46 The problems of structure and agency

establish their case that social structure is epiphenomenal, that is a mere

outworking of the doings of 'other people', then it also follows that they
have not succeeded in denying emergence. Their inability to withstand
claims that the social context has autonomy and independence from
people, pre-exists them, and is causally influential of them, means that
there should at least be a pause in Empiricist devotions to entertain the
case that a 'social structure' which has these properties also has a claim to
existence, though not one which can be substantiated through experience
as sense-data. Does Collectivism manage to sustain it?

METHODOLOGICAL COLLECTIVISM

Collectivism*s social structure

The irony of Collectivism is that whilst it defends the methodological
indispensability of 'structural factors', no overall conception of social
structure is advanced ontologically. What accounted for this is that
Collectivists were simultaneously haunted by Holism and hamstrung by
Empiricism. As far as the former was concerned, the proper desire to
evade any charge of reiflcation seemed to imply that safety lay in
refraining from making ontological claims as far as possible. Conse-
quently what we are actually dealing with most of the time is Methodolo-
gical Collectivism. Its overriding concern is with explanation and par-
ticularly with the deficiencies of the Individualists' programme of
reductionism. In criticizing it, the Collectivists' case rests largely on the
fact that references to the social context have to be included for explana-
tory adequacy, because accounts cast purely in terms of'chaps' just don't
work. They break down short of the goal (through failure of composition
laws) and 'societal properties' are needed to supply the deficit. Although
the point is also made that 'chaps', their dispositions and their doings
cannot even be identified (i.e. described as 'believers' or 'voters' etc.)
without further resort to the social context, this is not used to issue an
ontological challenge to the Individualists' concepts of 'structure' and
'agency'. On the contrary, when Individualists defended their backs by
promptly incorporating all such social features into their conception of
individual people, the Collectivists noted the fact, commented that it
would be unhelpful in explaining the relations between what we now call
'structure and agency', but backed away from an ontological confron-
tation by deeming this to be a matter of semantics. Since the Individualist
was arbitrating about the ultimate constituents of social reality, it is hard
to see that this could be let pass as merely an issue about the use of words -
particularly when identifying explanations are often at stake.
 Individualism versus Collectivism 47
In other words, the Collectivist was playing an inordinately defensive
game. References to 'societal facts' are defended as ineradicable 'remain-
ders', without which Individualists' descriptions remain incomplete, and
also as indispensable adjuncts when Individualists' explanations come up
against the 'irreducible'. The very language of 'remainders' and 'unre-
duced concepts' casts the Collectivist in the role of critically supplement-
ing Individualism, rather than confronting it head on. Instead of articu-
lating a robust counter-concept of 'social structure', the Collectivist
cautiously indicates points at which some aspect of society is necessary to
explain this or that and only becomes exuberant when detecting Indi-
vidualists busily committing sins of commission, on their own terms, by
incorporating such references anyway.
This means that the Collectivist deals with the 'social structure' in the
most fragmented way, as a disparate collection of facts or factors which
are only brought forward when Individualism fails. Yet when they are
then adduced by Collectivists, the question cannot be evaded as to their
ontological status. Here, the spectre of Holism and the fear of reification
made the Collectivist response as circumspect as possible. Gellner, for
instance, was far from content with 'descriptive individualism' as the
necessary bulwark against Holism: whilst it warded off the reined
ghoulies, he clearly considered that it also cordoned off important tracts of
the field which contained things quite other in kind than 'geists' and
'group-minds'. Thus, he speculates that the patterns we are capable of
isolating in our environment and reacting towards are not 'merely
abstracted', not simply mental constructs. He then invites us to consider
that 'For any individual, the mores, institutions, tacit presuppositions,
etc. of his society are an independent and external fact, as much so as the
physical environment and usually more important. And if this is so for
each individual, it does follow that it is so for the totality of individuals
composing a society.'21 What then is the status of these patterns in whose
terms the everyday actor thinks and in relation to which s/he acts, as does
the observer who also recognizes that they cannot be eliminated from his
account of social life? The way the reply is couched is revealing. 'The
pattern isolated, however, is not "merely abstracted" but is as I am
somewhat sheepishly tempted to say, "really there".'22
To examine the origins of the 'sheepishness' is important for they were
responsible for withholding full ontological status from 'societal proper-
ties' for decades. Tentativeness is rooted in two spectres of reification and
the seeming difficulty of affirming the existence of 'societal properties'

21
Gellner, 'Holism versus individualism', p. 264.
22
Gellner, 'Holism versus individualism', p. 264.
48 The problems of structure and agency

