causal properties or functions might be. What it does is a
consequence of its relations in the system. This holds true at every level,
including the psychological.
Moreover, higher level properties have bearings on lower level func-
tions and properties. The coordinated movements of an organism are
paradigmatic. The cat reaches for the ball of string. In achieving his goal,
fantastic constraints are imposed in coordinating the array of systems,
perceptual, muscular, anatomical and so on, which are involved.
The organism is not a closed system. That is, the effects of micro-
processes at the molecular level are mediated not only at that level but

1 The following owes much to the various writings of Paul A. Weiss (1968, 1971, 1972).
See also Hull, 1974; Wimsatt, 1976a, 1976b; Craver, 2001.
Explanation and understanding in the social sciences 45

by mediations in a wider environment, an environment which, strictly
speaking, extends to the far reaches of the universe. For example, expo-
sure to radiation, alcohol, drugs, poor nutrition and so on is devastating to
the organism™s course of development. As Weiss writes, “in this incessant
interplay, the latitude for epigenetic vagaries of the component elements
on all levels . . . is immense.”
The epigenetic “vagaries,” of course, are not unlimited: they are, if you
will, restricted by our “biologically determined” human nature. It may be
useful here to give a restricted meaning to a term used widely but vaguely
(and usually wrongly). We can say that some trait, capacity or difference
is a feature of our (biologically determined) “human nature” only if, in
realizing that trait, the developing conceptus undergoes a “characteristic”
human development such that it is substantially irrelevant where and
when that process takes place. To be sure, there is no characteristic human
development since development is consistent with a fantastic range of very
different environments. Nevertheless, the idea is clear enough. We want
here to rule out (at least for the moment), such genetic and environmental
accidents as Down™s syndrome and thalidomide babies. Most crucially,
while we must acknowledge that humans need a human environment to
realize their distinctive human capacities, we want to put aside (for the
moment) the social and cultural differences encountered in all human
Given this restricted sense, there are some obvious biologically deter-
mined traits: our human anatomy and physiology is one. This makes
some capacities possible and others impossible. Humans cannot ¬‚y and,
lacking gills, they cannot breathe in water. Biology determines sex and
manifest physical traits that mark family resemblance, such as facial fea-
tures, body type and skin color. But race is not biologically determined
since on all the evidence there are no biological grounds for grouping
people into distinct races.2 In other terms, there is no non-arbitrary sta-
tistically signi¬cant difference between populations which we would like
to call “races” and neighboring populations. Indeed,“each population
is a microcosm that recapitulates the entire human macrocosm, even if
the precise genetic composition varies slightly” (Cavelli-Sforza, 2000: 25,
29). Gould concludes with a wonderful illustration from Lewontin: “If
the holocaust comes and a small tribe in the New Guinea forests are the
only survivors, almost all the genetic variation now expressed among the
innumerable groups of our four billion [1980] people will be preserved™
(Gould, 1981: 323).
2 For a review and summary of the evidence, see in addition to Gould (1981), Lewontin,
1982; Drechsel, 1991; Cavelli-Sforza, 2000; and the special issue of Nature Genetics 36
46 A Realist Philosophy of Social Science

Race (like ethnicity) is a social construction: we simply employ a
socially agreed-upon difference as the criterion for the grouping. Of
course, there are historical reasons which explain why the cluster of differ-
ences associated with physiognomy became the criterion for distinguish-
ing “races,” but this important issue cannot be pursued here.3 Since,
however, the idea of race as a biological notion has been reinvigorated
by some recent work, and because much is at stake, a brief diversion on
the topic of race is warranted. This will also provide a useful example of
problems in explaining and predicting phenotypical outcomes, including
here the best cases for study “ the range of diseases which include sickle
cell anemia, type 2 diabetes and multiple sclerosis.

Biology, race and disease
This new vigor, along with some old problems, comes with the widely
disseminated idea that gene studies have transformed forensics, with its
use in understanding and predicting the probabilities of diseases, and
with the construction of so-called “ethnic drugs,” for example, BiDil,
used for the treatment of heart disease among African-Americans. All of
these have made an impact on the popular imagination, too often in a
misleading or downright mistaken form.4 Accordingly, it is critical to be
clear on the central issues.
First, there remains agreement that there are no gene variants present
in all individuals of any demographic group and absent in individuals in
any other such group. Indeed, there is considerably more genetic variation
within populations as between them (Bonham et al., 2005: 12). Second,

3 See Hanaford, 1996; Voegelin, 2000; Henningsen, 2004. See especially Lentin, 2004,
who argues that racism was not an aberration in the modern democratic state: “On the
contrary, ˜race™ and racism, following central authors such as Bauman, Arendt, Voegelin
and many others is shown, not only to be a particularity of modernity, and speci¬cally of
the mid-nineteenth century on, but also to be grounded in what Gilroy (2000: 59) calls
a ˜statecraft™ which at a particular historical moment requires a notion of racial hierarchy
as the legitimating framework of its actions” (Lentin, 2004: 11).
4 A particularly egregious case was the misreporting of the efforts of the National Human
Genome Center of the College of Medicine at Howard University. A headline in the New
York Times (May 27, 2003) read: “DNA of Blacks To Be Gathered To Fight Illness.”
The article reported that “samples would be used to ¬nd genes involved in diseases with
particularly high rates among blacks like hypertension and diabetes.” But, indeed, its goals
were hardly so narrow and misconceived. Rather, its aim is “to study the complex interplay
between environmental and genetic factors.” See Rotimi, 2004. Another example is the
Op Ed essay in the New York Times (March 14, 2005) by Armand Marie LeRoi, “A Family
Tree in Every Gene.” LeRoi seems to think that correlations are suf¬cient to establish
distinct racial groups. He seems also to confuse “race” and “ancestry.” See below. See
also Jerry A. Coyne™s review of Vincent Sarich and Frank Miele, Race: The Reality of
Human Differences, in the Times Literary Supplement (February 25, 2005).
Explanation and understanding in the social sciences 47

