Back in 1982, a brief but brusque exchange … took place between James
Tobin … and Robert Nozick … Tobin exclaimed at Nozick: ‘There is nothing
more dangerous than a philosopher who’s learned a little bit of economics.’
To which Nozick immediately responded: ‘Unless it’s an
economist who hasn’t learned any philosophy.’
Terence Hutchison, ‘On the Relations Between Philosophy and
Economics’ (1996)
What distinguishes a human agent from automata or insects is the developed
capacity to reflect and deliberate upon the context, options, purpose
and possible outcomes of action. As Karl Marx (1976, p. 284) wrote in Capital:
‘what distinguishes the worst architect from the best of bees is that the
architect builds the cell in his mind before he constructs it in wax’. This
does not imply that all human behaviour is deliberate, but that human deliberation
is possible. We should also acknowledge that some non-human
animals might have very partially developed this capacity.
Alongside the concept of the agent, the concept of structure is central
and essential to any viable social science. A social structure is a set of significant
relations between individuals that can lead to causal interactions. Social
structures can involve rules, norms, meanings, communication and
much else. These relations can be acknowledged or unacknowledged by
the individuals involved. Furthermore, social structures can survive the
demise of particular individuals that once related to them. Accordingly,
the study of human social systems is more than the study of human individuals,
because society embodies relations and properties in addition to
those of individuals themselves. Although structures frame and condition
behaviours, they are neither reducible nor ontologically equivalent to
them.
Broadly, the idea that society is more than a collection of individuals has
a long pedigree. Indeed, it is much older than the individualistic notion
that society is merely the sum of its members. Such atomistic conceptions
2 Agency and structure
date largely from the European Enlightenment of the late seventeenth and
eighteenth centuries. Before that, the individual was generally regarded as
being part of, or subordinate to, some greater entity or whole. For example,
there is an ancient metaphor, traceable at least back to Plato and the Bible,
that society is like a living organism in which the individual is a component.
At the beginning of modern economics, the French physiocrat
François Quesnay made use of the metaphor of blood circulating in the
structure of the economic body (Foley, 1973). When Adam Smith wrote of
the ‘invisible hand’ he was clearly suggesting that socio-economic systems
have additional properties that are not reducible simply to the efforts of
the individual minds and visible hands within them.
As modern social science developed in the nineteenth century, the idea
of social structure strengthened and evolved. It found a prominent exponent
in Marx, who fastened upon the architectural metaphors of structure
and superstructure. Marx (1971, p. 20) wrote in 1859 in his famous Preface
to A Contribution to the Critique of Political Economy:
In the social production of their existence, men inevitably enter into
definite relations, which are independent of their will, namely relations
of production appropriate to a given stage in the development of
the material forces of production. The totality of these relations of production
constitutes the economic structure of society, the real foundation,
on which arises a legal and political superstructure and to which
correspond definite forms of social consciousness.
Law and politics thus rested on the ‘economic’ structure. While Marx
failed to define what he meant by ‘economic’ here, in his writings the concept
of a structure reached a high point of development. Instead of understanding
social reality simply in terms of the wills and personalities of the
individuals involved, his concept of structure hinted at the powerful interests,
incentives and institutions that might constrain or mould individual
human agency.
The idea of social structure also developed within the writings of the
German historical school of economists. Some of the nineteenth-century
German historicists used the old idea of society as an organism to connote
the existence of social structure above constituent individuals (Hutter,
1994; Hodgson, 2001c). There the structure metaphor often assumed a biological
form, like the physiology of an organism. Herbert Spencer (1877)
likewise embraced an explicit concept of social structure and also described
society as an organism. However, Spencer (1881, pp. 48–9) believed
that ‘the character of the aggregate is determined by the characters
of the units’ and that social structures emerge because of ‘a proclivity towards
the structure’ in the ‘substance’ of each individual. Hence Spencer’s
notions of social structure or social organism did not amount to much
more than the aggregate of individual attributes.
Agency and structure 13
The French sociologist Émile Durkheim went much further, to emphasize
that society is more than the sum of its parts. Durkheim developed
his own distinctive argument that social facts were not reducible
merely to individuals or their psychology. A concept of social structure
is also traceable in the writings of Max Weber, but there it does not play
such a strong and embracing role as in the works of Durkheim or Marx.
In some discourses the essential issues are addressed using different
terms. A specific type of social structure, described as an institution, pervades
the writings of institutional economists such as Thorstein Veblen,
John R. Commons and Wesley Mitchell. A broad and inclusive conception
of institutions is widely accepted by social scientists and is adopted here.
We may define institutions broadly as durable systems of established and
embedded social rules that structure social interactions.
The term ‘rule’ is broadly understood as an injunction or disposition,
that in circumstances X do Y.1
Hence it includes norms of behaviour and
social conventions, as well as legal or formal rules. By their nature, institutions
must involve some shared conceptions, in order to make rules operative.
According to this definition, systems of language, money, law,
weights and measures, traffic conventions, table manners, firms (and all
other organizations) are all institutions.2
As Alan Wells (1970, p. 3) put it: ‘Social institutions form an element in a
more general concept, known as social structure.’ The original institutional
economists understood institutions as a special type of social structure
with the potential to change agents, including changes to their purposes or
preferences.3
I have described this possibility as ‘reconstitutive downward
causation’ and attempted to specify its causal processes (Hodgson, 2002a,
14 Introduction
1 Rules are not necessarily explicit. As M. Weber (1978, p. 105) pointed out in 1907, rules are
often followed ‘without any subjective formulation in thought of the “rule”’. Hayek (1967,
pp. 66–7) also emphasized non-articulated rules: ‘it should be clearly understood that the
term “rule” is used for a statement by which a regularity of the conduct of individuals can
be described, irrespective of whether such a rule is “known” to the individuals in any
other sense than they normally act in accordance with it.’ However, Hayek (1979, p. 159)
overly extended the term ‘rule’ to instincts, which are not necessarily culturally or socially
embedded, as required in my definition of an institution. Notably, all culturally
embedded rules are in principle codifiable, in which case breaches of rules can be more
readily detected. In which case, mental representations of rules become significant (Searle,
1995). See Ostrom (1986) and Crawford and Ostrom (1995) for detailed analyses of the
nature of institutional rules.
2 Note that Schmoller (1900, p. 61) defined an institution similarly as ‘a partial order for
community life which serves specific purposes and which has the capacity to undergo
further evolution independently. It offers a firm basis for shaping social actions over long
periods of time; as for example property, slavery, serfhood, marriage, guardianship,
market system, coinage system, freedom of trade.’ (Translated and quoted in Furubotn
and Richter, 1997, p. 6.)
3 However, some institutionalists such a J. F. Foster (1981, p. 908) have misleadingly
defined institutions as ‘prescribed patterns of correlated behavior’. T. Lawson (2003a, pp.
189–94) rightly points out a difficulty in this conception: if institutions are behaviour, then
how can changes in behaviour be explained? A related argument was devised by Aristotle
in his critique of ‘the Megaric view’. See footnote 29 on p.170 below.
2003b; Hodgson and Knudsen, 2004). These ideas will be addressed and
developed in the present work.
Although it is familiar to Marxists and modern sociologists, the use of
the term ‘social structure’ has not been universal in the social sciences as a
whole. Its use, for much of the nineteenth and twentieth centuries, was not
that frequent, and the term was rarely defined. During the twentieth century,
it was largely within the rising discipline of sociology that the term
began to be used more frequently. It was prominent in the work of some
early American sociologists, such as Lester Frank Ward (1903), but others
made less use of the term.
The American sociologist Talcott Parsons inherited a concept of structure
principally from European forerunners – including Marx and
Durkheim – and it retained a pivotal role in his theory of social action (Parsons,
1937). It was then developed by Robert Merton (1949) and in the
‘structural sociology’ of Peter Blau (1975) and others. Structure was given
special attention by those influenced by Marxism, such as the Frankfurt
School in Germany, and by the schools of structuralism in France founded
by the anthropologist Claude Lévi-Strauss and by the Marxist theorist
Louis Althusser.
It is disturbing to note that the agency–structure problem is evaded by
some fashionable developments in contemporary social theory. In particular,
there have been recent attempts by post-modernists and poststructuralists
to dismiss or transcend this issue. In response, Nicos
Mouzelis (1995, pp. 69–70) rightly assessed ‘attempts to dismiss the
agency–structure distinction ... either by conflating the two notions, or by
... deriving the one from the other’ as leading to a theoretical impasse. Notably,
many of these evasive attempts involve ‘the reintroduction of the
distinction by the back door ... by keeping the logic of the agent–structure
dichotomy while expressing it through a different terminology’. The solution
to the agency–structure problem is not to walk away from it, or to pretend
it does not exist. Such strategies have notably ended up with the
readmission of the problem in another form.