without invoking one of them. The first was J. S. Mill's old fear, 23 namely
that to acknowledge emergence was to countenance the existence of a new
'social substance'. The second was that talk about 'societal properties'
was also talk about things produced or generated by society, indepen-
dently of the activities of people and therefore superordinate to actors. In
fact, it appears that both very proper anxieties were really semantic in
origin, turning respectively on the Greek word (ousia) for substance
(which need never be employed in relation to emergence) and the Latin
phrase 'sui generis' (which has been misused in this connection).
The first source of unease is the doubt that any referent of a holistic
concept can have an effect upon concrete individuals, since this seems to
endow an abstraction with some kind of existence which cannot be flesh
and blood and therefore must entail a different substance (from people) if
it is real. As Gellner writes, putting himself in the shoes of the Individual-
ist, 'Surely the insubstantial cannot constrain the substantial? I think we
can provisionally agree to this principle'. 24 In other words, the only two
alternatives seemed to be to credit 'societal properties' with some
mysterious substance or to withhold reality from them. The language of
substances proved as damaging in sociology as ousia has been in the
Tridentine concept of 'transubstantiation', which construes eucharistic
theology in terms of sacramental physicalism. In social science the
problem was identical, only (rightly) sociological physicalism had no
takers. The real problem was that the wrong language was being
employed, even more by the opponents of 'societal properties' than by
their sheepish advocates. In consistency, the Individualist who felt
confident wfien pointing to flesh and blood people, surely did not hold
that they were invoking dubious 'substances' when they (necessarily)
referred to people's personalities, attitudes or dispositions? And if so,
then why should the defenders of'societal facts' be automatically guilty of
invoking such when referring to their 'non-observables' - for neither
could confine their terms of reference to sense-data, which is what
'substance' effectively stood for in this context.
The second source of concern derived from the current (and continu-
ing) mis-assumption that to consider 'societal facts' as being sui generis
entailed reification because it implied that they were generated by society
itself- as a separate and superordinate entity, 'Society'. Literally, the
phrase means nothing more than 'of its own kind'. In this case, 'societal
facts' do not pertain to the genus (class of object) made up of individual
people but belong to a different genus, i.e. the class of objects designated

J. S. Mill, A System of Logic Ratiocinative and Inductive^ People's Editions, London,

24
1984, p. 573. Gellner, 'Holism versus individualism', p. 262.
 Individualism versus Collectivism 49
by terms like society, social organization or social structure. Note, this
says nothing at all about their genesis, where they come from as
properties, it specifies only what kind of properties they are. The
confusion arises etymologically because the same word genus (of which
generis is the genitive means 'birth', deriving from the older Sanskrit verb
'jan', meaning 'to be begat'. Hence, the source of the Holistic error that
(reified) Society begets or generates its own (equally reified) properties.
However, when referring to things, such as 'society', it denotes merely
'sort' or 'kind'.
Collectivists were perfectly clear that they were making no such claims;
neither invoking a reified Society nor denying that the origins and indeed
persistence of 'societal facts' depended upon continuous human interac-
tion. Thus Gellner underlined that where properties of groups and
complexes are concerned, 'these latter can indeed exist only if their parts
exist - that is indeed the predicament of all wholes - but their fates qua
fates of complexes can nevertheless be the initial conditions or indeed the
final conditions of the causal sequence'.25 In exactly the same vein,
Mandelbaum maintained that 'one need not hold that a society is an entity
independent of all human beings in order to hold that societal facts are not
reducible to individual behaviour'.26 Although such statements clear
their advocates of reifying Society and also distance them from their
opponents' reductionist ontology, what they do not clarify is the precise
ontological status of'societal properties' themselves. Mandelbaum, after
seriously damaging the Individualist conception of social reality by
demonstrating the ineradicability of references to the social in description
and explanation, was clear that 'one's ontology must be accommodated to
the facts: the facts cannot be rejected because of a prior ontological
commitment'.27 Fair enough, for methodology should indeed regulate
ontology. However, this points to a different conception of social reality,
one which was not restricted to the Individual but never referred to the
Social Whole, one which accepted that 'societal facts' were activity-
dependent yet also maintained they were causally influential, autonomous
and pre-existent vis-d-vis individuals. But it was not forthcoming from
within Collectivism. The reason for this was the impossibility of substan-
tiating the existence of a societal property, 'of its own kind', within the
confines of an empiricist epistemology, where knowledge only comes
from sense-experience. Since it was not forthcoming, then the charge of
reification was repeatedly reiterated by Individualists, whilst Collecti-
vists did not articulate a new social ontology because hemmed in between
Holism and Empiricism.
25
Gellner, 'Holism versus individualism', p. 263.
26 27
Mandelbaum, 'Societal facts', p. 230. Mandelbaum, 'Societal facts', p. 232.
50 The problems of structure and agency