there is no argument that there are correlations between phenotypical
outcomes and genetic variation. Third, as already insisted, the problem
is not that genes are not causally critical to phenotypical outcomes, but
that the explanation of these outcomes cannot, in general, be reduced to
genetic mechanisms.
What then is the problem? It stems from the fact that while as Cavalli-
Sforza says, “each population is a microcosm that recapitulates the entire
human macrocosm” (2000: 28) it is also true that “frequencies of genetic
variation and haplotypes differ across the world” (Bonham et al., 2005:
12). Critical here are so-called SNPs, a special sort of genetic “marker.”5
Enormous ingenuity and energy have been devoted to identifying SNPs.
These also are critical to the idea that there are “ethnic drugs,” or drugs
speci¬cally pertinent to “ethnic estimation based upon allele frequency
variation” (Duster, 2004: 7).
The route to misunderstanding is easily identi¬ed. One begins by
noting a statistical difference in the incidence of some disease, for
example, sickle cell anemia, between African-Americans and European-
Americans. One then identi¬es a correlation between these differences
and differences in genetic variation in the two groups: an association is
established between being a member of a phenotypically de¬ned “race”
and sharing in the particular genetic variation. Since genes are surely
causal, we conclude, mistakenly, that racial differences explain differ-
ences in phenotypical outcomes. The fallacy is plain: these are all correla-
tions and not particularly strong ones at that. Thus, socially constructed
categories of race and ethnicity in use are reasonably correlated with
ancestry,6 but given that the individuals may have membership in several
bio-geographical clusters, that the borders of these are not distinct and
are in¬‚uenced by sampling strategies, ancestry is not race. While it has

5 Some further critical terms may be introduced here: an allele is a form of a gene which
codes for one possible outcome of a phenotype. For example, Mendel found that there
were two forms of gene which determined the color of a pea pod. Accordingly, alleles
are causal. SNPs (single nucleotide polymorphisms) are alleles whose sequence has only
a single changed nucleotide. For example, in the genetic code, GGG becomes GGC.
“Genetic variants that are near each other tend to be inherited together. For example, all
of the people who have an A rather than a G at a particular location in a chromosome
can have identical genetic variants at other SNPs in the chromosomal region surrounding
the A. These regions of linked variants are called haplotypes.” “The number of tag SNPs
that contain most of the information about the patterns of genetic variation is estimated
to be about 300,000 to 600,000, which is far fewer than the 10 million common SNPs.”
Haplotypes, then, may be correlated with diseases (see www.hapmap.org). It is generally
recognized that while convenient, there are obvious dangers in this approach, of which
some are noted below.
6 In contrast to SNP studies, inferring ancestry from DNA requires a very large number
of loci. Inferring ancestry from such data remains probabilistic. See Jorde and Woodling,
2004: 531“532. See also Cavalli-Sforza, 2000: 31.
48 A Realist Philosophy of Social Science

better predictive value than race, ancestry, then, is a weak correlate for
variation across the genome.7 An example makes the case: “the town of
Orchomenos in central Greece has a rate of sickle cell anemia twice that
of African-Americans and . . . black South Africans do not carry the sickle
cell trait” (Rotimi, 2004: 545). That is, since one can be phenotypically
“black” and lack the variation and one can be “white” and have it, race
(comprehended as biologically meaningful) explains nothing.8
Indeed, the epigenetic character of outcomes shows that even if the
probability of an African-American having heart disease is higher than
for European-Americans, this need not be explained in terms of genes,
still less in terms of race or ancestry. Again, an example makes the point.
As Duster pointed out, a classic epidemiological study (using the same
quantitative methods as used in gene / disease studies: see appendix A)
concluded that hypertension among African-Americans need not be the
direct result of genes; rather “darker skin color in the United States is
associated with less access to scarce and valued resources in society. There
is a complex feedback loop and interaction affect between phenotype and
social practices related to this phenotype” (Duster, 2005: 1050).
Finally, and morally critical, employing racial surrogates for SNPs not
only risks reinscribing race as an explanatory biological category, but risks
denying appropriate therapy to persons who could bene¬t. For example,
as regards hypertension, despite the correlated patterns, “many African-
Americans would respond better to ACE inhibitors than would many
European Americans” (Jorde and Woodling, 2004: 528).9
Similarly, Duster also rightly argues that there are appropriate and
inappropriate forensic uses of these new capacities. Since the genotype is
unique, a match (or the absence of one) may be decisive as regards the
guilt or innocence of a suspect or of someone already wrongly imprisoned.
Problems arise with “the dangerous intersection of ˜allele frequencies
in special populations™ and ˜police pro¬ling via phenotypes™ ” (Duster,
2004: 10). The problem is that having identi¬ed some person by a “racial
category,” the criminal behavior is “explained” and “predicted” by appeal
to allele frequencies. Nor is sampling bias altogether overcome by having

7 This may be generous given that the standard technique examines only a few selected
loci in the DNA. As Duster notes, “what is being assessed is the frequency of genetic
variation at a particular spot in the DNA of each population” (2004: 8).
8 Put in other terms, the database problems are huge and are a consequence of the fact
(already noted) that “not only do all people have the same set of genes, but all groups
of people also share the major variants of those genes” (Rotimi, 2004: 544, quoting
Steve Olsen). Thus, inferences drawn from one or two African populations will likely be
different than a sample of 100 African populations drawn from very different geographical
9 For discussion of the BilDil case. see Rotimi, 2004 and Duster, 2004, 2005.
Explanation and understanding in the social sciences 49

a universal DNA database, since if the police are not stopping white
cocaine users, it does not matter if their DNA is in the database. And,
of course, “DNA is only as reliable as the humans testing it” (New York
Times, May 16, 2005).
Chapter 1 argued that causes are not merely correlations and chapter 2
insisted that explanation and prediction are not symmetrical. Both ideas
were in the background of the foregoing discussion of race and biology.
In this section we argued that while genes certainly ¬gure in explanation,
properly understood, race as a biological category does not. Phenotypical
outcomes, whether they are diseases or behaviors, are causally complex
products. We must resist the easy assumption that any single mechanism
or event from among the ensemble of events and mechanisms, physi-
cal, chemical, biochemical, biological and social, is suf¬cient to explain
some outcome, whether it be schizophrenia or measured competence in
an IQ test. But once we put aside “the 800 pound gorilla which is race”
(Duster) “ and this may be harder than we think, and we fully acknowl-
edge the complexity involved “ there can be little doubt that current work
in genetics can give us a better understanding of the role of genetic mech-
anisms in phenotypical outcomes. By recognizing database problems and
the limits of exploiting correlation, many researchers are now aspiring to
the situation where in medical decision-making, disease-related genetic
variation is directly assessed.10