At the risk of oversimplifying matters, we may classify several types of
treatment, or evasion, of the problem. The first group includes those approaches
that claim that individuals are the ultimate explanatory or ontological
elements. As noted in the next section, this ‘methodological
individualism’ can itself be subdivided. A second group reverses the conflation:
structures are regarded as the ultimate explanatory units. A succeeding
section is devoted to this ‘methodological collectivism’. A third
group attempts what Margaret Archer (1995, p. 61) criticizes as a ‘central
conflation’, by erecting a concept of ‘structuration’ that encompasses both
structures and agents. This in some ways represents an advance on the
preceding two positions. A fourth group shares with structuration theory
the idea that agent and structure are mutually constitutive of each other.
But in contrast to structuration theory, it disassociates agent from structure
Agency and structure 15
by insisting on their differences. However, this group limits the acknowledgement
or explanation of their causal interaction in a manner explained
below.
The fourth approach is more sophisticated than the preceding three. A
primary aim of this book is to indicate that an institutional economics that
builds on Darwinian and Veblenian ideas holds the promise of a fifth, superior
approach. This would be both non-conflationary and causally interactive,
in a fuller sense to be explained and explored at length. The key
innovations in the fifth approach are to extend the requirements of causal
explanation, and to place the issue in an over-arching evolutionary frame-
work.4
Methodological individualism
There is a long-standing tradition in social theory, to attempt to explain social
structures, institutions, and other collective phenomena, solely in
terms of the individuals involved. This approach is promoted with mainstream
economics and elsewhere. Some, even sociologists, simply take for
granted ‘the methodological individualism of scientific practice’ (Lopreato
and Crippen, 1999, p. 209) without even defining the phrase. But the substance
and validity of ‘methodological individualism’ is widely contested.
There is not even strict agreement on the definition of this term (Udéhn,
2001).5
Broadly, methodological individualism emphasizes the human agent
over social structures. Ludwig Lachmann (1969, p. 94) asserted that methodological
individualism means ‘that we shall not be satisfied with any
type of explanation of social phenomena which does not lead us ultimately
to a human plan’. But very few social scientists would deny the role of individual
intentions in the explanation of social phenomena. In another attempt
Jon Elster (1982, p. 453) defined methodological individualism as
‘the doctrine that all social phenomena (their structure and their change)
are in principle explicable only in terms of individuals – their properties,
goals, and beliefs’. Being less banal, this definition also is insufficiently
precise, as it fails to clarify whether interactions between individuals or social
structures are ‘properties … of individuals’ or not. If individual
16 Introduction
4 As well as the term ‘evolutionary’, the notion of ‘causal explanation’ itself requires
specification and clarification. For useful accounts see Bunge (1959) and Lipton (1991).
Some further remarks on causality are found below.
5 Schumpeter (1908, pp. 64–8, 77–9, 85–7, 154–5, 261, 541–7) first coined the term
‘methodological individualism’. However, Schumpeter did not argue that all
explanations in social science must necessarily and exclusively be in terms of individuals.
Instead, he argued that it was the role of economics to start from given, rational individuals.
For Schumpeter, this explanatory constraint was optional, once we moved outside ‘pure
economics’. As Udéhn (2001) shows, Schumpeter’s use of the term is far from universal.
The definition adopted in the present work is in terms of a general methodological
injunction, rather than Schumpeter’s methodological option or disciplinary demarcation
criterion.
interactions or social structures are not ‘properties of individuals’, then
this narrower and more meaningful notion of methodological individualism
must be deemed inoperable, for the reasons given below.6
Much of the confusion in the debate over methodological individualism
stems from whether methodological individualism means explanations (a)
in terms of individuals alone, or (b) in terms of individuals plus individual
interactions or social structures. If it were meant to mean (b), then few
would disagree. Such an inclusive notion would not warrant the title of
methodological individualism any more than the description ‘methodological
structuralism’. If social structures or interactions between individuals
are also an essential part of the doctrine, then it is misleading to give
the individual exclusive representation in the label.
Attempts to conflate socio-structural phenomena upon the individual
generally flounder. There are three types of problem involved here, depending
on the type of argument and version of methodological individualism
involved. The first type of problem results from giving the
individual too much of the explanatory burden. For example, Stephen
Jones (1984) and Ekkehart Schlicht (1998) provide interesting theories of
conformism and custom. However, the explanation of the emergence of
customs and conventions depends crucially on an assumption that individuals
exhibit ‘rule preference’ or a ‘preference for conformism’. The
problem of institutional emergence is thus ‘solved’ by making properties
of institutions also the properties of individual preferences. In a manner
reminiscent of Spencer’s idea (1881, pp. 48–9) that social structures emerge
because of ‘a proclivity towards the structure’ in each individual, the social
phenomena are conflated upon the individual. The explanation carries
force only because individuals have been obliged to take on board factors
that properly relate to social structure. Crucially, what such theories do not
explain is how individuals acquire these socially infused preferences.
There are many other examples of a similar conflation of social structures
and their effects upon individuals. Howard Margolis (1982) and Kenneth
Koford and Jeffrey Miller (1991), view institutions as resulting from
features of individual preferences, such as for cooperation, for altruism, for
conformism, or for the observance of social norms. Robert Frank (1988)
and Amitai Etzioni (1988), emphasize individual emotions or add moral
Agency and structure 17
6 Claimed supporters of some version of methodological individualism include Popper
(1945, 1960), Hayek (1943, 1948, 1952b), von Mises (1949), Arrow (1968, 1994), Ghiselin
(1974), Boudon (1981), Elster (1982), Coleman (1990), Furubotn and Richter (1997) and
Schlicht (1998). Critics include Lukes (1973), Giddens (1984), Hodgson (1988), Bhaskar
(1989), Douglas (1990), Kontopoulos (1993), Archer (1995), T. Lawson (1997, 2003b),
Storper and Salais (1997), Bunge (1998) and Udéhn (2001). In apparently conciliatory
statements, Sober (1981) and Kincaid (1997, 1998a) argue that the validity or invalidity of
methodological individualism is ultimately an empirical issue. If that is the case, then the
evidence is against the narrow version. No significant explanation of social phenomena in
terms of individuals alone has yet been advanced. In practice there is always a social and
relational residual that is not reduced entirely to individual terms.
norms in attempts to give individual preferences more substance and
meaning. Still others propose the ‘multiple self’ in which the individual is
treated like a social organization of multiple wills (Elster, 1986). Again,
these works contain valuable insights. But the problem is that an augmented
individual is constructed to carry the entire explanatory burden of
social phenomena. As Archer (1995, p. 251) observes: ‘What is going wrong
here is the desperate incorporation of all emergent and aggregate social
properties into the individual.’ In all these cases, the key omission is a failure
to explain how and why the individual acquires the assumed ‘social’
characteristics. It is not clear how these assumed individual characteristics
could themselves be explained without reference to social relations or
structures.
In a second case, it is fatally admitted that individuals can be somehow
changed by social institutions, in which case the inevitable result is that
narrow methodological individualism is abandoned. The injunction to explain
all social phenomena solely in terms of given individuals founders,
once it is admitted that social institutions can change individuals. For example,
Friedrich Hayek (1943, 1948, 1952b) has been regarded by some as a
promoter of methodological individualism. At the same time he admitted
that people are formed by society, just as individuals (intentionally or unintentionally)
form society through their combined actions. Thus Hayek
(1948, p. 6) sought explanations of ‘social phenomena ... through our understanding
of individual action’ but declared on the very same page that
society is composed of ‘men whose whole nature and character is determined
by their existence in society’. As much as providing an ‘individualistic
analysis’ of social phenomena, Hayek also conceded that individuals
have to be understood in terms of their social circumstances. Hence the individual
alone was not given ultimate explanatory primacy. This made it
impossible for Hayek to be a methodological individualist, at least in the
strict and narrow sense above. If it is believed that ‘social phenomena’ are
explained by ‘individual action’ and the individual is ‘determined by …
society’, then the causality goes both ways. There is no warrant to describe
this as ‘methodological individualism’ any more than ‘methodological
collectivism’.
The notion that individuals are socially determined must undermine
any attempt to give the individual explanatory priority over social structures.
The reason for this is that a socially determined individual cannot
provide the ultimate explanatory bedrock that methodological individualism
requires. Elster’s suggestion that ‘all social phenomena’ have to be explained
‘only in terms of individuals’ is untenable if individuals
themselves are then to be explained in terms other than individuals alone.
If institutional influences on individuals are admitted, then these too are
worthy of explanation. In turn, the explanation of those may be in terms of
other purposeful individuals. But where should the analysis stop? The
purposes of an individual could be partly explained by relevant
18 Introduction
institutions, culture and so on. These, in their turn, would be partly explained
in terms of other individuals. But these individual purposes and
actions could then be partly explained by cultural and institutional factors,
and so on, indefinitely. We are involved in an apparently infinite regress,
similar to the puzzle ‘which came first, the chicken or the egg?’ Such an
analysis never reaches an end point.
It is simply arbitrary to stop at one particular stage in the explanation
and say ‘it is all reducible to individuals’ just as much as to say it is ‘all social
and institutional’. The key point is that in this infinite regress, neither
individual nor institutional factors have legitimate explanatory primacy.