The empiricist barrier

We have already noted the conviction current in the 1950s to the effect
that there were only two types of entities which could contend to be
'moving agents in history', the human and the super-human, and that
these alternatives were held to be exhaustive. From this it was concluded
that since the latter entailed reiflcation, then the former was the only
claimant. Now, both to view them as exhaustive and to conclude that,
because of their observability, individuals were the only conceivable
'moving agents' (i.e. real and really causally efficacious) are twin products
of Empiricism. Basically, the Collectivist sought to deny that this
dichotomy was exhaustive and to show that the conclusion only followed
whilst ever the dichotomy was sustained. Instead, Collectivists rejected
both referents and argued for a third type of 'moving agent': 'societal
facts', referring to forms of social organization, to social institutions, to
persistent roles, that is to systematic and enduring relationships. These
were neither human nor inhuman in nature but relational, and relations
depended upon people but at the same time exerted an independent
influence over their activities. However, given such a relational concep-
tion, 'one can still legitimately ask what sort of ontological status societal
facts can conceivably possess if it is affirmed that they depend for their
existence on the activities of human beings and yet are claimed not to be
identical with these activities'.28 The question is answerable, but it cannot
be answered within the framework of empiricism. Moreover, Collecti-
vists were aware that the answer was 'emergent properties', for Mandel-
baum actually refers to 'existential emergents' and Gellner mentions the
'principle of Internal Relations' for explicating their inner constitution.
Significantly, both insights are confined to footnotes, conveying the
impression that to air them would invite a frosty reception, possibly
withering Collectivism's more modest methodological attack upon the
explanatory inadequacies of Individualism.
Most likely they were correct, for the notion of 'emergent properties'
depends upon overturning empiricism itself. Instead of a one-dimen-
sional reality coming to us through the 'hard-data' supplied by the senses,
to speak of 'emergence' implies a stratified social world including non-
observable entities, where talk of its ultimate constituents makes no sense,
given that the relational properties pertaining to each stratum are all real,
that it is nonsense to discuss whether something (like water) is more real
than something else (like hydrogen and oxygen), and that regress as a
means of determining 'ultimate constituents' is of no help in this respect
and an unnecessary distraction in social or any other type of theorizing.
28
Mandelbaum, 'Societal facts', p. 230.
 Individualism versus Collectivism 51
We would not try to explain the power of people to think by reference to
the cells that constitute them,
as if cells possessed this power too. Nor would we explain the power of water to
extinguish fire by deriving it from the powers of its constituents, for oxygen and
hydrogen are highly inflammable. In such cases, objects are said to have 'emergent
powers', that is, powers or liabilities which cannot be reduced to those of their
constituents ... Emergence can be explained in terms of the distinction between
internal and external relations. Where objects are externally or contingently
related they do not affect one another in their essentials and so do not modify their
causal powers, although they may interfere with the effects of the exercise of these
powers ... In the case of internally related objects, or structures ... emergent
powers are created because this type of combination of individuals modifies their
powers in fundamental ways. Even though social structures exist only where
people reproduce them, they have powers irreducible to those of individuals (you
can't pay rent to yourself)29
or swear fealty to yourself, or manumit yourself.
Therefore, to talk about 'emergent powers' is simply to refer to a
property which comes into being through social combination. These are
literally 'existential emergents'. They exist by virtue of inter-relations,
although not all relationships give rise to them. Thus, the increased
productivity of Adam Smith's pin-makers was a power emergent from
their division of labour (relations of production) and not reducible to
personal qualities like increased dexterity. Although he himself held that
this was also a side-effect, it did not account for the hundred-fold increase
in output (mass production) which was the relational effect of the time
saved in not picking up and putting down different tools, or manipulating
each pin through various angles and on different surfaces when making
one from start to finish. By contrast, the Ladies' Sewing Circle was
doubtless a social relationship but not one which generated the emergent
power of mass production, since each member confined herself to her own
work.
Just as the development of 'emergent powers' is nothing mysterious,
neither is there any mystery about their constituents and certainly no
invocation of dubious 'social substances':
The nature or constitution of an object and its causal powers are internally or
necessarily related: a plane canflyby virtue of its aerodynamic form, engines, etc.:
gunpowder can explode by virtue of its unstable chemical structure; multinational
firms can sell their products dear and buy their labour power cheap by virtue of
operating in several countries with different levels of development; people can
change their behaviour by virtue of their ability to monitor their own monitorings;
and so on.30
29
Andrew Sayer, Method in Social Science, Routledge, London, 1992, p. 119.
30
Sayer, Method, p. 105.
52 The problems of structure and agency