Consciousness and collective intentionality
To complete the account of human powers, and indeed, to understand
distinctly human actions and outcomes, we need to take another giant
step which lays the foundations for social mechanisms (chapter 4). This
requires that we identify a critical emergent causal product of our species-
speci¬c brain and central nervous system. It is consciousness and the
capacity of mind to represent objects and situations outside itself “ tech-
nically what is termed intentionality (Searle, 1983, 1992). While it is
next to impossible to deny that humans have this capacity, we still lack
any sort of adequate understanding of it. Included in this is a capacity

10 The Haplotype Map Project (HapMap) assumes, contestably, the “common disease
“ common variant hypothesis” which further assumes that complex diseases are in¬‚u-
enced by SNPs that “are relatively common in human populations” (Rotimi, 2004: 543).
The HapMap project has tended also to encourage the rei¬cation of racial categories.
But until direct assessment of disease-related genetic variation becomes feasible, there
remains disagreement regarding trade-offs in the use of current techniques for predictive,
diagnostic and therapeutic uses. See especially Duster, 2004, 2005; Jorde and Wooding,
2004: 532 and Rotimi, 2004.
50 A Realist Philosophy of Social Science

for collective intentionality, usually unnoticed, and, a capacity for lan-
guage that is always noticed “ and is almost certainly an essential feature
of human society. Following Searle, collective intentionality means not
only that persons have the capacity to engage in cooperative behavior
and use a language, but that they can “share intentional states such as
beliefs, desires and intentions.”11
There is no good reason to be squeamish about the idea of a collec-
tive intentionality. To be sure, there are those who have supposed that it
requires an untenable ontological commitment, an independently exist-
ing Hegelian spirit or a Durkheimian “collective conscious.” Of course, all
consciousnesses are individual, in someone™s brain. As powerfully argued
by George Herbert Mead and John Dewey, mind is necessarily social.
Accordingly, if humans everywhere and anytime, abstractly have these
capacities, given that societies differ, they will be concretely realized in a
wide variety of ways.
It will be useful to distinguish realized capacities, e.g., the ability to
speak (say) Dutch, from capacities as potentialities, the ability to acquire
language. Capacities as potentialities are biologically determined but in
actual development (contrary to our mind experiment), realized capac-
ities are not. That is, social mechanisms (like genetic mechanisms) are
necessary causes. Homo sapiens everywhere and anytime has the poten-
tial to be “minded” and linguistically competent, and more generally, to
function in society. But of course, depending upon the time and place,
children acquire some very different languages. That is, in the actual
world, the potential is concretely realized in differing societies.12 There

11 Searle (1995: 23“26) gives a linguistic argument. Mead™s “social behaviorism” is cer-
tainly the best explanation we have of the fact that we can share intentions, cooperate
“ indeed communicate. In arguing against both Wundt and Watson, his problem was
precisely to explain mind and meaning in terms consistent with Darwin. In sum, “self”
presupposed “communication” which presupposed “meaning” which presupposed “sig-
ni¬cant symbols” which presupposed “vocal gestures” which presupposed the “conver-
sation of gestures” already available to lower animals. “Acts” were “social” in exactly
the sense that “the human animal has the ability over and above the adjustment which
belongs to the lower animal to pick out and isolate the stimulus. Mentality consists in
indicating those values to others and to one™s self so that one can control one™ s response”
(Mead, 1967: 132). See also Gillespie, 2005. Of course, this still leaves many questions
unanswered. Bickerton (1990) provides a powerful account of the origins of language
which draws on evolutionary theory, biology and linguistics.
12 As we might expect from our evolutionary history, there are important correlations
between populations de¬ned in terms of ancestry and languages. Cavalli-Sforza asks:
“How is it possible for these two very different systems to follow parallel evolution-
ary trajectories, to ˜co-evolve™? The explanation is quite simple: two isolated popula-
tions differentiate both genetically and linguistically. Isolation, which could result from
geographic, ecological, or social barriers, reduces the likelihood of marriages between
populations, as a result, reciprocally, isolated populations will evolve independently and
Explanation and understanding in the social sciences 51

are probably biologically grounded propensities or tendencies of other
sorts, for example, toward cancer and schizophrenia and perhaps also
traits of personality, for example, temperament, and musical or mathe-
matical pre-dispositions. Some people have a tin ear; some cannot hit the
curve ball; others seem especially apt with numbers or things mechan-
ical. Many potentialities of persons are either not realized at all or are
barely realized. There are many reasons for this. One obvious reason:
other conditions necessary to realize the capacity were absent: insuf¬cient
protein; no violin; no teacher. Another obvious reason is that realizing
some capacities often requires work, often at a sacri¬ce of other goals and
From birth onward, then, in order to realize their distinct human capac-
ities, humans need to interact with other humans. This is also a com-
plicated epigenetic causal story requiring contributions both from the
developing child who is an active participant and from the wider social
environment: the immediate nurturer, family, friends, consociates, then
teachers and so on.13 Since the process is in time, everything that happens
can have effects on what will then happen. At some point “ and evidently
quite early on “ a person with a personality “ a distinct ensemble of
habits, attitudes and beliefs “ emerges.
Three fundamental theses would seem to follow:
1. Except for humanness, nothing is programmed. But we can have a better
understanding of the relevant mechanisms, biological, psychological
and social which, taken together, produce “personalities,” and, based
on this knowledge, it will be possible to offer some very useful general-
izations, for example, that regarding persons experiencing crisis situa-
tions in their lives, the probability of depression is less among persons
with two copies of a long allele of the gene known as 5-HTT.14

gradually become different” (2000: 15). Not only is isolation a highly relative matter
(contact is continuous and reveals itself both linguistically and genetically), but because
the microcosm recapitulates the macrocosm (notes 2 and 7 above), populations, best
de¬ned on the basis of endogamous behavior (a tendency to marry and reproduce within
the group), are not races.
13 In a powerful but not widely acknowledged account, Harris (1998) offers an explanation
of “why children turn out as they do.” For her, “parents matter less than you think
and peers matter more.” From the present point of view, while she uses the evidence
of behavioral genetics to refute standard psychological misconceptions, especially in
undermining assumptions of correlations, she seems a bit insensitive to the transactional
or epigenetic character of all development.
14 This is a conclusion of a British and New Zealand longitudinal study as reported in the
New York Times, July 18, 2003. The mechanism is also identi¬ed. 5-HTT “contains the
code to produce a protein that escorts the chemical messenger serotonin across the spaces
between brain cells, or synapses, and then clears away the leftover serotonin. Drugs like
Prozac, Paxil, Zoloft and Celexa, which are widely effective in treating depression, work
by acting on the serotonin system.”
52 A Realist Philosophy of Social Science