The idea that all explanations have ultimately to be in terms of individuals
(or institutions) is thus unfounded. Once we admit that the individual is
socially determined then we have an explanatory infinite regress, and neither
individuals nor institutions can be the legitimate final term. Hence
methodological individualism – in any adequately meaningful sense – has
to be abandoned (Nozick, 1977; Hodgson, 1988).
In a third case – which is found in the ‘new institutional economics’ – attempts
are made to explain the origin of institutions from interacting individuals,
starting from an institution-free ‘state of nature’. For example,
Carl Menger (1871) pioneered a basic analysis of how institutions evolve.
His chosen example was the institution of money. Menger saw money as
emanating in an undesigned manner from the communications and interactions
of individual agents. He started with a barter economy. The wellknown
problem with barter is the lack of a general ‘double coincidence of
wants’. To deal with this problem, traders look for a convenient and frequently
exchanged commodity to use in their exchanges with others. Once
such usages become prominent, a circular process of institutional self-reinforcement
takes place. Emerging to overcome the difficulties of barter, a
money commodity is chosen because it is frequent and convenient, and it is
all the more convenient and frequent because it is chosen. This circular,
positive feedback leads to the emergence of the institution of money.
Menger argued that there is a basic division between institutions that
emerge spontaneously and those that result from a process involving overall,
deliberate design. Menger’s discussion of money has been interpreted
as an attempt to show how some institutions could emerge spontaneously
from the interactions of individuals in an institution-free ‘state of nature’.
This type of explanation is evident in the ‘new’ institutional economics of
Oliver Williamson (1975, 1985), Richard Posner (1973), Mancur Olson
(1965) and many others. This type of work is concerned to show how spontaneous
institutions can emerge, simply out of the interactions of individuals,
each pursuing their given purposes and preferences. Andrew Schotter
(1981, p. 5) went so far as to define ‘economics as the study of how individual
economic agents pursuing their own selfish ends evolve institutions as
a means to satisfy them’ (emphasis removed). The stress is on a ‘bottom up’
Agency and structure 19
approach: given a set of interacting individuals, how do institutions
emerge?
The value of all this work should not be denied. Substantial heuristic insights
about the development of institutions and conventions have been
gained on the basis of the assumption of given, rational individuals. But
even in its own terms there are serious problems with this approach. Alexander
Field (1979, 1981, 1984) has advanced a fundamental criticism. In attempting
to explain the origin of social institutions, the new institutional
economics always has to presume given individuals acting in a certain
context. Along with the assumption of given individuals, is the supposition
of given rules of behaviour governing their interaction. What is sometimes
forgotten is that in the presumed ‘state of nature’ from which
institutions are seen to have emerged, a number of weighty rules, institutions
and cultural and social norms have already been (implicitly or explicitly)
assumed. These original institutions, rules and norms are
unavoidable; even in an unreal ‘thought experiment’ we can never properly
envisage an original ‘state of nature’ without them.
For example, in attempting to explain the origin of institutions through
game theory, Field pointed out that several constraints, norms and rules
must inevitably be presumed at the start. There can be no games without
constraints or rules, and thus game theory can never explain the elemental
constraints or rules themselves. As Field (1984) argued, game theory may
be used to explain the emergence of some institutions, but to do so it has to
assume at the beginning a significant number of rules and constraints.
Even in a sequence of repeated games, or of games about other (nested)
games, at least one game or meta-game, with a structure and payoffs, must
be assumed at the outset. Any such attempt to deal with history in terms of
sequential or nested games is thus involved in a problem of infinite regress:
even with games about games about games to the nth degree there is
still one preceding game left to be explained.
As another illustrative example, Williamson (1975, p. 20; 1985, p. 143)
wrote that ‘in the beginning there were markets’. However, the market itself
is an institution. The market involves social norms and customs, instituted
exchange relations, and information networks that themselves have
to be explained. All market and exchange relations involve complex rules
and thus markets cannot be an institution-free ‘beginning’. As Viktor
Vanberg (1986, p. 75) put it: ‘What we call a market is always a system of
social interaction characterized by a specific institutional framework, that is,
by a set of rules defining certain restrictions on the behavior of market participants.’
Like others, Williamson failed to explain the evolution of the
firm from an institution-free ‘state of nature’. In a comparative-statics approach,
he implicitly assumed one institutional framework and explicitly
attempted to derive another. Accordingly, the project of starting simply
from given individuals was implicitly abandoned.
20 Introduction
Numerous critical studies have confirmed a similar defect. In the
claimed ‘methodological individualism’ of Karl Popper (1945), ‘the social
phenomena have not really been eliminated; they have been swept under
the carpet’ (Lukes, 1973, pp. 121–2). Likewise, in neoclassical economics,
claims to implement ‘methodological individualism’ in fact reveal hidden
assumptions concerning social structures (Kincaid, 1997). Mario Bunge
(1998, p. 80) argued that proclaimed attempts of methodological individualism
often have ‘a hidden holistic component’. Similarly, Kyriakos
Kontopoulos (1993, p. 79) noted that ‘a methodological individualist strategy
necessarily incorporates references to social relations’. As these critics
have showed, claimed methodological individualists never start from individuals
alone.
The strict and narrow methodological individualist has a problem of
potentially infinite regress: attempts to explain each emergent layer of institutions
always rely on previous institutions and rules. According to the
Mengerian research programme, these in turn have to be explained. Unless
an institution-free state of nature can be discovered, the idea of explaining
all institutions in terms of individual interactions alone faces an
infinite chain of links to be revealed.
There is a particular and fundamental reason why the idea of explaining
all institutions in terms of the interactions of individuals, starting from an
institution-free state of nature, must be abandoned. This is because all individual
interactions depend unavoidably on some – at least rudimentary –
form of language. Language itself is an institution. Individuals rely on customs,
norms, and the institution of language, in order to interact. Interpersonal
communication, which is essential to all stories of institutional
emergence, itself depends on linguistic and other rules and norms.7
The institution-free
state of nature is unattainable, in minimally adequate theory
as well as in reality.
Individual choice requires a conceptual framework to make sense of the
world. The reception of information by an individual requires a paradigm
or cognitive frame to process and make sense of that information. The acquisition
of this cognitive apparatus involves processes of socialization
and education, involving extensive interaction with others (Cooley, 1902,
1922; Mead, 1934; Fleck, 1979; Burge, 1986; Douglas, 1986; Hodgson, 1988;
Bogdan, 2000). As well as language, these interactions require other, preexisting
institutions. The means and mechanisms of our understanding of
the world are necessarily acquired through social interaction. Cognition is
a social as well as an individual process. Individual choice is impossible
without these institutions and interactions. We cannot understand the
Agency and structure 21
7 Bovill (1958) noted that the Moors and Ashanti traded salt for gold without a verbal
language, by placing their products on opposite banks of the river and withdrawing,
taking the merchandise back if the other offer was not deemed to be satisfactory.
Nevertheless, even in this case there was a form of communication with shared
interpretations and meanings. Otherwise trade would not be possible.
world without concepts and we cannot communicate without some form
of language.
What is being contested here is the possibility of using given individuals
as the institution-free starting point in the explanation. The above arguments
show that attempts to start simply from individuals must actually
start from individuals plus institutions. The canons of narrow methodological
individualism may be proclaimed, but they are not followed.
All theories must first build from elements which are taken as given.
However, the particular problems identified here undermine any claim
that the explanation of the emergence of institutions can start from some
kind of institution-free ensemble of (rational) individuals in which there is
supposedly no rule or institution to be explained. Consequently, the project
to explain the emergence of institutions on the basis of given individuals
runs into difficulties, particularly with regard to the conceptualization
of the initial state of nature from which institutions are supposed to
emerge (Hodgson, 1998a).
Overall, while methodological individualism is a popular mantra, in
narrow terms it is never actually achieved. Explanations are never reduced
to individuals alone. The advocates of this approach fail to carry out their
own prescriptions.
A reformulated project would stress the evolution of institutions, in part
from other institutions, rather than from a hypothetical, institution-free
‘state of nature’. Notably, in recent years, a number of significant studies
have developed in this direction. Accordingly, Jack Knight (1992) criticized
much of the new institutionalist literature for neglecting the importance
of distributional and power considerations in the emergence and
development of institutions. Even more clearly, Masahiko Aoki (2001)
identified the problem of infinite regress in much of the former literature
and developed a novel approach. He not only took individuals as given,
but also assumed a historically bestowed set of institutions. With these materials,
he explored the evolution of further institutions. With these studies,
the goal of narrow methodological individualism is abandoned. Some
institutions are taken as given, rather than attempting to conflate their explanation
upon the individual.8
22 Introduction
8 This is what game theory essentially does. A payoff matrix is assumed that expresses not
only individual preferences but also institutional circumstances, rules or constraints. The
starting point in game theory always involves – and unavoidably so – both individuals and
institutions. As Shubik (1982, p. 8) put it, in game theory ‘the rules of the game include not
only the move and information structure and the physical consequences of all decisions,
but also the preference systems of all the players’. However, the treatment of institutional
constraints as given challenges the widespread but unelaborated genuflections to
‘methodological individualism’. From a post-Darwinian and institutionalist viewpoint, a
crucial question concerns the causal explanation of the presumed individual payoffs and
their individual and institutional underpinnings. That which is assumed by the game
theorist must at some stage be explained.