The existence of such causal powers has nothing to do with essentialism

for as the entity changes (through natural causes such as metal fatigue or
social causes such as a Third World embargo on multinational imports),
so the powers change because their internal relations have altered (or been
altered) in ways which nullify that which was necessary to the power in
question.
Such were the ontological implications of the insights which the
Collectivists already had, but failed to pursue. And their reason for this
was their full awareness that such efforts would come straight up against
the brick wall of empiricist epistemology. For 'societal facts' and 'emer-
gent properties' in general are incapable of being known via sense-data,
because as non-observables they cannot be 'pointed to' in the sense in
which we can point to material or organic objects, or to their qualities or
activities. Mandelbaum was conscious that, on this criterion, the argu-
ment would simply go round full circle:
Whenever we wish to point to any fact concerning societal organization we can
only point to a sequence of interpersonal actions. Therefore any theory of
knowledge which demands that all empirically meaningful concepts must ultima-
tely be reduced to data which can be directly inspected will lead to the insistence
that all societal concepts are reducible to patterns of individual behaviour. 31
Thus the problem of how to substantiate the existence of relational
properties appeared intransigent. Mandelbaum himself remained
stranded in the uncomfortable position of asserting that 'societal con-
cepts' could not be translated into individual terms without leaving an
irreducible societal remainder, whilst at the time bowing to empiricist
epistemology and advocating the necessity oi partial translations in order
to verify the concepts in question. Thus, 'It is always necessary for us to
translate terms such as "ideologies" or "banks" or "a monogamous
marriage system" into the language of individual thought and action, for
unless we do so we have no means of verifying any statements which we
may make concerning these societal facts.' 32
Yet as we noted earlier, Gellner had seen a way round this epistemolo-
gical difficulty, a method of securing the reality of relational concepts not
on the perceptual criterion of empiricism, but through demonstrating
their casual efficacy, that is employing a causal criterion to establish
reality. What precluded its exploitation was that the empiricist concep-
tion of causation, in terms of constant conjunctions at the level of
(observable) events, constituted another brick wall. The trouble with
'internally related structures' is that their powers may not always be
exercised because other contingencies intervene in society, which is
31 32
Mandelbaum, 'Societal facts', p. 232. Mandelbaum, 'Societal facts', p. 229.
 Individualism versus Collectivism 53
necessarily an open system and can never approximate to laboratory
conditions of closure. Because of this, 'emergent properties' will not
necessarily or usually be demonstrable by some regular co-variance in
observable events. Despite their roles, bank tellers sometimes hand over
money to masked men and ideologies may be masked by tokenism. In
other words, emergent properties rarely produce constant conjunctions
in society and therefore almost always fail to establish a claim to reality on
the empiricist criterion of causality.
Ironically, the notion of emergence was a defence against Holism which
came to grief on Empiricism. It was employed purely defensively to rebut
ontological objections, namely that references to societal facts or proper-
ties entailed reification, but it was never deployed in its own right for a
thorough-going reconceptualization of social structure. Its drastic and
premature limitation to this defensive role is starkly illustrated by
Goldstein's conclusion:
No sociological theory need make explicit reference to sociological emergence; its
usefulness is of another sort. When methodological individualists assail this or
that theory as holistic, when in fact it simply uses concepts that are not reducible
to individual dispositions, its defenders have always the possibility of pointing to
methodological emergence or some variation of it. That is, since the nature of the
criticism levelled against the theory is ontological rather than methodological,
sociological emergence offers a way of meeting it. It affirms that social scientists
may develop non-individualistic theories without being holists. And it has the
further advantage of forcing methodological individualists to defend their thesis
on methodological grounds. If non-individualist social science does not commit
untoward ontological sins, the methodological individualists are required to find
better grounds for its rejection. The doctrine that all explanation in social science
is ultimately in terms of individual dispositions is not established, indeed, in no
way supported, by the untenability of holism.33
There we have it all: the emergentist ontology relegated to the
background, invoked only to repulse charges of holistic reification and
thus to allow Collectivist explanations to continue to be advanced. In
short, the methodological game can go on, but only as a battle over the
proper form of sociological explanation, in a way which makes no explicit
reference to emergence!
Effectively what this does is to encourage Collectivists to go on playing
a game, defined in empiricist terms and according to its rules which means
that they can never win. On such terms there is no way in which they can
establish the reality of the explanatory concepts they adduce. As we have
seen, either they concede the necessity of 'partial translation' into
statements about individuals which re-shackles them to the empiricist
33
Goldstein, Two theses', pp. 281-2.
54 The problems of structure and agency

criterion of observability (and therefore does nothing to establish the

reality of their non-observable structural properties), or, if they appeal
instead to the causal criterion, emergent structural properties must fail
Hume's test for they do not manifest themselves in constant conjunctions
(they are incapable of predicting regularities at the level of events).
Consequently, at most, such properties can be inserted into explanations
when reduction fails, and the most that can be hoped for by Collectivists is
that this 'gives us some understanding of the unreduced concepts'.34 It is
hardly a confident expectation, because confronted with the same situa-
tion, the Individualist cherishes the opposite hope, namely that the
connections established between 'group variables' may 'suggest the
appropriate composition rules of individual behaviour'.35

Hume's heritage
In other words, Collectivists retreated to playing a methodological game
which could never establish their ontological claims. They thus became
closet emergentists but explanatory game players and in the process the
emergent social structure, to which no 'explicit reference' was made,
underwent further diminution. Once again, methodology reacts back to
regulate ontology, in this case fragmenting structure into a series of
discrete properties rather than allowing social structure to be considered
as a distinct stratum of social reality and explored as such. It enters
explanations as a set of social features adduced on an ad hoc basis when
explanation cannot do without them, thus serving to occlude the systema-
tic nature of social structure. But the effects go deeper still, for what now
governs even its ad hoc admission is none other than the Humean model of
causation itself! For structural features are allowed in under the rubric of
(as yet) 'undefined group properties' provided they increase our explana-
tory/predictive power by helping to account for observed regularities. It
is its contribution to accounting for a constant conjunction which gives a
structural property its right of entry. Yet most of the time, in open social
systems, regularities at the level of events are just what emergent features
do not generate. Therefore, the structural elements which can pass the
Humean check-point, only do so on an ad hoc basis but are also atypical 'of
their own kind'! In practice, they are those which approximate to
observability and are in play because of their descriptive indispensability.
Thus, for example, the type of electoral system (proportional represen-
tation or first-past-the-post) will be needed to explain the kind of