2. The causal complexity of human development assures that, even as
regards identical genotypes, concrete persons will be idiosyncratic
3. While there remains considerable contention regarding the impor-
tance of biology in human behavior, nobody denies that both nature
and nurture are inextricably involved in all development (Ridley,
2003). But there is also an emerging consensus that deciding how
much of either is a question that cannot be answered. Because devel-
opment is epigenetic and causes are not additive, there is no reasonable
way to discriminate the causal importance of any of the countless fac-
tors, neither the enormous range of implicated mechanisms nor the
probably not identi¬able contingent events involved in outcomes (see
appendix A).
4. Given the complicated idiosyncratic biography of particular persons,
there is no reason to believe that any science could offer much improve-
ment over our ordinary ways of explaining the concrete behavior of a
person.15 Assuming that our character is causally linked to our behav-
ior, what we do in any particular circumstance also depends upon
highly variable concrete circumstances, how we understand these,
what particular judgements we make, and how we assess aims and
alternatives. Physics cannot explain or predict the ¬nal landing place
of a falling leaf. Behavior is caused, but once we grasp the complexity
of the causal nexus involved, it hardly seems plausible that any science
should enable us to improve on our ability to explain and predict the
concrete acts of individuals. We turn directly to this question.

Science and the explanation of the actions of persons
It is very often held that it is the task of a social science to explain behavior.
It is further assumed that what a person does has natural causes and that

15 Despite persistent assertions that “the explanation of behavior” is a goal, this includes
psychology as a science “ for all the same reasons. The task of psychology, as of other sci-
ences, is understanding, in particular the understanding of human powers: perception,
cognition, motivation, learning, imagination, language, etc. See Manicas and Secord,
1984; Margolis et al., 1986. Although developing this would call for another book, the
idea is not new. See, for example, Campbell and Misanin (1969: 77): “Few, if any psy-
chologists now believe that those conditions once labelled basic drives, such as hunger,
thirst, sex, and material behavior, are predominately governed by some common under-
lying generalized drive state, even if there is some activating or energizing state common
to many basic drives, it is clear that the speci¬c behaviors elicited by those drives are
controlled by a complex of inter-actions among environmental stimuli, hormonal states,
physiological imbalance, previous experience, etc. and that the basic drive concept is of
little value in unravelling these complexities.”
Explanation and understanding in the social sciences 53

explaining behavior, accordingly, requires their identi¬cation. If so, even
if we think we have free will, our acts are determined.
In this view, determined means caused; but in the context of the
dichotomy, free will versus determinism, it implies, critically, that a per-
son “could not have done otherwise.” That is, it denies the agency of per-
sons. It insists that even if we think that we always could have done other
than what we did, we are, in fact, automata, programmed by causes to do
just what we do. Our failures to explain and to predict behavior, then, are
merely functions of our ignorance: if we had all the pertinent laws, and a
precise description of all the initial conditions, predicting behavior would
be like predicting the positions of planets. Indeed, despite the fallacious-
ness of this idea of science, it is usually believed that we must presuppose
this if a human science is to be possible. This idea has historically been
at the bottom of a debate which began with the philosopher Immanuel
Kant. So-called naturalists take the position that we must bite the bul-
let and deny free will. Anti-naturalists take the common-sense position
that since we could have done otherwise, we need to reject altogether the
causal model of explanation. The alternative model on this view is that
of the historian: thus, Collingwood (1969:12):

The historian need not and cannot (without ceasing to be an historian) emulate
the scientist in searching for the causes or laws of events. For science, the event
is discovered by perceiving it, and the further search of its cause is conducted by
assigning it to its class and determining the relation between that class and others.
For history, the object to be discovered is not the mere event, but the thought
that expressed it. To discover that thought is already to understand it. After the
historian has ascertained the facts, there is no further process of inquiring into
their causes. When he knows what happened, he already knows why it happened.

Collingwood™s idea can be generalized. Understanding Roman history
requires that we understand why Brutus stabbed Caesar. We need to
grasp his reasons and beliefs. Similarly, as regards explaining why Sam
robbed the convenience store. (Both could have done otherwise.) And
surely this seems right. But Collingwood™s main conclusion is mistaken
since his image of causality (and of science) is mistaken. And because he
assumes that reasons are not causes.
As we argued in chapter 1, causal laws are not of the form, “whenever
this, then that,” and since the universe is not Laplacean, contingency
is the constant feature of all events in the world, including the acts of
persons. We can predict the position of a planet because there are only two
pertinent causes (inertia and gravitation) and three pertinent variables
(mass, velocity and position). Most critically, there is no “butter¬‚y effect”:
for all practical purposes, the system is closed. Remember that we could
54 A Realist Philosophy of Social Science

not explain or predict the ¬nal pattern of splintered pieces of rock from
a boulder rolling down a hill, that even in this very simple case involving
nothing human, there was an inherent incalculability resulting from the
fact that what happens at each instant has effects on what happens in
the next instant. Given that this is true of humans and that persons are
immensely complex open systems, it is hardly surprising that we cannot
predict (or explain) with the ease and certainty of celestial mechanics.
Indeed, suppose that just as I am about to start this sentence, an errant
throw of a baseball shatters the window in my of¬ce. The sentence I
started to write does not get written.
Indeed, as noted in the account of prediction in the previous chapter,
there is a paradox in explaining and predicting the acts of persons. We
are, in fact, quite good at both explaining and predicting the acts of per-
sons, quite independently of knowledge provided by the human sciences.
Indeed, as ordinary socialized human beings we are better at explaining
and predicting human acts than sophisticated science is at explaining and
predicting the ¬nal outcome of falling leaf. And, indeed, there is no reason-
able hope that the human sciences could do better in explaining and predicting
the acts of persons than we do in our own very pre-scienti¬c way.
Our ordinary explanations of action, of course, are not scienti¬c. They
take the form of providing reasons for what people do “ just as Colling-
wood suggests. Although this topic remains contentious in some quar-
ters, there is no good reason to say that reasons are not causes; and there
are good reasons to say that they are. To say that Sam did A because he
believes B seems unavoidably to mean that Sam™s belief that B is the cause
of Sam™s doing A, otherwise, there is no real connection between the rea-
son and the act. As Bhaskar writes: “If and whenever they explain . . . rea-
sons must be interpreted as causes, on pain of ceasing to explain at all”
(1979: 115). One can assent here that my reason to do so-and-so was
itself caused, but surely this hardly matters since it is my reason. Had I
chosen otherwise, that too would have been my reason.
Moreover, like other sorts of causes, the possession of a reason can be
a state or disposition: being honest gives one a reason to tell the truth.
Being a liberal gives one a reason for voting for a Democrat. Like other
causes which must be analyzed as dispositions, reasons may be possessed
even when not exercised, and even when exercised they may not explain
the act: in that case they would not be the reason for the action. On the
other hand, without thought, when appropriate conditions are present,
we act. Indeed, the overwhelming percentage of our actions fall into this
category: they do not, in general, require that we recognize, articulate
or acknowledge the reasons for our action. Of course, we may be asked
retrospectively to give an account, which we are generally in a position
Explanation and understanding in the social sciences 55

