Emergence in Sociology – ISCTE

Loading...

Taking too long?

Reload document

| Open in new tab

Emergence

in Sociology:

Contemporary

Philosophy

of Mind

and Some

Implications

for Sociological

Theoryl

R. Keith Sawyer

Washington University

Many accounts of the micro-macro link use the philosophical notion
of emergence to argue that collective phenomena are collaboratively
created by individuals yet are not reducible to explanation in terms
of individuals. However, emergence has also been invoked by meth-
they accept the existence of emergent social
odological individualists;
properties yet claim that such properties can be reduced to expla-
nations in terms of individuals and their relationships. Thus, con-
temporary sociological uses of emergence are contradictory and un-
stable. This article clarifies this situation by developing an account
of emergence based in contemporary philosophy of mind. The phil-
osophical account is used to evaluate contradictory sociological the-
ories. Several unresolved issues facing theories of emergence in so-
ciology are identified.

THE SLIPPERY CONCEPT OF EMERGENCE

The relationship between the individual and the collective is one of the
most fundamental
issues in sociological theory. This relationship was a
central element in the theorizing of the 19th-century founders of sociology,
including Weber, Durkheim, Simmel, and Marx, and was essential, if
in many 20th-century sociological paradigms, including struc-
implicit,
tural functionalism
(Parsons [1937] 1949, 1951), exchange theory (Blau
1964; Romans 1958; Romans 1961), and rational choice theory (Coleman
1990). In recent years, this relationship has become known as the micro-
macro link (Alexander et al. 1987; Ruber 1991; Knorr-Cetina and Cicourel
1981; Ritzer 2000).

1 During the preparation of this article, I was supported in part by a N ational Academy

of Education Spencer Postdoctoral Fellowship. Direct correspondence to Keith Sawyer,
Program in Social Thought and Analysis, Washington University, St. Louis, Missouri
63130. E-mail: keith@keithsawyer.com

@ 2001 by The University

of Chicago. All rights reserved.

0002-9602/2001/10703-0001$10.00

AJS Volume 107 Number 3 (November

2001): 551-85

551

American J oumal of Sociology

However, emergence has also been invoked by methodological

Many accounts of the micro-macro link use the philosophical notion of
emergence to argue that collective phenomena are collaboratively created
by individuals yet are not reducible to individual action (Archer 1995;
[1979] 1998, 1982; Blau 1981; Edel 1959; Kontopoulos 1993;
Bhaskar
Mihata 1997; Parsons 1937; Porpora 1993; Smith 1997; Sztompka 1991;
Whitmeyer 1994; Wisdom 1970). Most of these accounts argue that al-
though only individuals exist, collectives possess emergent properties that
cannot be reduced to individual properties (cf. Brodbeck [1958] 1968).
Thus, these accounts reject sociological realism and are methodologically
collectivist. Other theorists make the stronger argument that emergence
can be used to ground sociological realism (Archer 1995; Bhaskar 1998).
indi-
vidualists in sociology and economics. Methodological
individualists ac-
cept the existence of emergent social properties, yet they claim that such
properties can be reduced to explanations in terms of individuals and
focus on micro-to-
their relationships. Methodological
macro processes is explicitly considered to be a study of how social prop-
erties emerge from individual action (Axelrod 1997, e.g., p. 4; Coleman
1987, p. 171; 1990; Epstein and Axtell1996, e.g., pp. 6-20; Romans 1964a).
For example, Romans argued that “emergence, and the nature of the
properties that emerge, are to be explained by psychological propositions,”
and he claimed that he had demonstrated this reducibility
book Social Behavior (1964a, p. 229). These sociologists draw inspiration
from economics, where emergence is conceived of as the process whereby
unintended macrosocial phenomena arise from the actions of many par-
ticipating
(Rayek 1942, 1943, 1944; Menger [1883] 1963). In
contrast to sociologists who believe that emergence is incompatible with
reductionist
them collectivist emergentists
will call
emergentists believe that macrosocial properties and laws
-individualist
can be explained in terms of properties and laws about individuals and
their relations.2

individualism’s

individualism-I

in his 1961

individuals

Thus, contemporary sociological uses of emergence are contradictory
and unstable; two opposed sociological paradigms both invoke the concept
of emergence and draw opposed conclusions. The problem arises in part~

2 Some sociologists define the micro- and macrolevels in terms of the size of social units

(e.g., Munch and Smelser 1987, pp. 356-57; Ritzer 2000, pp. 499-505). However, both
individualist and collectivist emergentists agree that the micro-macro debate must be
couched in terms of relations between properties at multiple levels of analysis, not in
terms of group size, and this is consistent with the philosophical account I give in the
second part of this five-part article. Because systems may have some properties that
are merely aggregative and others that are emergent, it does not make sense to speak
of systems or structures as emergent, but only of properties of those systems (Archer
1995, pp. 8-9; Wimsatt 1986, p. 260).