In sum, attempts to conflate explanations of social phenomena upon the
individual have generally failed, because some ‘social’ aspect of the individual
is simply assumed and cannot conceivably be explained without
reference to social relations or structures, or because it is admitted that individuals
are moulded by social circumstances, or because the theorist
never actually starts from individuals alone.
Methodological collectivism
Today, warnings of the dangers of methodological collectivism (sometimes
called methodological holism) are relatively commonplace.
9
By reversing
the aforementioned (narrow) definition of methodological
individualism, methodological collectivism can be defined symmetrically
as the notion that all individual intentions or behaviour should be explained
entirely in terms of social, structural, cultural or institutional phe-
nomena.10
As with methodological individualism, we are concerned with doctrines
that might come close to this extreme case. Hence methodological
collectivism may suggest versions of ‘structural determinism’, ‘cultural
determinism’, ‘economic determinism’ and ‘technological determinism’.
The versions that are close to methodological collectivism see individual
thought or behaviour as being determined largely by structural, cultural or
technological factors. In turn ‘structure’, ‘culture’, ‘economy’ or ‘technology’
are often seen as having a powerful logic and dynamic of their own.
Social, cultural or technological systems are seen to dominate any individual
motives or behaviours. Such systems are upheld to have their own teleology.
They act somehow upon individual actors, who are dragged in their
wake.
Examples or hints of methodological collectivism are found in Marxism,
in the sociology of Durkheim, in structuralist or functionalist sociology or
anthropology, and even in some versions of postmodernism. For instance,
structure is asked to accomplish most or all of the explanatory work in the
structuralist anthropology of Claude Lévi-Strauss (1962) and in the
Agency and structure 23
9 The first appearance of the term ‘methodological collectivism’ may be in Hayek’s (1943, p.
42) critique of attempts to treat ‘social phenomena not as something of which the human
mind is a part’. But this, misleadingly, is primarily an ontological rather than a
methodological statement.
10 Note that such terms are sometimes used in different ways. For instance, Dugger and
Sherman (1994, p. 107) claim that ‘institutionalism relies on methodological collectivism
rather than on methodological individualism. Since institutionalism is a cultural science,
the individual is seen as a product of culture.’ However, they immediately go on to write:
‘The individual is not a cultural marionette, because individuals can and do transform
their culture through collective action and even through individual action.’ This
somewhat qualifies the former statement and implies that explanations entirely in terms of
social phenomena such as culture would be generally inadequate. Hence Dugger and
Sherman do not advocate methodological collectivism in the strict sense that I define it
here.
functionalist sociology of Talcott Parsons (1937). The ‘postmodernist’ JeanFrançois
Lyotard (1984, p. 15) wrote: ‘A self does not amount to much.’ The
tendencies in such accounts are to downgrade the human subject and to
see everything that is human as entirely derivative from society.11
If Marx is accused of being a methodological collectivist, then his defenders
will point out in response that he acknowledged the role of the individual.
Nevertheless, there are some highly misleading passages. For
example, Marx wrote in 1845: ‘But the human essence is no abstraction inherent
in each single individual. In its reality it is the ensemble of the social
relations’ (Marx and Engels, 1976a, p. 4). The danger in this assertion is that
the individual could be regarded as no more than an expression of social
relations. Similarly, in a section of the German Ideology written at about the
same time, Karl Marx and Frederick Engels (1976a, p. 59) wrote: ‘The ruling
ideas are nothing more than the ideal expressions of the dominant material
relations’. The similar pitfall here is that ideas and volitions could be
seen simply as expressions of the ‘material relations’ of the social structure.
These problematic formulations are not confined to their early works. In
the 1860s Marx (1976, p. 989) described how the actions of the capitalist are
‘no more’ than the manifestation of capitalist structures:
The functions fulfilled by the capitalist are no more than the functions
of capital … executed consciously and willingly. The capitalist functions
only as personified capital, capital as a person, just as the worker is no
more than labour personified.12
A similar idea is repeated in the third volume of Capital. Marx (1981, pp.
1019–20, emphasis added) wrote:
The principal agents of this mode of production itself, the capitalist
and the wage-labourer, are as such simply embodiments and personifications
of capital and wage-labour – specific social characters that the
social production process stamps on individuals, products of these
specific social relations of production.
The problem here is that explanations of individual agency seem to be conflated
entirely upon ‘material relations’ and ‘social structures’, without
recognition of individual diversity, cultural variation or discretionary possibilities.
Although multiple interpretations of these passages are possible,
24 Introduction
11 Elements of postmodernism that lead to a downgrading of the human subject are
criticized at length in Archer (2000, pp. 18–50).
12 This Marxist view of the capitalist as severely constrained by capitalist structures is
contested in Hodgson (1999a) where I propose greater consequential differences between
capitalist institutions and national cultures. Capitalists have a significant zone of
discretion and their behaviour is not entirely dominated by capitalist structures.
Consequently, an infinite variety of different types and trajectories of capitalism are
possible.
Marx did not do enough to guard against a methodological collectivist
interpretation.
A related difficulty in Marx’s writing is his divination of social or ‘productive
forces’ that, at least in some accounts, seem to have powers over
and above individuals. For example, in the famous Preface to A Contribution
to the Critique of Political Economy of 1859, Marx (1971, p. 21) wrote that
‘the material productive forces of society come into conflict with the existing
relations of production’. Such phrases have suggested that Marx was
underplaying the role of the individual and making largely mysterious
and undetailed ‘productive forces’ do the entire work of explanation of social
change. On the other hand, the formulation in the Preface does have its
relatively sophisticated defenders (G. Cohen, 1978). Furthermore, Marx
and Engels (1975, p. 93) wrote in 1845 that 'history is nothing but the activity
of man pursuing his aims’. But this apparent rejection of ontological
collectivism was not enough to prevent his work being interpreted in
methodologically collectivist terms.
Some of Marx’s followers were less circumspect. For example, Nicolai
Bukharin (1969, p. 40) wrote in a Soviet Russian textbook of 1921 that: ‘Social
phenomena determine at any given moment the will of the various individuals.’
Much later, in his structural determinist account of Marxism,
Louis Althusser gave explanatory priority to structure, while downgrading
the human subject. He wrote: ‘The true “subjects” are … the relations of
production’ (Althusser and Balibar, 1970, p. 180). There are many similar
Marxist examples of such a conflation of the individual with social relations
or structures.
Turning to Durkheim, he too was strongly influenced by Marx but he
reacted against the ‘materialist’ aspects of Marx’s thought. Against Marx’s
notion that economic or material factors somehow determined individual
thought or action, Durkheim (1982, p. 247) wrote in 1908: ‘In social life, everything
consists of representations, ideas and sentiments’. But in 1897
Durkheim (1982, p. 171) also insisted that:
We believe it is a fruitful idea that social life must be explained not by
the conception of it formed by those who participate in it, but by the
profound causes which escape their consciousness. We also think that
these causes must be sought mainly in the way in which individuals
associating together are formed in groups.
But this clearly created a problem for Durkheim.13
First, the character of
these profound causes is not clear. To be consistent with the 1908 statement,
these causes must also be ‘representations, ideas and sentiments’ –
possibly those shared within groups or widely dispersed among society.
Agency and structure 25
13 The following discussion draws heavily on Lukes’s excellent introduction to a translation
of Durkheim’s Rules of Sociological Method (Durkheim, 1982).
Durkheim failed to resolve this problem and took refuge in unelaborated
metaphorical phrases, such as ‘collective forces’ and ‘social currents’.
Durkheim (1982, p. 59, emphasis removed) defined his basic concept of the
‘social fact’ in the following terms:
The social fact is any way of acting, whether fixed or not, capable of exerting
over the individual an external constraint … which is general
over the whole of a given society whilst having an existence of its own,
independent of its individual manifestations.
But there is a big difference between seeing such emergent phenomena
as independent of any one individual, or external to all individuals. The
above quotation is ambiguous. Permitted by such ambiguities and aided
by his social metaphors, Durkheim sometimes slid towards a methodological
collectivism, where society and ‘social forces’ somehow stand above
and manipulate all individuals.
Like Marxism, Durkheimian sociology lacks a developed micro-theory
of how social structures affect, and are affected by, individual purposes or
dispositions. Marx seemed to make psychology redundant, by declaring
that the human essence was nothing more than the ‘ensemble of the social
relations’. More explicitly, Durkheim (1982, p. 129) banned psychology
from social science with his famous declaration in 1901 that ‘every time a
social phenomenon is directly explained by a psychological phenomenon,
we may rest assured that the explanation is false’. The consequences of
such neglects or prohibitions are highly damaging for social theory.