to do. This is, of course, a powerful insight of the ethnomethodological
The point is of considerable importance. As Searle has suggested, in the
social science literature, there are two dominating sorts of theories which
aim at explaining action. One is “mental causation, according to which
the agent is operating consciously or unconsciously, with a set of ratio-
nal procedures over more or less well-de¬ned sets of intentional states,
such as preference schedules or internalized rules” (1995: 141). This
model is now termed “rational choice theory,” but in various forms, it has
been around a long time.16 The other model “does not appeal to inten-
tional states but to brute physical causation.” Most powerfully associated
with behaviorism, it af¬rms causality, but caught in the mode of regu-
larity determinism “ “whenever this, then that” “ it denies agency alto-
gether. Searle is quite correct that what we need here is a causal account
“that will explain the intricacy, the complexity, and the sensitivity of
behavior, as well as explaining its spontaneity, creativity, and originality”
(1995: 141).
Beginning with the idea that people have reasons for what they do,
rational choice theory is an effort at spelling out what makes a decision to
act rational. To be sure, this model may sometimes seem appropriate. We
sometimes make a careful assessment of our situation, clarify our goals
and try to assess the pluses and minuses of alternatives according to some
rational ordering. But, ¬rst, this is not generally what happens. Moreover,
even when it does, we are not logic-machines. By the standards of modern
logic, we often do very foolish things “ even if we have our reasons. Thus,
according to the theory, it is irrational not to prefer a to c, if one prefers
a to b and b to c. On this theory, if you value two things, such as your life
and your nickel, there must be some odds at which you would bet your
life against the nickel. Decision theory says, if you say “no” to any such
odds, you are irrational. But, indeed, who is irrational?
A third model rejects both of these models and begins by extending
the idea, already noted, that the possession of a reason must be analyzed
dispositionally. More generally, as noticed by Aristotle, George Herbert
Mead, John Dewey, C. W. Mills and, more recently, Pierre Bourdieu,
our character and the habits we have developed, ¬gure hugely in what we
do, and they do so because they give us capacities, or powers, and these
dispose us to act in certain ways. Like the powers of the theoretical things
of science, these are tendencies which, although causally critical, do not,

16 Hempel offered that what he called “explanation by reasons” satis¬ed the D-N model
(1965: 463“487). RCT is well-ensconced in modern micro-economics and has become
prominent in sociology and political science. See below, chapter 5 and appendix B.
56 A Realist Philosophy of Social Science

in open systems, generate invariances. Knowing that Sam dislikes rock,
when asked whether he likes a recently released Jimi Hendrix album,
we may expect him to be critical, but what he says is not “determined.”
Indeed, he may well surprise us and tell us that he was pleasantly surprised
by several of the cuts. Similarly, if he is honest, the answer will likely be
an honest answer, but it may not be, if there are good reasons not to tell
the truth: for example, if he is disposed also to try to please you. In the
next chapter, this idea is complicated further to include the constraining
and enabling of action by one™s place in social relations. We can expect
that these will be causally related to who we are.
Because we know that there is connection between what a person does
and the reasons for doing it, we have a store of generalizations which give
us considerable predictive ability. Thus, we can predict (or explain) that
Sam will shortly go to lunch by knowing that he usually gets hungry at
such and such a time, that he dislikes being hungry, that nothing pre-
vented him from satisfying his desire to eat. As with any event, a host of
causal mechanisms are at work here, physical, chemical, biological, psy-
chological and social, and at least the ¬rst three are quite well understood.
Indeed, on the present point of view, a good deal of our ignorance of the
pertinent psychological and social mechanisms results from misconceiv-
ing the goals of the human sciences, but in particular, the assumption
that their aim is the explanation of behavior. In any case, we do very well
in explaining and predicting Sam™s behavior even without an adequate
scienti¬c understanding of any of these mechanisms. Social life preceded
modern science by several millennia, but it is very hard to imagine social
life in the absence of such competence. This knowledge is re¬‚ected in
human languages, whose concepts and distinctions regarding action are
the product of long historical experience. Max Weber correctly insisted
that as humans we have been “schooled in the world of own everyday
Here we might consider the problem of a Martian social scientist. He
lacks a human historical experience “ and perhaps also a different natural
history. If so, perhaps his perceptual system, not to mention his social
system, is radically unlike our own. Our actions would be utterly unin-
telligible to him. He would ¬rst need to identify the patterns of human
everyday life, and then seek an understanding of these in terms of the
causal mechanisms at work. He may well have an understanding supe-
rior to ours of the physical, chemical and bio-chemical mechanisms, but
he would need to put this to work to understand humans, after which
he would need to do some very serious human ethnography. Harr´ and e
Secord (1973) were absolutely correct to insist that the social scienti¬c
problems of our Martian would not be at all like our social scienti¬c
Explanation and understanding in the social sciences 57

problems, yet a good deal of our social science acts as if our ordinary
human understandings were utterly irrelevant.
There is a paradox, ignored in the preceding, that needs to be con-
fronted. There are no persons without society and no society without
persons. But society, unlike nature, does not exist independently of our
activities. We turn next to these problems.