34
Gellner, 'Holism versus individualism', p. 255n.
35
Brodbeck, 'Methodological individualisms', p. 303.
 Individualism versus Collectivism 55
government to emerge from any election, in addition to statements about
people's political dispositions, which in turn are only identifiable in
relation to Political Parties. These two structural properties earn their
ticket and the Collectivist might even manage to suggest that voters are in
a (pre-structured) situation where their Party affiliation is affected by the
electoral system (i.e. supporting a minority party makes more sense under
proportional representation). However, what Humean gatekeeping will
preclude are propositions about the prior distribution of power having
affected the electoral system in operation, the Parties in existence, let
alone the political dispositions of voters themselves. Yet there may well be
internal and necessary relationships obtaining between all four elements.
Another way of putting this is that certain emergent effects may get
through the gate, but no emergent mechanism will. Included purely
insofar as they boost predictive power, some structural factors can be
added to statements about individuals to improve the correlation coeffi-
cient. In this way, all that is asserted is that the two together yield better
predictions. What cannot be asserted or even explored in terms of
constant conjunctions is how the explanatory factors interact together to
generate a given outcome. The explanatory formula is 'individual dispo-
sitions' plus some indispensable 'structural property', where the 'plus' is
predictive rather than real (i.e. two independent factors which together
predict better than one alone, rather than as inter-dependent variables).
Consenting to play a purely methodological game according to
Humean rules gradually undermines the Collectivist programme. We
have just charted the fragmentation of structure into disparate 'factors'
and indicated that it is immediately followed by the exclusion of the
interplay between 'structure and agency'. Yet this interaction had been
just what early emergentists looked towards and saw profit in social
theory exploring. Mandelbaum had argued that to hold 'that societal facts
are not reducible without remainder to facts concerning the thoughts and
actions of specific individuals, is not to deny that the latter class of facts
also exists, and that the two classes may interact'.36 Moreover he had begun
to spell out how they do so, by sketching in exactly the kind of mechanism,
or still better process, which constant conjunctions literally cannot ack-
nowledge (for to Hume all we can ever say is that (a) and (b) are regularly
observed to coincide). On the contrary, Mandelbaum proposed that 'if we
wish to understand many of the dilemmas by which individuals are faced,
we can do no better than hold to the view that there are societal facts which
exercise external constraints over individuals no less than there are facts
concerning individual volition which often come into conflict with these

36
Mandelbaum, 'Societal facts', p. 234.
56 The problems of structure and agency

constraints'. 37 Finally, if this crucial interplay is written off the agenda,

two other elements are lost with it.
The first is any notion of the 'structuring of agency', that is the
processes by which our necessary involvement in society (as opposed to
our equally inescapable involvements with other people), help to make us
the kind of social beings we are, with the dispositions we possess and
express. For the Collectivist was surely right that, for instance, an attitude
of political disillusionment can be engendered by such things as a
succession of coalition governments locked in immobility which were
produced, in part at least, by proportional representation systems.
Instead of the re-conceptualization of agency to which this points, we are
left with 'the individual' plus some 'structural factor' needed for
enhanced prediction and can only combine them for purposes of correla-
tion, but cannot investigate the processes of their combination in the real
world.
Secondly, since process in general is off the Humean agenda, then the
strange and undesirable situation arises in which a given 'structural
property' may permissibly figure in an explanation, yet the processes
through which it emerged cannot be captured within the same explana-
tory framework. Regrettably then, the strenuous policing of which
'structural properties' might appear in explanatory statements (those
which improved predictive power) also prevented any explanation of
their own origins (interaction in a prior social context) and their mode of
influence (through structuring the context of current interaction). By
entering 'factorially' into explanations, it was allowed that these frag-
mented aspects of social structure co-determined outcomes (along with
individuals), but never that they did so by a process of working through
people - shaping the situations they confronted, furnishing beliefs for
their interpretation, or distributing different vested interests to them in
maintaining or transforming the status quo. Instead, they remained
'undefined', unexplored and unlinked (to one another or to agents): only
their deterministic effects in accounting for regular social outcomes was
upheld.
On Humean terms, such 'structural properties' as earned their keep
remained both unduly mysterious and inexplicably powerful. Ironically,
then, positivism served to retain them as something much more akin to
Holistic factors (of unexplicated provenance and deterministic conse-
quence) than had ever been the wish of Collectivists. Not surprisingly,
many of those who found the parameters of the Humean game unduly
restrictive sought stronger beer in unabashed Holism itself- structura-

37
Mandelbaum, 'Societal facts', p. 234.
 Individualism versus Collectivism 57

lism, structural functionalism and structuralist marxism. In short, the

failure of Collectivism to articulate an alternative social ontology to
Individualism, and the Collectivist 'retreat' to defending what they could
of their methodological ground, served both directly and indirectly to
foster 'downwards conflation' in practical social theory - be it through
those acceding to positivism and according 'structural properties' a
deterministic influence in the regular occurrence of events, or through
kicking over the empiricist traces to become Holistic recidivists.

Contesting the terms of the traditional debate

Chapter 1 began by stressing the tripartite relationship between ontology,
methodology and practical social theory. Since none is dispensable, then
each has to be adequately conceptualized in itself and consistently related
to the others. In turn, this means that we are dealing with their mutual
regulation and matters can only go astray if what should be aflexibletwo-
way relationship is rigidly conceived of as uni-directional. This was the
purpose of going over the ground of the old debate between Individualists
and Collectivists, for both programmes illustrate the deficiencies of one-
way approaches.
Thus, Individualists began from an unshakeable ontological commit-
ment that the ultimate constituents of social reality were 'individuals',
formulated their methodological injunctions on this basis, yet were
unwilling to make ontological adjustments in the light of the unworkabi-
lity of their own methods and the findings of others who did not share
their commitment to the necessity of reductionism. By contrast, Collecti-
vists started from an equally strong methodological conviction that facts
about the social context could neither be excised from the description or
explanation of our subject-matter, but failed to ground this in a concep-
tion of social reality which both avoided any taint of Holism and evaded
the strictures of empiricism.
The inability of either Individualism or Collectivism to establish a
convincing, consistent and working relationship between social ontology
and methodology can be laid firmly at the door of empiricism itself. For it
fortified Individualists in the belief that since they were ontologically
secure, then their methods must work 'in principle', despite all evidence
to the contrary. Simultaneously, it undermined Collectivist confidence in
their methodological 'success' by querying the reality of their explanatory
variables, which never could be validated in empiricist terms.
The implications for practical social theorizing were equally unsatis-
factory. However implicit they may be, no social theory can be advanced
without making some assumptions about what kind of reality it is dealing
58 The problems of structure and agency