In the absence of a theory of how society may lead to the reconstitution
of individual preferences or purposes, a temptation is to explain individual
action primarily by reference to the constraints imposed by the evolving
social organism upon the individual. Institutional constraints have effects,
but without necessarily changing individual inclinations. For Durkheim
(1982, p. 144) ‘social life presents itself to the individual under the form of
constraint’. His concept of ‘constraint’ seems to include anything from legal
rules and their sanctions to matters of mere convenience, communication
or coordination.
With such a concept of social constraint, there is some notion of the
power that social institutions can hold over the individual. However,
Durkheim’s concept of social power is itself incomplete. As Steven Lukes
(1974) argues in a classic study, power itself has multiple dimensions. One
possibility is that power may be exercised by ‘coercion, influence, authority,
force or manipulation’ (Lukes, 1974, p. 17) but these mechanisms do
not necessarily involve the alteration of individual preferences, purposes
or values. For Lukes, the overemphasis on the coercive aspect of power ignores
the way that it is often exercised more subtly, and often without
overt conflict. Lukes (1974, p. 23) thus wrote:
26 Introduction
To put the matter sharply, A may exercise power over B by getting him
to do what he does not want to do, but he also exercises power over
him by influencing, shaping or determining his very wants. Indeed, is
it not the supreme exercise of power to get another or others to have
the desires you want them to have – that is, to secure their compliance
by controlling their thoughts and desires?
Consider an example. If a criminal desists from crime, simply because
they fear the risk of apprehension and punishment, then behaviour is
changed through the force of deterrence and potential constraint. On the
other hand, if someone persuades the criminal that wrongdoing is evil,
and that there are morally superior ways of earning a living, then the released
criminal will desist from crime, even if the constraints and perceived
penalties are ineffective. The preferences and purposes of the
criminal would have been changed through persuasion.14
Both Durkheim and Marx lacked an adequate account of how individual
dispositions are moulded. Such accounts must necessarily include psychological
mechanisms. In their absence the temptation is to place the
emphasis on social constraints, rather than on the additional reconstitution
of individuals themselves. This emphasis on constraints diminishes and
denudes the concept of social power, including the dimension of power
where individual purposes and preferences may be changed.
Many social theorists have criticized methodological collectivism for
making the individual the mere puppet of social forces. In addition, it is argued
here that the main problem is that methodological collectivism not
only diminishes the individual, but it also pays insufficient attention to the
processes and mechanisms by which the individual is fundamentally altered.
One consequence of conflating the individual into the structure is to
lose sight, not simply of the individual, but also of the mechanisms of social
power and influence that may help to reconstitute individual purposes
or preferences. It may appear paradoxical, but only by rescuing the individual
from its conflation into the social, can the social determination of individuality
be fully appreciated.
Part of the solution is to bring psychology back into the picture. But
strangely this is absent from much of social theory. There is very little psychology
in Marxism, partly because the subject was so underdeveloped
during Marx’s time. Durkheim himself bears part of the responsibility for
the exclusion of psychology from the main currents of twentieth-century
sociology. The influential Talcott Parsons (1937) was persuaded partly by
Durkheim in this and other respects. Instead of psychology, and in a
Agency and structure 27
14 Alternatively, Stigler and Becker (1977) would argue that no change in the preference
function took place. Instead the persuader simply revealed new information to the actor.
However, Stigler and Becker assumed a (meta-)preference function that is capable of
accommodating an immense number of contingencies, certainly exceeding the
computational and memory storage capacities of any human brain.
manner highly reminiscent of Durkheim, Parsons emphasized the power
of social norms.15
Some influential neoclassical economists also abandoned psychology at
about the same time. Lionel Robbins (1932) recast economics as ‘the science
of choice’. Individual ends were taken as given, economics was to be all
about the rational choice of appropriate means. Because individual preferences
were taken as given, psychology no longer had a significant role in
this reconstruction of the subject (Hodgson, 2001c).
After their common rejection of psychological and other underpinnings,
economics and sociology went their separate ways. Proclamations
of methodological individualism were more prominent in economics, and
of methodological collectivism in sociology. The social sciences as a whole
were characterized as an apparent dilemma between an Adam Smith-like
and incentive-driven view of action, on the one hand, and a Durkheim-like
and norm-propelled view, on the other. In one discipline there appeared
the ‘self-contained’, ‘under-socialized’, ‘atomistic’ and ‘asocial’ individual;
in the other the individual seemed sometimes to be the ‘over-socialized’
puppet of ‘social forces’.
However, despite the century-long battle between methodological individualists
and methodological collectivists, they have much more in common
than is typically admitted. Methodological individualism conflates
the social upon the individual, thus losing sight of key mechanisms of social
influence, and is consequently impelled to take the purposes and preferences
of the individual as given. Methodological collectivism conflates
the individual upon society and thereby lacks an explanation or adequate
recognition of how individual purposes or preferences may be changed.
The explanatory moves are different but the results are similar in some vital
respects: there is no adequate explanation of how social institutions
may reconstitute individual purposes and preferences. Typically, both approaches
disregard the value and role of psychology in the explanation of
social phenomena. Both methodologies end up with a diminished concept
of social power, and an analytical overemphasis on overt coercion and constraint,
rather than more subtle mechanisms of social influence.
Accordingly, as long as the debate within social theory simply moves
back and forth along the line between these two positions then it will be incapacitated
by a failure to examine, and escape from, their common presuppositions.
They are two mutually implicated poles of a misconceived
and unsustainable dualism; they have both demonstrably failed to bring
social theory out of its twentieth-century impasse.
28 Introduction
15 See Hodgson (2001c) for an account of how Parsons was also partly persuaded in this
respect by his teacher Ayres. The institutional economist Ayres is discussed later and at
length in the present work.
Reductionism and reduction
Methodological individualism and methodological collectivism are both
different versions of reductionism, by which is meant the more general
doctrine that all aspects of a complex phenomenon should be completely
explained in terms of one level, or type of unit. This is a strong definition of
reductionism, involving a universal imperative of explanation. But it is not
lacking in adherents, such as Elster (1983, pp. 20–4): ‘Generally speaking,
the scientific practice is to seek an explanation at a lower level than the
explandum. ... The search for micro-foundations … is in reality a pervasive
and omnipresent feature of science.’ A major theme of this book is to criticize
the reductionist view that such outcomes are generally attainable and
always necessary.
16
The definition here of reductionism in terms of explanation should be
distinguished from ontological and epistemological reductionisms. Ontological
reductionism involves the claim that wholes are nothing but their
parts (or vice versa). Epistemological reductionism claims that we know of
the whole entirely by knowledge of the parts (or vice versa). Such doctrines
exist (and are vulnerable to criticism) but they are different from the
definition here of reductionism in terms of explanations of nature and origin.
We can find many pronouncements of (explanatory) reductionism. Biological
reductionism proposes that (social) phenomena should be
explained solely in terms of biological characteristics. Physical
reductionism requires that (biological, chemical or other) phenomena
should be explained solely in terms of physical characteristics. Neurological
reductionism proposes that psychic phenomena should be explained
entirely in neurological terms. Methodological individualism pursues universal
explanations of social phenomena in terms of individuals, and
methodological collectivism the reverse.
Reductionism should also be distinguished from reduction. Emphatically,
some degree of reduction to elemental units is inevitable and
desirable in science. Even measurement is an act of reduction. Science cannot
proceed without some dissection and some analysis of parts. However,
although some reduction is unavoidable, complete analytical reductions
are generally impossible. They are beset by analytical explosions in the
number of combinations of elements; they are cursed by the ubiquitous
phenomenon of complexity. Complete analytical reductions are rarely, if
ever, achieved.
Agency and structure 29
16 Several authors have incorrectly identified Quine (1951) as the origin of the term
‘reductionism’. Earlier uses include Urban (1926, p. 110) – who wrote of the ‘revolt of
Emergent Evolution against reductionism’ – as well as Werkmeister (1937), Alpert (1938)
and several others. The term is used in several different ways, but here I follow the usage
of the term in classic presentation of the reductionist case, in Oppenheim and Putnam
(1968), who advocated a reductionism in which explanations of phenomena were derived
successively from scientific laws at lower ontological levels. Agazzi (1991) provides a
useful set of essays on reduction and reductionism in science. See also Dupré (1993).
The strong version of reductionism criticized here cannot be refuted in
principle, because any explanatory deficit might sometime be remedied in
the future. In no science are the canons of reductionism strictly enforced.
Yet the sciences achieve results. Although we can never be sure that someday
a missing explanation will be found, and a further explanatory reduction
might be possible, there is sufficient evidence from the sciences to
undermine the reductionist dogma and to diminish reductionist ambitions.
While reduction is a worthwhile and important aim, the sciences do
not need 100 per cent reductionism to qualify as science.17
Few reductionists acknowledge a fateful consequence of their own doctrine.
If reductionism were viable, and complete analytical reduction to
lower levels were possible, then the result would not be methodological individualism
but the dissolution of all sciences except subatomic physics.
