Review of International Studies (2002), 28, 23?37 Copyright ? British International Studies Association
23
1
I am deeply indebted to Tore Fougner, Eve Garrard, John Horton, Matthew Paterson, Erik Ringmar,
Oliver Stocker, Rob Walker, Alex Wendt, and the two referees of the Journal for their valuable
comments on the earlier drafts of this article, which was initially presented as a paper to my Theory
Group colleagues at the School of Politics, International Relations and the Environment at Keele
University on 28 October 2000.
2
See Wendt, Social Theory of International Politics (Cambridge: Cambridge University Press, 1999).
Hereafter STIP.
3
Compare Hayward R. Alker, ?On Learning from Wendt?, Review of International Studies, 26 (2000),
pp. 141?50 and Andreas Behnke, ?Grand Theory in the Age of its Impossibility: Contemplations on
Alexander Wendt?, Cooperation and Conflict, 36 (2001), pp. 121?34.
4
See, in addition to the two items cited in n. 3 above, David Campbell, ?International Engagements:
The Politics of North American International Relations Theory?, Political Theory (June 2001);
Roxanne Lynn Doty, ?Desire All the Way Down?, Review of International Studies, 26 (2000), pp.
137?9; Patrick Thaddeus Jackson, ?Constructing Thinking Space: Alexander Wendt and the Virtues
of Engagement?, Cooperation and Conflict, 36 (2001), pp. 109?120; Robert Keohane, ?Ideas Part-Way
Down?, Review of International Studies, 26 (2000), pp. 125?30; Stephen D. Krasner, ?Wars, Hotel
Fires, and Plane Crashes?, ibid., pp. 131?6; Friedrich Kratochwil, ?Constructing a New Orthodoxy?
Wendt?s ?Social Theory of International Politics? and the Constructivist Challenge?, Millennium:
Journal of International Studies, 29 (2000), pp. 73?101; Ronen Palan, ?A World of Their Making: An
Evaluation of the Constructivist Critique in International Relations?, Review of International Studies,
26 (2000), pp. 575?98; Steve Smith, ?Wendt?s World?, ibid., pp. 151?3; and Maja Zehfuss,
?Constructivism and Identity: A Dangerous Liaison?, European Journal of International Relations
(forthcoming). I have found Zehfuss?s article particularly well-researched, Kratochwil?s impressively
erudite, and Campbell?s especially thought-provoking. Smith?s review and Palan?s article were also
helpful for my purpose. The items listed here and in the preceding note collectively make a vast
number of points that cannot be summarized easily, and I will not attempt it here since it is not my
purpose to conduct a literature survey on Wendt?s reception.
On Wendt?s philosophy: a critique
HIDEMI SUGANAMI
1
Abstract. This article subjects to a close philosophical scrutiny the internal logic of Wendt?s
extensive argument regarding his work?s location in the field?in particular, how it relates to,
and differs from, the American ?rationalist? orthodoxy in IR, comprising neorealism and
neoliberalism. I argue that his empirical hypotheses regarding collective identity formation are
plausible in their own right, but that his complex philosophical argumentation, by means of
which he tries to locate his work within the American scientific orthodoxy, but away from its
individualist core, is unconvincing.
Introduction
Alexander Wendt?s Social Theory of International Politics
2
has already been
discussed widely; its reception ranges from wildly enthusiastic to openly hostile.
3
Some reviews are well researched, erudite, and thought-provoking; others are not
quite so well executed.
4
None of the existing works, however, subjects to a close
philosophical scrutiny the precise internal logic of Wendt?s extensive argumentation
regarding his work?s location in the field?in particular, how it relates to, and differs
from, the American ?rationalist? orthodoxy in IR, comprising neorealism and
neoliberalism. The purpose of this article is to fill this gap. In particular, I argue (1)
that his empirical hypotheses regarding collective identity formation seem suffi-
ciently plausible in their own right to warrant historical case studies by those who
wish to explore such an avenue, and will also help us speculate on the transformative
potentials of the contemporary international system; but (2) that his?to my mind,
rather ungainly?philosophical argumentation, by means of which he tries to locate
his work within the American scientific orthodoxy, but away from its individualist
core, is unconvincing. First I wish to outline the substantive contention of his book.
Wendt on collective identity formation 
Stripped to its core, Wendt?s view is that there is no such thing as the logic of
anarchy; that how states interact with each other depends on what kind of
international political culture they live under; that, broadly speaking, there are three
such cultures, Hobbesian, Lockean, and Kantian; and that each culture defines for
the states what their social identity or role should be?so that they are ?enemies?,
?rivals? or ?friends?, respectively. Where the existing international culture is deeply
internalized by the states, it will be difficult, though not impossible, for them to alter
their identities. This happens, according to Wendt, along a symbolic interactionist
path, in which states under anarchy not only learn to co-operate in spite of the ?free
rider? problem, but, in so doing, also alter their conceptions of Self and Other in the
direction of developing a collective identity. Wendt dedicates an entire chapter to
working out the details of how this may occur.