with and how to explain it. All social theory is ontologically shaped and
methodologically moulded even if these processes remain covert and
scarcely acknowledged by the practitioner. This is inescapable because
theories logically entail concepts and concepts themselves include certain
things and exclude others (at the methodological level) and denote some
aspects of reality whilst denying others (at the ontological level). Any who
think they can avoid both fall into the trap of instrumentalism: those
believing that the use of 'heuristic concepts' in explanation saves them
from making any ontological commitment fail to recognise that terming
something 'heuristic' is itself a matter of ontology.
Yet the concepts on offer from Individualism and Collectivism were
fundamentally unsatisfactory. Individualism supplied an unacceptably
atomistic concept of the individual, shorn of any relationship with the
social context yet inexplicably bulging with social attributes; a conception
of the social structure as a mere aggregate of individual activities whose
every tendency was the responsibility of current actors, plus the unwork-
able method of reduction as the means for linking 'structure and agency'.
On the other hand, Collectivists proffered a fragmented conception of
structure, defined residually as that which defied reduction, an equally
fragmentary concept of agency represented by individuals plus their
social context, and they refrained from specifying the processes linking
the two together. Insofar as working social theorists took Individualist
concepts on board, this served to perpetuate the fallacy of upwards
conflation in social theorizing. If they drew upon Collectivism instead,
then the missing two-way link between structure and agency continued to
foster the equally fallacious form of downwards conflation in social
theory.
Of course much of this went on in the state of inarticulate unawareness
and often consisted in practical analysts cutting their theoretical cloth to
suit their coat or vice versa. 38 Thus, at one extreme interpretative
sociologists undertook small-scale interactional studies and simply
placed a big etc. after them, implying that the compilation of enough
sensitive ethnographies would generate an understanding of society by
aggregation. At the other, large-scale multivariate analyses pressed on
towards some predictive goal without reference to the interactional
processes generating their variables. However, it has already been
stressed that the scope of the problem or size of entity is not what actually
38
Tactual trends may certainly be detected with respect to the preferred, strategic field of
empirical inquiry. In particular, those who focus on small groups, or microsociological
phenomena, are more often than not reductionistically orientated, and those who study
the comprehensive historical processes, or macrosociological phenomena, tend toward
antireductionistic interpretations'. Piotr Sztompka, Sociological Dilemmas. Academic
Press, New York, 1979, p. 92.
 Individualism versus Collectivism 59
differentiates between Individualism and Collectivism; to the former, the
macroscopic is just the 'large group'; to the latter, a dyad like husband and
wife or doctor and patient is unidentifiable without reference to the social
context. Thus the above connections were ones of superficial theoretical
affinity, but once forged the concepts used then transmitted their own
deficiencies into practical theorizing. Alternatively, in some specialisms,
theorizing would begin on the basis of concepts taken from one camp,
realize the limitations of the concepts, and then swap to the other camp,
only to repeat the process. Thus, for example, the 'old' sociology of
Education (Collectivist) gave way to the 'new' (Individualist), eventually
leaving practitioners calling for synthesis.
Yet as we have seen, synthesis or compromise is the one deal which
cannot be struck, which is why I have continuously resisted the notion of a
via media between the two programmes, consisting of conceding Descrip-
tive Individualism to the Individualist and Explanatory Emergence to the
Collectivist. Further modifications or revisions, such as 'situational
individualism', undertaken with the same conciliatory aim in view, have
not been discussed, because like the via media they fail - as they must - to
reconcile contradictory premises. I have stuck to the pure lines of the
debate, as articulated in the 1950s because if, as I maintain, there are
intimate and indissoluble connections between ontology, methodology
and practical social theory, then this is what we have been stuck with ever
since - a choice between the two alternatives, replete with their deficien-
cies which are merely replicated at the practical level, which ever is
chosen. This was the reason for saying 'don't choose', but it was almost
impossible advice to follow when positivism was in full flood and
empiricism itself was responsible for the intrinsic defects of the only two
options available.
Only after the empiricist hegemony had been challenged and the
closely associated domination of positivism had been similarly under-
mined did siding with neither Individualism nor Collectivism become a
genuine option. For with the progressive demise of empiricism, not only
were the terms of the old debate between them rejected, but the debate
itself was re-cast in entirely different ones. These transcended the original
antinomy between the 'study of wo/man' and the 'science of society' by
re-conceptualizing 'structure' as intimately rather than truistically
'activity-dependent' and the 'individual' as intrinsically rather than
extrinsically the subject of 'social constitution'.
What did not disappear, despite the vastly premature celebration of a
new consensus by many commentators, was the enduring necessity of
making a choice. For the new terms in which 'structure and agency' were
re-conceptualized and linked together were again represented by two
60 The problems of structure and agency