Everything would have be brought down and explained in its terms. There
would be no mechanics, no thermodynamics, no chemistry, no biology
and no social science. All sciences would be reduced to one. The reason
why we have different sciences is that complete explanatory reduction is
generally beyond reach, and multiple levels of explanation are both appropriate
and powerful. It should now be clearly seen that reductionist ambitions
are at best, wildly optimistic, and at worst, dogmatic and
diversionary.
Central conflation
In the 1980s, largely as a result of the stimulating work of Anthony
Giddens (1976, 1979, 1984), debates on the problem of agency and structure
were given a welcome impetus. Giddens’s ‘structuration theory’ is an
attempt to steer a middle course between structural determinism and functionalism,
on the one hand, and voluntarist, individualist and subjectivist
formulations, on the other. Its attraction is to propose an alternative to the
extremes of both methodological individualism and methodological collectivism.
Giddens countered the widespread belief that the only alternative
to methodological individualism is the slippery slope to
methodological collectivism, or vice versa. He argued that social theory
should focus exclusively neither on the social totality nor simply on the experiences
or behaviours of individual actors. Instead, social theory should
30 Introduction
17 Dennett’s (1995, pp. 80–3) condemnation of the sin of ‘greedy reductionism’ is confusing
because reductionism by its nature is gluttonous. All reductionism is greedy, as it desires
beyond possibility and need. According to Dennett, those who are guilty of ‘greedy
reductionism … underestimate the complexities, trying to skip whole layers or levels of
theory in their rush to fasten everything securely and neatly to the foundation’. However,
such persons are not simply guilty of the sin of greed, but also of haste and sloppiness.
Dennett’s rejection of ‘greedy reductionism’ is essentially a doomed attempt to cleanse
reductionism of its reckless deviants and to retain respectable reductionist credentials.
Symptomatically, Dennett does not admit that complete explanatory reductions are
generally unattainable in science. He is thus charged by his own indictment.
take its starting point as ‘recursive social practice’ and consider the ways in
which such practices are sustained through time and space.
18
At the heart of his structuration theory is the notion of ‘duality of structure’.
For Giddens, the idea of a duality is contrasted with that of a dualism.
The two elements of a dualism are regarded as mutually exclusive or
separable (Dow, 1990). By contrast, in a duality the parts are interdependent:
each element may actually help to constitute or sustain the other.
Giddens regarded agent and structure as a duality: where both human
subjects and social institutions are jointly constituted in and through recurrent
practices, and where no element has ontological or analytical priority
over the other. In Giddens’s theory, structure and agency are mutually and
symmetrically constitutive of each other.
In structuration theory, the idea of structure is tied up with ongoing
processes and capabilities. Structure is less an objective thing: more a ‘virtual
order’ of ‘transformative relations’. In short, Giddens (1982, p. 35) saw
structures as ‘recursively organized rules and resources’. For Giddens,
agency is both free and constrained. Human beings are reflective of, and
reactive to, their circumstances, as well as being conditioned by them.
Equally, instead of the prominent idea that the ‘structural properties of society
form constraining influences over action ... structuration theory is
based on the proposition that structure is always both enabling and constraining’
(Giddens, 1984, p. 169).
Social life is reproduced by drawing upon social structures, just as social
structures are reproduced (intentionally or unintentionally) through the
practices of social life. Structure refers to the ‘structuring properties’ that
make it possible for similar social practices to endure in time and extend in
space. Hence, in structuration theory, ‘structure’ is as much a verb as a
noun.
In the idea of structuration, no stress is placed on the existence of different
ontological or analytical levels. Instead, agent and structure are regarded
as different aspects of the same process. As Ian Craib (1992, pp.
3–4) put it in his commentary on Giddens, structure and agency are not
treated as ‘separate and opposing things in the world or as mutually exclusive
ways of thinking about the world’ but as ‘simply two sides of the same
coin. If we look at social practices in one way, we can see actors and actions;
Agency and structure 31
18 As well as Cooley (1902), who is discussed later below, in some respects there are also
similarities with the work of Bourdieu (1990) and Elias (1991, 2000), who attempted to
escape similar dichotomies. The concept of habitus in the works of Bourdieu (1990) and
Elias (1991) has some similarities with Giddens’s concept of routinization. Storper and
Salais (1997) have tried to synthesize the approach of Giddens with that of Bourdieu. Elias,
like Giddens, emphasized process, and the mutual reconstitution of individual and
society. American authors including Boas, Cooley, Ellwood, James, G. H. Mead, Small and
Sumner influenced Elias. When Elias (2000, p. 455) wrote: ‘concepts such as “individual”
and “society” do not relate to two objects separately but to two different yet inseparable
aspects of the same human beings’, he suggested a dual aspect formulation, similar to that
of Cooley and Giddens. For simplicity, we focus solely on the work of Giddens in this
section.
if we look at them in another way we can see structures.’ In fact, there is little
to stop Giddens’s duality of agent and structure collapsing into the
more special case of a ‘dual aspect theory’, where agent and structure become
different facets of a unity. As noted later below, other versions of
dual aspect theory appear in the sociologies of Charles Horton Cooley
(1902) and in attempted solutions to the mind–body problem.
What is missing in Giddens’s account? In contrast to structuration and
dual aspect theory, several philosophers have proposed an ontology in
which reality is irreducibly layered: successively with physical, molecular,
organic, mental, human individual and social layers.19
Everything belongs
to a level and each level has, within bounds, some autonomy and stability.
However, no level is disconnected from others: each layer is linked to, and
dependent upon, other layers. Such a stratified ontology is essentially absent
from structuration theory. The individual and the social levels are
conflated into the central ground of the recursive structure.
Essentially, in stratified ontologies, what separates one layer from another
is the existence of emergent properties at the higher level. Units exist at
higher levels that are not mere epiphenomena of lower-level units. A viable
and irreducible hierarchical ontology depends upon the notion of
emergent properties. As related in later chapters, the concept of emergent
properties was developed by the philosopher George Henry Lewes (1875),
the psychologist and philosopher of biology Conway Lloyd Morgan
(1923), and several others. A property may be said to be emergent if its existence
and nature depend upon entities at a lower level, but the property is
neither reducible to, nor predictable from, properties of entities found at
the lower level.20
Significantly, in his conflationary strategy, Giddens makes no significant
or explicit use of the idea of emergent properties. Indeed, in one passage
he rejected such a concept by suggesting that ‘human actors ... do not
come together ex nihilo to form a new entity by fusion or association’
(Giddens, 1984, p. 171). This denial creates a serious problem. Central to
Giddens’s structuration theory are notions such as self-reflexivity and consciousness.
But if neural entities ‘do not come together ex nihilo to form a
new entity by fusion or association’, then how can human consciousness or
self-reflexivity be explained? Arguably, consciousness is an emergent
property of interactions in the human neurosystem. We have to rely on
emergent properties to sustain notions such as consciousness and self-reflexivity,
which are central to structuration theory. Likewise, the existence
32 Introduction
19 See, for example, Broad (1925), Sellars (1926), Bunge (1973a, 1973b) and Bhaskar (1975).
Emergentist philosophy is discussed later in the present work.
20 The idea of emergent properties is similar in some respects to the concepts of ‘creative
synthesis’ (Wundt, 1895; Ward, 1903; Sellars, 1918, 1922) and ‘synergy’ (Ward, 1903;
Ansoff, 1965; Corning, 1983, 2000a). The origin of, and relation between, these three
concepts is discussed in later chapters below.
of a social structure depends upon emergent properties; otherwise it
would be reducible to the individuals involved.
Third, a consequence of Giddens’s rejection of emergent properties is
not only the rejection of a higher and social level of analysis with their own
emergent properties, but also the analytical neglect of the natural and
physical world as the essential substratum and context of human activity.
The denial of emergent properties forces structuration theory to accept a
single level of reality, with nothing (social or otherwise) ‘above’ it, and
nothing (natural or otherwise) ‘below’. One consequence of this is the denudation
of the concept of social structure. Another is the neglect of the
natural and biological substratum of all human activity.
If structuration theory accepts a singular plane of being, then where is
it? Giddens is explicit about this. For him, ‘structure exists ... only in its
instantiations of such [social] practices and as memory traces orienting the
conduct of knowledgeable human agents’ (Giddens, 1984, p. 17). Symptomatically,
the formulation is repeated elsewhere: ‘Structure exists only
as memory traces, the organic basis of human knowledgeability, and as
instantiated in action’ (Giddens, 1984, p. 377). And again, in another work,
for Giddens (1989, p. 256), structure ‘exists only in a virtual way, as memory
traces and as the instantiation of rules in the situated activities of
agents’. An agent carries ‘structural properties’ in its memory, which may
be transmitted through practice from one agent to another. Commentators
on Giddens’s theory thus observe that ‘if structures have a locus of existence,
it is in the heads of social actors’ (Craib, 1992, p. 42). Richard
Kilminster (1991, p. 96) made a similar point: ‘“structure” in Giddens’ theory
is internal to actors’.