552

Emergence in Sociology

because sociologists have not developed an adequate account of emer-
gence. In this article, I make an initial attempt to develop such a foun-
dational account, with the goal of clarifying
these different concepts of
sociological emergence. To do so, I will draw heavily on a long tradition
of emergentism in the philosophy of science. Philosophical
interest in
emergence has gone in several cycles since the term was first coined in
1875 by the philosopher G. H. Lewes; I focus on emergentist theories
from the 1970s through the 1990s that have been inspired by developments
in cognitive science. Although philosophical arguments about emergence
and reducibility have focused on the mind-brain
relation, they can be
generalized to apply to any hierarchically ordered sets of properties, as
noted by many philosophers (Fodor 1989; Humphreys 1997, p. 3; Jackson
and Pet tit 1992, p. 107; Kincaid 1997, p. 76; Yablo 1992, p. 247n5).

Contemporary sociologists are not the first to be confused about emer-
the long history of the usage of the term (see Sawyer,
gentism. Throughout
it (Broad
in press a), one finds comments on the confusion surrounding
1925, p. 59; Ede11959, p. 192; Kim 1992, p. 122). In the face of almost
a century of confusion, it would be overly ambitious to resolve these issues
for sociologists in a single article; this article should be viewed as an initial
attempt to demonstrate the relevance of these philosophical debates to
sociological theory, rather than as a conclusive solution. The article format
allows only the briefest of summary treatments of complex debates in the
philosophy of mind, and I necessarily brush over many subtle differences
in presenting what most philosophers of mind agree is the current
consensus.

I begin this article by summarizing

this consensus. I then summarize
the two competing uses of emergence in sociology, beginning with indi-
vidualist emergentism and then turning
to collectivist emergentism. In
both cases, I use arguments from the philosophy of mind to evaluate these
competing theories of emergence, and I conclude that none of these the-
ories has adequately addressed all of the implications of the philosophical
account. I conclude the article by identifying several unresolved issues
facing sociological theories of emergence.

EMERGENCE IN PHILOSOPHY

The concept of emergence has a long history predating the 19th century
(see Wheeler 1928), but the term was first used in 1875 by the philosopher
George Henry Lewes. In a critique of Hume’s theory of causation, Lewes
(1875) found it necessary to distinguish between two types of effects:
resultants and emergents (e.g., 1875, p. 412). An emergent effect is not
additive, not predictable from knowledge of its components, and not de-

553

American J ournal of Sociology

composable into those components. Lewes’s classic example was of the
formation of molecules from their component atoms; hydrogen and oxygen
are the cause, and water is the effect-the properties of water are emergent
from the combination of hydrogen and oxygen.

These ideas were picked up by several British philosophers after World
War I-most
notably by Morgan (1923) and Whitehead (1926). Emer-
gentism in the 19205 rejected vitalism and dualism, accepting the mate-
rialist ontology that only physical matter existed. Higher-Ievel entities and
properties were grounded in and determined by the more basic properties
of physical matter; this was referred to as superoenience. However, the
19205 emergentists argued that when basic physical processes achieve a
certain level of complexity of an appropriate kind, genuinely novel char-
acteristics emerge; these emergent higher-Ievel properties could not, even
in theory, be predicted from a full and complete knowledge of the lower-
level parts and their relations. Further,
they could not be reduced to
properties of the parts and their relations, even though those properties
are supervenient on and thus determined by the system of parts (Kim
1993b, p. 134; Teller 1992, pp. 140-42).

Philosophers of mind turned to emergence beginning in the 19605, fol-
lowing the cognitivist rejection of behaviorism. The cognitive revolution
reactivated a 19th-century debate between identity theorists and dualists.
Identity
theorists hold to the reductionist and eliminativist position that
the mind is nothing more than the biological brain, and dualists hold that
the mind and the brain are distinct. Emergence has been perceived as a
third path between dualism and identity theory (Beckermann, Flohr, and
Kim 1992; Horgan 1993; Humphreys et al. 1997; Kim 1993a), and this
third path is generally known as nonreductive materialism (Kim 1992).3
N onreductive materialism holds that mental properties are not reducible
to physical ones (Davidson 1970; Fodor 1974) and may indeed have causal
power over the physical brain (Andersen et al. 2000; Heil and Mele 1993).
Although nonreductive materialism is widely accepted, its acceptance is
not universal, and emergence continues to be debated in the philosophy
of science, as indicated by several recent journal special issues (lntellectica
1997, no.25; PhilosoPhical Studies, August 1999; Philosophy of Science
suppl., 1996). In fact, just as methodological
individualists claim that
emergentism is compatible with their stance, some philosophers of science
likewise argue that emergentism is compatible with reductionism (e.g.,
Kim 1993a; Wimsatt 1997).