standpoints, thus opening up a new debate beginning in the seventies or

early eighties. These I have termed 'Elisionism' (because transcending
the dualism between individual and society consisted in replacing it by an
insistence upon their mutual constitution), and 'Emergentism' (because
structure and agency are both regarded as emergent strata of social reality
and linkage consists in examining their interplay).
The first manifestations of Elisionism in social theory were distinctly
idealist. Neo-phenomenological forms of theorizing construed the social
context as 'facticity' rather than fact and insisted upon its 'externalization'
and 'objectification' rather than allowing it externality and objectivity.
However, in viewing entities such as social institutions as purely dramatic
conventions which depended upon co-operative acts of agents in sustain-
ing a particular definition of the situation, Symbolic Interactionists in
particular elided 'structure' and 'agency' in three key ways which have
increasingly come to characterize Elisionism as a distinctive theoretical
orientation: (i) a denial of their separability, because, (ii) every aspect of
'structure' is held to be activity-dependent in the present tense and
equally open to transformation, and (iii) the conviction that any causal
efficacy of structure is dependent upon its evocation by agency.
Because of the centrality of'inseparability', such premisses are neither
reductionist {contra Individualism), nor anti-reductionist {contra Collec-
tivism). Whilst the untrammelled idealism, characteristic of interpreta-
tive sociology in the seventies, is no longer the hallmark of those viewing
structure and agency as mutually constitutive, the fundamental insepara-
bility of the two is what constitutes Elisionism as a distinctive approach.
Those now endorsing the 'duality of structure' as the medium and
outcome of social practices, under the rubric of Structuration theory,
have reconstituted Elisionism on a more acceptable basis (which incor-
porates material resources and power rather than dealing with networks
of meanings alone), whilst continuing to endorse inseparability and its
associated premisses. In contradistinction, the very notion of 'emergent
properties' which are generated within socio-cultural systems is necessar-
ily antithetic to the tenet of inseparability because such structural and
cultural features have autonomy from, are pre-existent to, and are
causally efficacious vis-d-vis agents - their existence, influence and
analysis therefore being incompatible with the central premises of
Elisionism.
Consequently choice is inescapable because 'Elision' (the term used for
those grouping themselves around Structuration theory) and 'Emer-
gence' (those exploring the interface between transcendental realism and
social theory) are based upon different ontological conceptions, related to
disparate methodological injunctions and thus have quite distinct impli-
 Individualism versus Collectivism 61
cations for practical social theorizing. To celebrate the development of a
new consensus is to concentrate upon their common rejection of the terms
of the old debate whilst ignoring the different bases upon which the two
re-set the terms. The unpopular message of this book is that the burden of
choosing has not been removed - and we can only make a sensible choice
by closely scrutinizing the nature of and connections between ontology -
methodology - practical social theory which Elisionists and Emergentists
respectively endorse. This is exactly what will be done: it is undoubtedly
more burdensome than the conclusion that we can have the best of both
worlds, but it is preferable to recognize in advance that again there can be
no via media than to find it collapsing under us later on.
Let us briefly introduce the two new standpoints whose relative merits
will be examined in the course of the next three chapters - and the reasons
for the choice which is made between them here. On the one hand, the
Elisionists' new 'ontology of praxis' seeks to transcend the traditional
debate through replacing the two sets of terms in which it was conducted
by their notion of'the duality of structure', in which agency and structure
can only be conceptualized in relation to one another. From this, it follows
methodologically that neither the reductionism advocated by Individual-
ist nor the anti-reductionism defended by Collectivists can play any part
in the Elisionists' approach to explanation - which takes up the novel
position of areductionism. This is the direct logical consequence of their
re-defining structure and agency as inseparable. Whilst this frees both
from being an epiphenomenon of the other, it does so by holding them to
be mutually constitutive. In turn it will be maintained that although the
implication of this is a rejection of both upwards and downwards
conflation in social theorizing, its consequence is actually to introduce a
new variant - central conflation - into social theory.
On the other hand, the realist ontology of the Emergentists is deployed
to furnish that which Collectivism lacked, an activity-dependent concept
of structure, which is both genuinely irreducible yet in no danger of
hypostatization, and a non-atomistic conception of agents, to rectify the
deficiencies of Individualism's individual - without, however, regarding
the two elements as part of an inseparable 'duality'. Instead, because
Realists endorse the existence of irreducible 'emergent properties', they
advance a much more robustly stratified view of both society and people
and hence resist central conflation which is the expression of Elisionism in
social theory.
Emergentists' combined repudiation of both reductionist and confla-
tionary theorizing means a principled avoidance of the epiphenomena-
lism which is embedded in Holism and Individualism, where 'agency'
and 'structure' respectively become inert as wholly dependent features -
62 The problems of structure and agency