A problem with the idea that social structure is entirely mental and internal
is that it downplays the fact that structure consists not merely of persons
or things, but also of interactive relations between persons, in a social
and material context. Hence questions such as ‘where is social structure?’
are essentially misconceived, as a relation between two individuals separated
in space has no singular, meaningful location. A relation is real, but it
is an association, not a singular entity. Individuals may confront these
structures, even if they do not have the memories, ideas or habits that are
associated with them. Newborn infants face a social structure, even if they
have little understanding of it. Rebels and heretics confront a social structure,
when failing to follow its rules or adopt its associated ideas.
Overall, for structuration theory, the single level of being is human
knowledge and action. Structuration theory either takes structure as mental
and internal to actors, or it has few defences to prevent such a conflation.
The problem with this modern solipsist philosophy is much the same
as that which troubled Bishop George Berkeley in the eighteenth century,
who was also tempted to find reality inside the mind. Berkeley asked: if the
world consists merely of our knowledge or perceptions, then how do we
explain the persistence of objects when we do not apprehend them?
Agency and structure 33
Berkeley’s solution was that they endured through time in the perception
of God.
Similarly, Giddens had to explain the persistence of social structures.
He searched for a secular solution to this Berkeleyian puzzle. He found it
in the centrality and persistence of routinized practice. Indeed, for
Giddens (1984, p. 60), the ‘concept of routinization, as grounded in practical
consciousness, is vital to the theory of structuration. ... An examination of
routinization ... provides the master key’. But the next question is how can
the existence of routinization itself be explained? Giddens’s (1984, p. 50)
answer seems essentially to lie in his concept of ‘ontological security’. This
allegedly has its origins ‘in basic anxiety-controlling mechanisms’ that in
turn are acquired by the individual as a result of ‘predictable and caring
routines established by parental figures’.
However, this argument is incomplete and has a strong functionalist
flavour: the replication of routines is explained in terms of their function.
The explanation for the persistence of routines is seen as the search for ontological
security, which in turn results from the persistence of (parental)
routines. However, no adequate explanation is given for the persistence of
these ‘caring routines established by parental figures’. These routines may
be handed down from generation to generation, but why would this be so?
No adequate explanation of the origin or persistence of routines is given,
and Giddens’s discussion of ontological security does not provide it.21
Despite these omissions, the stress on routinization in structuration theory
has affinities with the ‘evolutionary economics’ of Richard Nelson and
Sidney Winter (1982, 2002), which – redolent of the institutionalism of
Veblen and Commons – also stresses routines. However, Giddens (1984,
pp. 228–43) has been a critic of evolutionism in the social sciences and has
rejected ‘evolutionary’ ideas in that domain.22
Related to Giddens’s abandonment of evolutionary theory is his inadequate
treatment of historical time. This defect has been identified by ‘critical
realists’ Roy Bhaskar (1989) and Margaret Archer (1995). For Bhaskar
and Archer, but not for Giddens, human agents and structures are not different
aspects of the same things or processes, but different entities. Although
structures, of course, contain individuals, and structure and agent
34 Introduction
21 Another problem with Giddens’s explanation in terms of ‘ontological security’ is that a
highly stable ontology could have unstable, chaotic or unpredictable outcomes. Also
routines themselves can have highly disruptive effects, for example when they are
embodied in a military organization. What might matter most for the agent would be
epistemological rather than ontological security.
22 In teleological and pre-Darwinian terms, Giddens (1979, p. 233) described evolution as
‘social change as the progressive emergence of traits that a particular type of society is presumed
to have within itself from its inception’. This epigenetic or ‘unfolding’ conception of evolution
is non-Darwinian and very different from the evolutionary economics of Veblen (1899a,
1919a) or of Nelson and Winter (1982). These accounts see economic evolution as an
ongoing, imperfect and non-teleological process of competitive selection, acting upon a
varied population of institutions, habits, customs and routines. Evolution in this
conception is not necessarily progressive, and is not vulnerable to Giddens’s critique.
are interdependent, they are different and distinct. This separation of actor
and structure stems from the fact that, for any particular actor, social structure
always exists prior to her engagement with the world.23
As Bhaskar
(1989, p. 36) wrote:
people do not create society. For it always pre-exists them and is a necessary
condition for their activity. Rather, society must be regarded as
an ensemble of structures, practices and conventions which individuals
reproduce and transform, but which would not exist unless they
did so. Society does not exist independently of human activity (the error
of reification). But it is not the product of it (the error of
voluntarism).
Hence any given individual is preceded by the social structures into
which they are born. In recognizing the temporal priority of structure,
Bhaskar and Archer took their cue from Marx. In 1852, Marx (1973, p. 146)
wrote that: ‘Men make their own history, but not ... under circumstances
they themselves have chosen but under the given and inherited circumstances
with which they are directly confronted.’ Durkheim made a similar
point at the beginning of the twentieth century. In his Rules of Sociological
Method, Durkheim (1982, p. 51) pointed out that the social actor must learn
pre-existing beliefs, laws, customs and so on: ‘if they existed before he did,
it follows that they exist outside him. The system of signs ... the monetary
system ... the credit instruments ... practices ... all function independently
of the use I make of them.’ In a particularly useful study of social structure,
Kyriakos Kontopoulos (1993, p. 211) similarly insisted that ‘institutions are
“always already there” and, thus, become the parameters of new actions
and systems of interaction’.
It is in the work of Archer that the implications of the temporal priority
of structure over the individual are drawn out most clearly. Following
Marx, Durkheim and Bhaskar, Archer (1995, p. 72) wrote: ‘This is the human
condition, to be born into a social context (of language, beliefs and organization)
which was not of our making.’ She criticized Giddens’s
structuration theory as involving a ‘central conflation’ because it conflates
structure and agency into processes acting together at a single level.
Giddens’s duality of structure wrongly treats structure and agency as not
only mutually constitutive but also conjoined.
Archer thus exposed a major difficulty in structuration theory: it cannot
incorporate historical time. Because it resists untying structure from
Agency and structure 35
23 There was an earlier, insufficiently acknowledged but contrasting tradition in American
philosophy describing itself as ‘critical realism’. This is discussed in a later chapter below.
See also Sellars (1908, 1916, 1922), Drake et al. (1920), Bode (1922), Moore (1922) and
Werkmeister (1949). By ‘critical realism’ in the contemporary context I refer to the writings
of Bhaskar (1975, 1989), Archer (1995, 2000) and others. C. Lawson (1994, 1996), T. Lawson
(1997, 2002, 2003a, 2003b), Runde (1998) and others have applied insights from this
perspective to economics.
action, it cannot recognize that structure and agency work on different
time intervals. As individuals, we are born into a set of structures that are
not of our making. Acting within them, they may be changed or sustained
by our actions. We then bequeath them to others. However, Archer did not
conflate individual into structure, giving the latter the sole burden of explanation.
Indeed, the reproduction of social structure depends upon the
actions of the individuals involved. She aimed for ‘a theoretical approach
which is capable of linking structure and agency rather than sinking one
into the other’ (Archer, 1995, p. 65).24
However, the differentiation of structure from agent is valid if structure
is seen as external to any given individual, but not if it is regarded as external
to all individuals. Structure does not exist apart from all individuals,
but it may exist apart from any given individual. In some accounts – such
as Durkheim’s – this distinction is not always given sufficient stress. The
danger, as a consequence, is that the concept of structure may be reified.
Non-conflation but incomplete explanation
In her alternative approach to social theory, Archer (1995, p. 91) proposed
a ‘morphogenetic cycle’ involving first (a) a given structure, then (b) social
interaction, then (c) structural elaboration or modification. The cycle then
indefinitely repeats itself through these three phases. She criticized other
approaches for downplaying particular phases of this cycle. For example,
methodological individualism misses out the first step (a) and then moves
simply from (b) to (c). In contrast to both methodological individualism
and methodological collectivism, all three elements in the cycle are
important.
To recapitulate: as acknowledged above, the work of Giddens was a major
attempt to transcend the dichotomy between methodological individualism
and methodological collectivism. However, Bhaskar and Archer
criticized Giddens for conflating structure and agency. They developed an
approach that likewise transcended the dichotomy between methodological
individualism and methodological collectivism, but emphasized that
structure and agency were different entities. Like Marx and Durkheim, it
asserted that structures historically pre-exist each individual. Archer thus
developed her ‘morphogenetic’ approach that moved cyclically, from
36 Introduction
24 Archer (1995, 2000) also introduced the terms ‘upwards conflation’ and ‘downwards
conflation’. By ‘upwards conflation’ she meant accounts where ‘structure is held to be the
creature of agency’ (Archer, 1995, p. 84) and by ‘downwards conflation’ she referred to
theories where structural forces drive the system and ‘agents are never admitted to touch
the steering wheel’ (Archer, 1985, p. 81). However, her terminology is ambiguous as the
reverse choice of terms could apply. Devolving structural explanations down to the
characteristics of individuals (methodological individualism) could just as well be
described as ‘downwards conflation’, just as methodological collectivism could be
described as ‘upwards conflation’.
structure to social interaction, back to the (modified) structure. A process
of structural evolution was suggested.