In the 19905, emergence became one of the core concepts in compu-

J In philosophy of biology as well, the dominant view is emergent mechanism (Bechtel

and Richardson 1993) or physicalist antireductionism (Rosenberg 1997). Here I restrict
my arguments to the philosophy of mind, but the issues are quite similar in both cases.

554

Emergence in Sociology

tational modeling of complex systems, including connectionism (Clark
1997), artificial life (Brooks and Maes 1994; Langton 1994), and multiagent
models of social systems (Gilbert and Conte 1995; Sawyer 2001a). In this
recent formulation, emergent systems are complex dynamical systems that
display behavior that cannot be predicted from a full and complete de-
scription of the component units of the system. Canonical examples of
emergence include traffic jams, the colonies of social insects, and bird
flocks. For example, the V shape of the bird flock does not result from
one bird being selected as the leader, and the other birds lining up behind
the leader. Instead, each bird’s behavior is based on its position relative
to nearby birds. The V shape is not planned or centrally determined; it
rules. The bird flock demonstrates
emerges out of simple pair-interaction
one of the most striking
features of emergent phenomena: higher-level
regularities are often the result of quite simple rules and local interactions
at the lower level.

I
To elaborate these various theories of emergence, in the following
briefly summarize the current emergentist consensus position in the phi-
losophy of mind. This nonreductive materialist argument is grounded in
the philosophy of science tradition and focuses on the terms, concepts,
laws, and theories associated with a scientific discipline. In this tradition,
the question of reductionism is not only an ontological question about the
putative existence of higher levels of analysis, but it is often formulated
as a question about scientific laws, concepts, and terms: Can a law or
concept from psychology be reduced to a neurobiological law or concept?
The nonreductive materialist argues that there are strong grounds for
believing that this reduction is not possible, even though there is nothing
in the universe other than physical matter. The argument is based on
supervenience, multiple realizability, and wild disjunction.

Supervenience

Most sociologists, both individualists and collectivists, try to avoid hy-
postatizing or reifying social groups; they accept that the only real entities
are individuals. This position is known as ontological individualism:
the
ontological position that only individuals exist. The emergentist argument
of nonreductive materialism starts with a parallel ontological assumption:
all that exists is physical matter. Because there is only physical matter,
there are only physical events; thus, psychological events are the same
events as neurophysiological events. This is known as the token identity
thesis: any token psychological event is identical to a physical event. Token
event identity entails that emergent higher-Ievel properties superoene on
the system of lower-Ievel components (Davidson 1970; Fodor 1974; Kim
1993b). Supervenience refers to a relation between two levels of analysis

555

American J ournal of Sociology

and states that if two events are identical with respect to their descriptions
at the lower level, then they cannot differ at the higher level. If a collection
of lower-Ievel components with a given set of relations causes higher-Ievel
property E to emerge at time t, then on every other occasion when that
same collection of components in that same set of relations occurs, E will
again emerge. Note that this implies that an entity cannot change at a
higher level without also changing at the lower levels.

Several philosophers of social science have suggested that the individ-
ual-collective relation is one of supervenience (Bhargava 1992, pp. 62-68;
Currie 1984, p. 357; Kincaid 1997; MacDonald and Pet tit 1981, pp.
119-20, 144-45; MelIor 1982, p. 16; Pet tit 1993, pp. 148-54). However,
most of these philosophers have argued that supervenience is compatible
with methodological
individualism and that it does not entail the irre-
ducibility of the social. In fact, philosophers of mind generally agree that
supervenience alone is not an argument for irreducibility of the mental
(Bunge 1977; Heil 1998; Heil 1999; Humphreys 1997; Margolis 1986;
Wimsatt 1997, p. 373). Supervenience is compatible with the type identity
thesis; that is, the claim that all higher-Ievel types or properties are iden-
tical to some type or property in the physical language. To develop an
consistent with supervenience, philosophers
argument for irreducibility
realizability and wild
of mind have elaborated the notions of multiple

disjunction.

Multiple Realizability and Wild Disjunction

Fodor’s (1974) influential argument against reductionist physicalism is
based on the concept of types as natural kind terms and on a certain
notion of what counts as a scientific law. A law is a statement within
which the basic terms are natural kind terms of that science. To reduce
a law to the science of the lower level, a bridge law must be identified
that translates that law. To accomplish this, each of the natural kind terms
of the higher-Ievel science must be translatable into natural kind terms
of the lower-Ievel science.