In our effort to provide a meta-theory for inquiry in the social sciences,
we turn to the second huge difference between the natural sciences and
the social sciences: nature exists independently of us; society does not.
Searle has rightly argued that more fundamental than the distinction
between “nature” and “culture,” or “mind” and “body” is the distinction
between those features of the world that exist independently of us and
those that are dependent on us for their existence. Trees and molecules
exist independently of us. Were the human race suddenly to perish, they
would still exist “ even if our representations of them would disappear.
Money and science, by contrast, would perish along with us. Of course,
the paper which for us represents a dollar bill would still exist. It would still
be an ensemble of molecules with speci¬c causal powers. For example,
it would still burn. But it would not be money, since there would be no
one to use it to buy anything.
Searle (1995: 27) offers a distinction between “brute facts,” facts about
features of the world that exist independently of us, and “institutional
facts,” or facts about features of the world which require special human
institutions for their existence. Institutional facts necessarily require col-
lective intentionality: we share in believing that the paper is a dollar
bill which allows me to purchase the ice-cream cone from you. That
H2 O is water is a brute fact. That an ice-cream cone costs one dollar
is an institutional fact. There are other critical features of institutional
r Brute facts are logically prior to institutional facts. The “natural” (inde-
pendently existing) material world is the stuff out of which institutional
facts are made (and sustained). But these need not be physical objects
(as is the case with old-fashioned money) but may be magnetic traces
on a tape, or in the case of conversation, sounds coming from our
r Institutional facts are interconnected with other institutional facts. As
Searle writes: “In order that anybody in society could have money,
that society must have a system of exchanging goods and services for
money. But in order that it can have a system of exchange, it must
58 A Realist Philosophy of Social Science

have a system of property and property ownership” (1995: 35). This in
turn requires a system of law. This also suggests that institutional facts
may presuppose other institutional facts, suggesting different degrees of
depth or generality. But it does not follow from this, that the ensemble
of institutional facts comprise an integrated totality.
But if we now think of society as an ensemble of interconnected insti-
tutions, we must face the troubling consequence that society™s existence
depends wholly on us. If so, then it would seem to be all in our heads,
and hence, it is hard to see how it could function causally. On the usual
reading, Durkheim was the ¬rst to see this problem. On the usual reading
of Durkheim, social facts are external to us and have a coercive power.
This then easily explains the stability and regularities of action. Indeed, it
almost seems commonsensical to say that society in¬‚uences our behavior.
Especially since the in¬‚uential work of Talcott Parsons, social scientists
have been more or less committed to some version of Durkheim, often
without noticing. Symptomatically, appeals are made to social forces, or
to explanations of action in terms of social structure. Something external,
real and causal, like the forces of nature, seems essential for explanation.
But the account with which we began offers what seems to be an “idealist”
ontology of social reality. It has continued to trouble contemporary writ-
ers for whom Anthony Giddens™s “structuration theory” has often been
a point of departure. The critical concept in this regard is the concept of
“social structure” or simply “structure.”17

The concept of social structure
Giddens (1984) distinguished between “system,” “structure” and “struc-
r System: “The reproduced relations between actors or collectivities,
organized as regular social practices.” As relatively bounded ensem-
bles of practices more or less having a pattern, these are observable.
System in this sense is often referred to as “society.” But one needs
to be careful here. We don™t see society. We see parents teaching their
children; workers engaged in work, supervised by bosses; legislatures
making laws which are obeyed by citizens and so on. Second, the term
“system” in wide use is often taken to mean that its parts are function-
ally and coherently integrated. But as with institutional facts (above),

17 As Sewell notes, “˜structure™ is one of the most important and most elusive terms in the
vocabulary of current social science,” but for better or worse, “any attempt to legislate
its abolition would be futile” (1992: 3).
Explanation and understanding in the social sciences 59

Giddens makes no such assumption.18 The observable patterns are the
empirical point of departure for inquiry.
r Structure: “Rules and resources, organized as properties of social sys-
tems.” Structure is not the patterned practices, but the principles that
give pattern to the practices. Structure is thus a theoretical term, and a
highly abstract one at that. The observable patterns (system) are con-
crete, while structures are abstracted from them. Rules and resources
are interconnected.
Rules “imply methodical procedures of social interaction,” and are
“generalized procedures” applicable over a range of contexts and occa-
sions. They have two aspects: the constitution of meaning and the sanc-
tioning of modes of conduct. Finally, and critically, “rules cannot be
conceptualized apart from resources, which refer to the modes whereby
transformative relations are actually incorporated into the production
and reproduction of social practices. Structural properties thus express
forms of domination and power.” (Giddens, 1984: 18)
r Structuration: “The conditions governing the continuity or transfor-
mation of structures.” To anticipate: if patterns are observable, and
structures are abstracted, then structuration refers to the mechanisms
which produce the patterns. For Giddens, structure has but a “vir-
tual existence.” As he writes, “structure enters simultaneously into the
constitution of the agent and social practices and ˜exists™ in the gener-
ating moments of this constitution.” This is certainly the central idea in
Giddens™s formulation.
Structuration involves the idea of “duality of structure,” that social
life is fundamentally “recursive,” that is, agent and structure presuppose
one another: there is no action without structure and no structure with-
out action. Accordingly (as Mead had insisted), there was no time when
there were agents and no society, and conversely. We should, accord-
ingly, avoid saying that persons create society. Instead, since all individ-
uals are born into actually existing societies, social structures pre-exist
for them, incarnate in the ongoing activities of members. Hence, they

18 Although Giddens has written most perceptively in rejection of functionalism in social
theory, his own use of “system” is undertheorized and perhaps even dispensable. His
best statement may be: “The connotation of ˜visible pattern™ which the term ˜social
structure™ ordinarily has, as employed in Anglo-American sociology, is carried on in my
terminology by the notion of a system: with the crucial proviso that social systems are
patterned in time as well as space, through continuities of social reproduction” (Giddens
1979: 64). Following this, he also says: “A social system is thus ˜a structured totality.™”
But this is wildly misleading at best. William Sewell makes some very useful suggestions
regarding the “multiple, contingent, and fractured” character of society and of structure.
We return to this.
60 A Realist Philosophy of Social Science