While the Archer–Bhaskar critical realist approach is an important advance,
there are problems with it. In particular, while there is a general account
of structural change, there is as yet no account of how individuals
are changed. We are told how structures evolve, but there is no parallel explanation
of the changes to individuals. In contrast to methodological collectivism,
individual agency is rightly retained and emphasized. Bhaskar
(1989, p. 80) and other critical realists have argued repeatedly that ‘intentional
human behaviour is caused’ but ‘it is always caused by reasons, and
that it is only because it is caused by reasons that it is properly characterized
as intentional’. But in critical realism there is no adequate explanation
of the causes of reasons or beliefs. So far, the account of agency in critical
realism is incomplete (Faulkner, 2002).
Bhaskar (1975, pp. 70–1) endorsed an ‘ubiquity determinism’, meaning
that every event is caused. Yet critical realism has so far failed to apply this
universal principle to individual reasons or beliefs. It recognizes rightly
that beliefs are part of social reality, but does not give an account of the cultural,
psychological or physiological causes of beliefs or reasons themselves.
In critical realism there is a general explanation of structural
change, but so far no equivalent explanation of how individual agents acquire
or change their beliefs, reasons, purposes or preferences. The possibility
of such changes may be admitted, but as yet there is no indication in
critical realism of how such changes may be explained.
As a result of this omission, a temptation is to adopt a schema in which
structure somehow channels individual activity with a sufficiency to explain
it, thus putting the emphasis on the role of structures as constraints
on individuals. Instead, what is required is an account of individual
agency that includes an explanation of how structures can lead to fundamental
changes in individual reasons, beliefs or purposes.
Critical realism rightly insists that structure and agency are different entities,
but in making this disassociation, a partial and inadequate account is
so far provided of their interaction. There is a general account of the causal
connection from individuals to structures, but not one from structures to
individuals, which shows how beliefs or reasons are formed or changed.
That is why I characterize the position of Bhaskar (1989) and Archer (1995)
as a case of non-conflation but incomplete explanation.
While critical realism has articulated an important critique of the work
of Giddens, there is a gap in the theories of Bhaskar and Archer: there is no
explanation, even in principle, of the evolution of reasons or beliefs. However,
this limitation is not unique to critical realism: in social science as a
whole, many other approaches share this defect.
For instance, a similar problem can also be found in the writings of
Marx. There is no adequate explanation in Marxism of individual motivations.
They are assumed to spring in broad and mysterious terms from the
Agency and structure 37
relations and forces of the system. A capitalist is said to act as a capitalist
because he occupies a capitalist role within the social structure. A worker is
said to act as a worker because she is obliged to take that social position.
Explanations of individual action within Marxism characteristically devolve
upon structure. Although Marx and Engels often rightly acknowledged
that structures themselves are the result of human activity, they
often describe how in a class-divided society people become prisoners of
these structures. Within Marxism, the connection between social structure
and individual action is made by the presumption of rational reflection
upon their perceived interests acting under the constraints of social structures.
Here too, structure bears the burden of the explanatory work. There
is no explanation of how particular perceptions of interests and interpretations
of situation may arise.
We require an explanation of how individual intentions or preferences
may change. Without such an account, a danger is that structural constraints
are called upon to do the main work of explaining human behaviour.
As a result, the disconnection of agency and structure may end up
explaining the individual solely by reference to structure, thus conflating
the individual into the structure, as criticized above. In contrast, if there is a
causal and psychological explanation of how structures can affect or
mould individual purposes or preferences, then the role of the individual
is placed alongside that of structure and becomes part of a more ample,
two-way explanation. Such a spiral of causation from structure to individual,
and from individual to structure, does not deny individuality; but it
places the individual in their proper place within the ongoing process of
social transformation.
However, while Marxism and critical realism have an inadequate explanation
of individual motives, they are better than many other approaches
in their recognition of the powerful role of social structures over individuals,
while simultaneously attempting to retain a concept of agency. In
many other cases there is an inferior explanation of structural powers and
an equally inadequate explanation of individual transformations. Both
methodological individualism and methodological collectivism come into
this doubly defective category.
Also, much of mainstream economics has exhibited these twin failings.
There are too few attempts to explain individual preference functions.
Similarly, Austrian school economists take the purposes of individuals as
given and do not regard their explanation as the task of economics or any
other social science. As Hayek (1948, p. 67) wrote: ‘If conscious action can
be “explained”, this is a task for psychology but not for economics ... or any
other social science.’ Like many others, Hayek shunned one of the central
problems of social science – to explain human motivation.25
Likewise, methodologically collectivist attempts to explain individuals
exclusively in terms of social structures also typically fail to provide an adequate
account of human motivation. It is often simply assumed that roles
38 Introduction
or cultures or institutions affect individuals, without explaining how such
social structures work their magic on individual motivations. Some have
turned to behaviourist psychology, in the belief that its mechanisms of
stimulus and response provide the answer. But behaviourism fails to address
the inner springs of cognition and deliberation, overlooks the fact
that beliefs are part of social reality, and makes the agent a puppet of its social
environment.
Veblenian institutional economics relied on the non-behaviourist psychology
of William James and others. The concepts of instinct and habit
pointed to a fuller account of how individual motivations evolved. However,
by the interwar period, instinct–habit psychology had become displaced
by behaviourism. Behaviourists such as John B. Watson eschewed
consciousness and intentionality as ‘unscientific’ concepts because they
could not be observed directly. Veblen did not embrace behaviourism, but
many other institutionalists, including Mitchell, adopted behaviourist
psychology in the 1920s. Others, such as Commons, failed to develop any
theory of human motivation. Frank Knight placed himself in an idiosyncratic
minority by criticizing behaviourism while retaining many
institutionalist ideas. For Knight, economics had to address both individual
intentions and social institutions. In contrast, Clarence Ayres went
with the flow of opinion, by embracing behaviourism and arguing that
technology and institutions largely conditioned individuals. The initial,
Veblenian, promise of a resolution to the problem was lost.
What is required is a framework within which the transformation of
both individuals and structures can be explained. This approach must involve
explanations of possible causal interaction and reconstitution, both
from individual to structure and from structure to individual. This would
mean an explanation of the evolution of individual purposes and beliefs,
as well as an explanation of the evolution of structures. Preferences or purposes
would be endogenously formed. Their co-evolution must be examined,
without conflating one into the other. Such an evolutionary analysis
provides the means by which social theory may escape from its unsustainable
dichotomies and make further progress.
The development of social theory in the last quarter of the twentieth
century has prepared the ground for the building of such an approach. In
Agency and structure 39
25 While Hayek excluded the explanation of individual motivations simply as a consequence
of disciplinary demarcation, von Mises excluded them on the grounds of perceived limits
to explanation. Von Mises (1949, p. 16) argued that such phenomena as thoughts and
feelings ‘cannot be analyzed and traced back to other phenomena’ and thus upheld an
‘insurmountable methodological dualism’. However, even if thoughts and feelings defy
explanation, as a result of the complexity of causes involved, this does not deny the fact
that they are caused. Furthermore, in some cases, causes can be identified. For example,
enduring personality traits may relate to experiences in childhood. In addition, we can
partly explain particular mass political sentiments by economic or political events. For
example, the rise of Nazism can be partly explained by the punitive reparations imposed
after the First World War, plus subsequent unemployment and inflation.
particular, the transcendence of the old dichotomy between methodological
individualism and methodological collectivism has created space for
more sophisticated developments, including those mentioned above.
To bring evolution in, we have to learn from biology. But what we learn
is not purely biological. Darwinian evolutionary theory points to a causal
explanation of process, focusing both on causal links and changes at the
microscopic level and their consequences in terms of transformations in
structures, populations and species. This general Darwinian imperative of
causal explanation requires that the evolution of individual purposes and
beliefs must be explained as well as acknowledged. Ultimately, in principle,
Darwinian evolutionary theory assumes no entity or characteristic as
given. The emphasis is not on fixed units or relations but on the causal explanation
of processes and transformations.
In particular, as elaborated in Chapter 4 below, Darwinism involves an
ontological commitment to variation among members of a population. In
regard to the social sciences, this ‘population thinking’ reinforces the
premise that there is significant variation in personality and purposes between
individuals, and that these variations matter in the explanation of
social phenomena. Methodological individualists have long acknowledged
this variation. Population thinking is another antidote to both methodological
collectivism and any central conflation. But Darwinism does
not sustain methodological individualism either. Not only does it insist
that the individual has to be explained, but it also sustains higher levels of
theorising and analysis above that of the individual.
However, the mere mention of biology will cause many social theorists
to run for cover, or to reach for their guns. This prejudice among social theorists
is a huge barrier to further advance. The next chapter is an attempt to
forestall some likely misunderstandings and misconceptions.
40 Introduction