consequently, introducing downwards and upwards conflation into social

theorizing. It also constitutes a principled departure from the 'duality of
structure' by which 'structure' and 'agency' are inextricably compacted
by Elisionists. In place of all three forms of conflationary theorizing, the
Emergentist substitutes analytical dualism. Because the social world is
made up, inter alia, of 'structures' and of 'agents' and because these
belong to different strata, there is no question of reducing one to the other
or of eliding the two and there is every reason for exploring the interplay
between them.
These differences between the Elisionists and Emergentists have often
been obscured by their common rejection of the terms of the traditional
debate, but what the two replace them by is grounded in antithetical
conceptions of social reality - precisely because Structuration theorists
explicitly disavow emergence itself. Thus Ira Cohen underlines that
'structures' are 'properties of systems that do not "emerge"' and states:
To affirm that enduring properties of collectivities are embedded in disappearing
and reappearing practices and relations both clarifies and demystifies the ontolo-
gical obscurities associated with emergence. In particular it is no longer necessary
to pose the uncomfortable question of how emergence actually occurs: a question
which no collectivist theorist, to my knowledge has answered in a persuasive
fashion.39
Such a viewpoint stands in the starkest contrast with the Realist assertion
that 'it is just in virtue of these emergent features of societies, that social
science is possible'. 40
Obviously there is an onus upon those of us who uphold the latter view
to clear up the 'ontological obscurities' which 'sheepish' Collectivists did
leave unresolved when they defended explanatory emergence (but failed
to ground it in a non-empiricist conception of social reality). T h e
contributions of transcendental realists over the last ten to fifteen years
have served to clarify these residual obscurities: the development of the
morphogenetic/static approach now provides an account of 'the occur-
rence of emergence' which complements the realist social ontology with a
working methodology. Together they insist upon the activity-depen-
dence of emergent properties, in their origins as in their influences.
Equally, they claim that this does not mean generative activities and
emergent consequences have to be treated as inseparable; on the contrary
they firmly uphold the possibility and utility of distinguishing between
them.
39
Ira J. Cohen, 'Structuration theory and social order: five issues in brief, in J. Clark, C.
Modgil and S. Modgil (eds.), Anthony Giddens: Consensus and Controversy, Falmer
Press, Basingstoke, 1990, p. 42.
40
Roy Bhaskar, The Possibility of Naturalism, Harvester, Hemel Hempstead, 1979, p. 25.
 Individualism versus Collectivism 63

Thus in the new conspectus which I have called Emergentism (and can
now be seen to be defined here as a realist ontology and a morphogenetic
methodology), it is vital to distinguish between
such causal inter-dependency, which is a contingent feature of the process
concerned, from existential intransitivity, which is a priori condition of any
investigation ... For although the processes of production may be interdepen-
dent, once some object ... exists, if it exists, however it has been produced, it
constitutes a possible object for scientific investigation.41
A realist ontology which upholds transfactual structures and intransi-
tive cultural properties, and encourages their investigation as emergent
entities, is thus at variance with the Elisionists' view which holds, (a) that
such properties only possess a 'virtual existence' until, (b) they are
'instantiated' by actors, which (c) means these properties are neither fully
real nor examinable except in conjunction with the agents who instantiate
them, and only then through an artificial bracketing exercise since the two
are inseparable in reality.
In conclusion, their consistent insistence upon the differentiation and
stratification of the social world leads Emergentists to separate 'parts' and
'people' in order to examine their distinctive emergent properties. As
Bhaskar noted of Peter Berger's early and idealist version of an elisionist
theory, its fundamental error is that 'People and society are n o t . . . related
"dialectically". They do not constitute two moments of the same process.
Rather they refer to radically different things'. 42 Precisely the same
criticism can be levelled at later versions like structuration theory, which
repeats this 'fallacy of the two moments', and will only entertain 'unack-
nowledged conditions of action', withholding the status of emergent
properties from them by rendering them merely matters of 'knowledgea-
bility' on the part of agents.
Hence, the separability/inseparability issue represents the ontological
parting of the ways between Emergentists and Elisionists. For the
Emergentist,
The importance of distinguishing, in the most categorical way, between human
action and social structure will now be apparent. For the properties possessed by
social forms may be very different from those possessed by the individuals upon
whose activity they depend ... I want to distinguish sharply then between the
genesis of human actions, lying in the reasons, intentions and plans of human
beings, on the one hand; and the structures governing the reproduction and
transformation of social activities, on the other.43
Why? Not simply because ontologically they are indeed different
41 42
Bhaskar, Naturalism, p. 47. Bhaskar, Naturalism, p. 33
43
Bhaskar, Reclaiming Reality, Verso, London, 1989, p. 79.
64 The problems of structure and agency

entities with different properties and powers, but because methodologi-

cally it is necessary to make the distinction between them in order to
examine their interplay and thus be able to explain why things are 'so and
not otherwise' in society.
This interplay between the two is crucial for effective theorizing about
the social world, whether our concern is with everyday personal dilemmas
or with macroscopic societal transformations. Yet the Elisionists insist-
ence upon 'inseparability' precludes just that examination of the interface
between structure and agency upon which practical social theorizing
depends. From the standpoint of Elisionism it becomes impossible to talk
about the stringency of structural constraints versus degrees of personal
freedom, for in theories based upon central conflation, causation is always
the joint and equal responsibility of structure and agency and nothing is
ever more attributable to one rather than the other, at any given point in
time.
The central argument of this book is just the opposite. It is only
through analysing the processes by which structure and agency shape and
re-shape one another over time that we can account for variable social
outcomes at different times. This presumes a social ontology which
warrants speaking about 'pre-existence', 'relative autonomy' and 'causal
influence' in relation to these two strata (structures and agents) and an
explanatory methodology which makes such talk practicable for the
practising social theorist.