become persons in society, and by their actions they reproduce and trans-
form it.
We want to preserve the Durkheimian insight that society in¬‚uences
behavior. We can see more clearly what this must mean. We can think
of society as an ensemble of practices. There is, ¬rst, the social process
of becoming a person. A baby is born into a world of adults acting with
materials at hand. Their activity is thus structured, both enabled and
constrained. That is, there are materials being employed in particular
ways by people. Some of these individuals rear that child, passing on a
fantastic range of tacit knowledge about activities in their society. The
child creatively appropriates this, becoming an active participant. A bat-
tery of dispositions are formed giving that person a personality. For every
generation there are given materials, materials given by the activities of
previous generations going back into time. This is one of the facts that
makes history important to any human science.
But society (as incarnate in the activities of persons) in¬‚uences behav-
ior also in the sense that structures make available the range of choices
available in society to the socialized person and this point runs from the
obvious to the not so obvious. You may know Hindi but if you are liv-
ing in a society with no Hindi speakers, you will not speak Hindi. If you
lack the language of that society, you will likely make the effort to learn
it. Less obvious, different societies make different kinds of alternatives
available to different kinds of people. Here is where theories of race, gen-
der, class and status enter the picture. For example, in some concrete
social system, “class” can refer to objective social relations between peo-
ple. “Class” is a theoretical term which, by abstraction, might well be a
principle which explains, against the background of a host of institutional
facts, a structured pattern of practices, for example, the mechanism of
labor markets.
Thus, structure is both medium and product of conscious activity.
Structure is a medium in the sense that it is material used, both enabling
and constraining. For example, a person knows a language and thus can
speak. She creates her sentences with the materials of the language; she
uses it to describe, protest, explain and so on. On the other hand, she
is also constrained by her language. To be understood, she must con-
form, more or less, to the rules of that language (even though these rules
are mainly tacit, unacknowledged by speakers). Some sentences make
no sense. Sometimes, she strains to communicate her meaning, perhaps
by creatively employing a metaphor. And some things simply cannot be
These features are fully generalizable. Everything we do involves
socially available materials, what are often called institutions. When we
Explanation and understanding in the social sciences 61

work, we work with materials, language and all the particular rules,
relations and tools which make up that work activity. Thus there are
tasks expected of us and ways to accomplish these, we have a boss who
can ¬re us, and we work with a computer, ¬les, telephones and so on. So,
too, when we play, marry, worship, engage in politics and so forth. When
we play a game, chess for example, the movement of pieces is prescribed,
but within these rules, we are free to decide what moves to make. When
we marry, both partners accept roles and responsibilities, acquire some
rights and lose others. And so with all the things we do.
On the other hand, social structures are products in the sense that,
since language is embodied in concrete utterances, as an unintended
consequence, when we speak and write, collectively and cumulatively,
we reproduce and transform it. And similarly with all other activities: our
work activity realizes the rules and relations which are incarnate in that
sort of work activity; our interactions with a mate realize family life and
so on.
Since structures are virtual and exist only as incarnate in ongoing prac-
tices, social science is inevitably historical and concrete. If we want to
understand present practices we must acknowledge that they are histor-
ical products. But different historical experiences will make otherwise
similar societies concretely different. One can speak abstractly of France,
South Korea, El Salvador and Canada as capitalist societies, but their
very different histories make them, today, very different in many impor-
tant ways.
The historical and concrete character of social science generates special
problems for theory. While theorizing is never ¬nished in any science, in
the social sciences, theory is continually revisable not merely in the sense
that new theories replace or amend older theories, but in the sense that
reality is changing. Given the immense changes in American society since,
for example the 1950s, theories developed to understand the American
family then will likely not be suitable today.
Understanding social change is, indeed, a critical feature of the social
sciences and there are all sorts of possible explanations for social change,
depending upon concrete material and historical conditions. But it
remains true that aside from natural events, hurricanes and the like,
everything that happens in society is produced by persons working with
materials at hand, sometimes as the intended, but usually as the unin-
tended consequences of their activities. Understanding social change,
then, requires speci¬c hypotheses of existing social mechanisms which
detail, concretely, the capacities that agents have and the constraints to
which they are subject, what they know and understand and, ¬nally, the
uses to which they put their capacities and knowledge (chapter 4). To
62 A Realist Philosophy of Social Science

merely hint at what is at issue here, compare the recent past of the former
USSR and the United States during its recent national elections. Fan-
tastic differences in the structure of the two economies, their political
arrangements and so on, made for enormous differences in the capaci-
ties and constraints of persons, from leaders, such as Putin or Bush, to
ordinary workers and citizens.
The historical and concrete character of the social also raises a danger
about generalizations. Since social phenomena are historical and con-
crete, generalizations that are meant to apply to many or all societies
may easily lead to triviality or distortion. It is, for example, almost surely
true that “an organization is more likely to be strongly centralized during
external crises than during normal periods” or that “economic / demo-
graphic resources of contending states determine capacities for military
domination.” But these would be extremely unhelpful in understanding
the behavior of France during the Napoleonic wars or the very different
situation of (say) the USSR in the 1950s. Similarly, one needs to dis-
tort the ordinary understanding of “entrepreneur” to hold that “in all
societies, entrepreneurs have been the catalyst for change.” This is not
to deny the importance of generalization in social science as potentially
descriptive and thus illuminating. We might remind ourselves here that,
as argued in chapter 1, it is an important task of science to provide expla-
nations of signi¬cant patterns and generalizations. This will be true also
of the social sciences. Just as we can understand oxidization and under-
stand why iron rusts, we need to understand sexism and why, despite
efforts to the contrary, there is still a glass ceiling (chapter 4).

The double hermeneutic
Because social structures do not exist independently of human activity,
there is a critical epistemological implication regarding inquiry into soci-
ety. In all sciences, since scientists must communicate with one another
regarding their claims about the world and, hopefully, come to a shared
understanding, they are engaged in hermeneutics “ “the art of interpre-
In order to build a consensus about claims made, all scientists must
continually seek mutual understanding about such claims, the standards
and methods employed, the evidence adduced and so on. For the natural
scientist, nature stands independently of us. Brute facts are whatever

19 Originally, hermeneutics was the theory and method of interpreting the Bible, extended
by Dilthey to human acts and products.
Explanation and understanding in the social sciences 63