The crux of Fodor’s argument is that there is no a priori reason to
believe that this translation will be possible for any given pair of scientific
disciplines; whether or not such a reduction is possible must be determined
empirically. His argument is that a simple translation-in
his case, from
a psychological term to some combination of neurobiological terms-may
not be possible. The argument is based on the notion of multiple realiz-
ability:
the observation that although each mental state must be super-
venient on some physical state, each token instance of that mental state
might be implemented, grounded, or realized by a different physical state.
For example, the psychological term “pain” could be realized by a wide

556

Emergence in Sociology

range of different neurobiological
terms and concepts, and each token
instance of “pain” might be realized by a different supervenience base.
Multiple
is thus an account of how one could accept token
identity and yet reject type identity.

realizability

Multiple

realizability alone does not necessarily imply irreducibility;

if
there are only a few realizing states, or if those states display some common
features, the reduction may not be problematic. However, reduction would
if the neurobiological equivalent of a psychological term were
be difficult
an otherwise unrelated combination of many neurobiological concepts and
terms (see fig. I). Fodor termed such a realization wildly disjunctive.
If
a higher-Ievel property is realized by a wildly disjunctive set of lower-
level properties, then the physical equivalent of a psychological law must
contain wildly disjunctive
terms. Fodor argued that a true scientific law
cannot have wildly disjunctive components and that wild disjunction thus
implied that there could be lawful relations among events, described in
psychological language, that would not be lawful relations in the language
of physics. Whether or not one holds to this definition of a law, it is clearly
of limited scientific usefulness to have laws with wildly disjunctive terms,
because they provide only limited understanding of the phenomena; they
are of limited predictive usefulness, because they apply only to a specific
is likely to be more generally
token instance, whereas the higher-Ievellaw
applicable. Such reductions can nonetheless be useful to explain exceptions
to the higher-levellaws; Fodor’s argument explains why laws in sciences
other than physics always have exceptions.

When supervenience is supplemented with the argument for wild dis-
junction-the
observation that a single higher-Ievel property might be
realized by many different lower-level supervenience bases and that these
different supervenience bases may have no lawful relations with one an-
have an account of emergence that shows why certain social
other-we
properties and social laws may be irreducible. There may be a social
property that in each token instance is supervenient on a combination of
individual properties, but each token instance of that property may be
realized by a different combination of individual properties. Many social
properties seem to work this way. The collective entity that has the social
property “being a church” also has a collection of individual properties
associated with each of its component members. For example, each in-
In may hold properties “believing in Xn” or “intending Vn,” where
dividual
the sum total of such beliefs and intentions are (in some sense) constitutive
of the social property “being a church.” Yet the property of “being a
church” can be realized by a wide range of individual beliefs and dis-
positions. The same is true of properties such as “being a family” and
“being a collective movement.” Microsocial properties are no less multiply
realizable: examples include “being an argument,” “being a conversation,”

557

American Journal of Sociology

Higher-Ievellaw

SIX

S2X

Disjunctive properties of

PIX orP2x orP3x

…PoX

~ p]*X OrP2*X OrP3*X

Pn*X

reducing science

FIG. 1.-Wild

disjunction and the reduction of higher-Ievellaws

and “being an act of discrimination.”
interest to sociologists seem to have wildly disjunctive

In fact, most social properties of

individual-Ievel

descriptions.

Emergentism does not claim that all higher-Ievel properties are irre-
ducible; some of them are predictable and derivable from the system of
lower-Ievel components. Only in cases where the relation between higher-
and lower-Ievel properties is wildly disjunctive beyond some threshold of
complexity will
reducible.
Whether or not this is indeed the relation between any given set of higher-
and lower-Ievel properties is an empirical question to be determined by
empirical study.

the higher-Ievel property not be lawfully

Downward Causation

Irreducible emergence and social causation have always been linked in
([1895] 1964) emergentist account of the autonomy
sociology. Durkheim’s
of sociology was foundationally based on emergent (or “sui generis’) social
properties having causal force on the individual. His defining criteria of
the social fact was its external constraint on individuals. This is a meth-
odological claim, not necessarily an ontological one: if we can identify
that a phenomenon has causal power, then we must treat it as real.

Durkheim’s

theory of social causation was famously criticized for seem-
ing to hypostatize the social. Likewise, higher-Ievel causation is problem-
atic in the philosophical account of emergence. Several philosophers have
argued that this account does not provide for the appearance of mental
causation, claiming that in nonreductive materialism,
the mental is epi-
phenomenal (e.g., Kim 1992, 1993a; Lowe 1993). Consequently, many
philosophers of mind have attempted to extend nonreductive materialism
to allow the mental to be more than epiphenomenal. Although nonred-
uctive materialists accept supervenience, many of them also hold that

558