they are irrespective of our human interests or of the meanings we
might impose on them. The social scientist must also build a consen-
sus about claims and theories about society; so social scientists are, with
one another, similarly engaged in a hermeneutic process.
But for the social scientist, there is a “double hermeneutic” (Giddens,
1984). The world that the social scientist is describing, communicating
and seeking consensus about is itself a meaningful world, a world having
meaning for the members of the society under study. As argued, activity
is meaningful in that human action involves concepts, rules, norms and
beliefs that are shared by members. This datum is the point of depar-
ture of inquiry in the social sciences. Social scientists must come to an
agreement about what are already ongoing interpretations by members
whose activities constitute their world. We must understand it if we, as
scientists, are to communicate and con¬rm claims about what is going
on in that world. We need to grasp their motivations and the norms they
live by. We need to know what, for members, counts as marriage or ¬lial
piety, what is immoral, criminal, democratic, just and unjust. This is as
much true of our own society “ and its subcultures “ as it is of so-called
“exotic” societies, even if we often “ and disastrously “ think that we
can understand or explain outcomes without considering what activities
mean for members. Indeed, this was precisely Schutz™s criticism of pos-
itivist social science: the positivist thinks of himself as a natural scientist
making claims about an “objective” world to which he gives meaning. In
his criticism of the work of Talcott Parsons, Schutz rightly insisted:
Professor Parsons has the right insight that a theory of action would be meaning-
less without the application of the subjective point of view. But he does not follow
this principle to its roots. He replaces subjective events in the mind of the actor
by a scheme of interpretation of such events, accessible only to the observer, thus
confusing objective schemes for interpreting subjective phenomena with these
subjective phenomena themselves. (Grathoff, 1978: 36)
But, the answering of our question, “What does the social world mean for me,
the observer?” has as a prerequisite the answering of the quite different questions,
˜˜What does this social world mean for observed actors within this world, and what
did he mean by his acting within it?” With these questions, we no longer naively
accept the world and its current idealizations and formalizations as ready-made
and meaningful beyond all doubt, but undertake to study the process of idealizing
and formalizing as such, the genesis of the meaning which social phenomena have
for us as well as for the actors, the mechanism of the activity by which human
beings understand one another and themselves. (Wagner, 1983: 48)

This plainly is the strong suit not only of ethnography, but of symbolic
interactionist orientations, ethnomethodology, Goffman™s work and other
64 A Realist Philosophy of Social Science

forms of what is called qualitative research.20 But, indeed, this “ethno-
graphic moment” is essential to any explanatory effort in the social sci-
ences “ and for that matter, in history (see chapter 5).
This raises two dif¬cult questions. First, is it possible for the social
scientist (historian or linguist) as outsider to come to grasp the meanings
of action in the society under study? Perhaps only natives, insiders, really
understand what is going on. Second, and more problematically perhaps,
can we ask whether the natives™ understanding of their world is adequate?
Do they really understand what is going on?

Ethnographic skepticism?
Ethnographic skepticism is healthy, but one need not turn a problem
into an impossibility.21 Consider again our taken-for-granted ability to
understand one another in our everyday lives. As Weber pointed out, this
involved what he called verstehen, the human capacity to grasp the mean-
ing of another™s actions. We must not think of verstehen as some sort of
special, intuitive, sympathetic understanding, a reliving of the experience
of others. Verstehen is something we all do all the time. We are engaged
in verstehen in judging that a person on a ladder is painting the house, in
judging that the expression on another™s face is distress produced by our
careless remark, and so forth. We learned to do this, indeed, when we
learned to use language. There is nothing dubious about such judge-
ments since, as with any judgement, they require evidence and may,
subsequently, be rejected.
Second, our ethnographer is not a Martian, but a human being. Even
if the culture she studies is very different to her own, it remains a human
culture: verstehen will still be critical.22 In the worst case, accordingly, the
researcher has available the same evidence that the members have “ the
actions and products (for example, texts and artifacts) of members. Some
actions will be immediately understood: they are seeking food or building
a shelter. Moreover, as the philosopher W. V. Quine argued, hypotheses
about meaning are tested and either seem to work or they don™t. Eventu-
ally (following what Mead had to say on the subject) interactions succeed,

20 There are various forms of qualitative research, including participant observation, inter-
views, analysis of texts and documents, focus groups and so on, that offer ways to grasp
members™ understanding of their world, some better than others. Quantitative work that
often appeals to surveys is not among the better ways. But discussion of these issues must
be forgone here. As noted, in appendix A, quantitative work may still be descriptively
and evidentially important. See, for example, C. Wright Mills, 1959.
21 For examples and discussion, see Clifford and Marcus, 1986; Rosaldo, 1989.
22 Sahlins (2004: 5) suggests the ethnographer must be a cultural animal if he is to make
any headway, but surely to be a human (cultural) being is a very big advantage.
Explanation and understanding in the social sciences 65

expectations are realized, there is communication and understanding. Of
course, this will take some time, and of course, our ethnographer might
be wrong “ perhaps in detail, perhaps in some fundamental way.
Moreover, the idea that the native has a privileged or unique under-
standing that is inaccessible to the other, runs into a logical dif¬culty. We
are all situated and there is no God™s-eye view of the world. Consider,
then, other possible privileged viewpoints: the colonized, women, black
women, women of color, upper-class women of color, urban lower-class
women of color and so on. The issue is not whether these voices have
been suppressed in white male dominated positivist social science: they
have. Nor is it argued that much qualitative work is poorly done, distorted
in this way or that. The issue rather is epistemological: because each of
us, logically, has a unique biography and position in society, each per-
son™s viewpoint is unique. We seem driven beyond relativism to a radical
subjectivism. I cannot be a native but I cannot be you either. That is, the
problem of understanding the other begins at home. In everyday life, we
do not turn a problem into an impossibility. Indeed, as Mead and Schutz ¨
insisted, if I am to communicate with you at all, I must in some measure
take your position.23
Any viewpoint, accordingly, will leave much out. The social scientist
is obliged to take care that other voices are heard, that the account is
as objective as is humanly possible. Granting that such objectivity is sit-
uated and not absolute, a situated objectivity will require re¬‚exivity in
Bourdieu™s sense. In addition to the obvious potential positional biases,
there is the question of “the objective space of possible intellectual posi-
tions offered to him or her at a given moment . . .” and ¬nally, there is
the intellectualist bias (rejected ¬rmly by Dewey long ago) of constru-
ing the world as “a spectacle” (Bourdieu and Wacquant, 1992: 39). On
the present view, knowing is not a re¬‚ection of reality, nor a mere con-
struction, but is “disciplined by the otherness with which it engages”
(Pickering, 1992: 412).24 Sahlins pertinently quotes Bahktin:
There is an enduring image, that is partial, and therefore false, according to
which to better understand a foreign culture one should live in it, and forgetting
one™s own look at the world through the eyes of this culture . . . To be sure, to
enter in some measure into an alien culture and look at the world through its
eyes, is a necessary moment in the process of understanding . . . [but] creative
understanding does not renounce its self, its place in time, its culture; it does not

23 Here again, the Mead / Dewey theory of meaning which rejects, at the outset, a Cartesian
point of departure gives us the philosophical ground that we need.
24 I perhaps here extend Pickering™s sense which speaks of the “mangle of practice,” and
“the dialectic of resistance and accommodation” (1992: 412). See also Fabian, 1991;


. 2
( 7)