5
In the beginning, according to Wendt, when states first form a system, they do so
in a context free of institutional constraints, and this tends to make states that adopt
aggressive, egoistic identities prosper at the expense of those which are more other-
regarding. Because there is nothing to prevent it, the system would be dragged down
to the level of the most self-interested. This process of selection is reinforced by that
of learning: an egoistic and manipulative stance adopted by a state towards another
is likely to be reciprocated. As hostile interactions continue and come to prevail
within the system, enmity comes to be treated as the name of the game, an attribute
of the system itself, rather than that of particular states.
6
This is how an inter-
national system comes to acquire the Hobbesian culture.
Wendt?s story so far is a hypothetical one?about what would happen if states
encountered one another in a quasi state of nature?and he does not have in mind
any particular historical period or case to explain, though he would welcome having
his hypothesis tested against history.
7
He does not, however, complete his hypo-
thetical picture by giving an account of how an international system may move from
the Hobbesian culture of enmity to the Lockean one of rivalry. Instead, he explores
how the Lockean culture might be transformed into the Kantian one of friendship,
24 Hidemi Suganami
5
See STIP, ch. 7. Unless otherwise indicated in footnotes, the following summary is based on my
reading of this chapter.
6
See STIP, p. 264.
7
Ibid., p. 82.
suggesting that this is the problem faced by the international system today, because
?[w]hether or not international politics has been Hobbesian for most of history,
states managed to escape that culture some years ago?.
8
However, Wendt?s explora-
tion of the paths from the Lockean to the Kantian culture turns out also to be a
hypothetical one, pointing to a few factors which may contribute to bringing about
such a transformation, and not aimed to explain any specific case of transition.
In his discussion, Wendt notes three familiar factors which would, under appro-
priate conditions, contribute to the process of transformation from a more indivi-
dualistic culture to a culture of collectivity. These are what he calls ?three master
variables??interdependence, common fate, and homogeneity. They are presented by
Wendt, not simply as factors contributing to co-operative behaviour, but more
specifically as factors which help reshape the identity or role of states from ?rival? to
?friend?.
Interdependence would help cultivate collective identity, especially if ?someone
had the bright idea of portraying the situation as one of interdependence?.
9
Then a
discourse may begin about what ?we? should do. Wendt adds that collective identity
is more likely to result from interdependence where interaction between states is
more dense, leading to the formation of ??core areas?, around which concentric
circles of identification might then develop?.
10
He acknowledges that inter-
dependence would not lead to co-operation where the fear of exploitation is
pronounced, but suggests that, in the Lockean culture, such a fear would be less
serious than in the Hobbesian one, making it more likely for interdependence to
contribute to the creation of a co-operative scheme. Where a fear of exploitation is
not very intense, common fate, too, is said by Wendt to contribute to the emergence
of a ?We who should work together?,
11
although he adds that where the threat is not
so acute, the emergence of perceptions of common fate may depend on much
?ideological labor? by ??entrepreneurs? and/or ?epistemic community? who take the
lead in reframing how actors understand themselves?.
12
As for homogeneity, or
mutual alikeness, Wendt rightly acknowledges a number of reasons why it is not by
itself sufficient to produce prosocial behaviour, but suggests that it may contribute to
increasing the ability of Self and Other to see themselves as members of the same
group. Where, prompted by these three master variables, co-operation and collective
identity begin to prevail in the interactions of states within a system, a tipping point
will again be reached, and friendship, as opposed to rivalry, will come to be seen as
the attribute of the system itself. This is how the system comes to live under the
Kantian culture.
But none of this would happen, Wendt acknowledges, if states could not trust
that a prosocial behaviour would be reciprocated by others. How do states come
to know that other states would be self-restraining? Here Wendt suggests three
plausible scenarios. First, through repeated compliance with the norms enjoining the
peaceful settlement of international disputes, states gradually internalize the institu-
tion of the pluralistic security community. By observing each other?s habitual
compliance, states gradually learn that others can be trusted. Second, some states?
On Wendt?s philosophy: a critique 25
8
Ibid., p. 339; italics mine.
9
Ibid., p. 347.
10
Ibid., pp. 347?8.
11
Ibid., p. 352.
12
Ibid., pp. 352?3.
democratic states in particular?are strongly predisposed to settle their disputes with
each other by peaceful means. This only needs to be appreciated among such states
for them to realise that they would be self-restraining. Third, a state may reach a
moment of reflexivity, in which it is realised that other states? hostility towards it is
to some extent due to its own hostile attitude towards them. If, as a result, the state
in question unilaterally resorts to policies of self-binding or sacrificing, as did the
Soviet Union under Gorbachev?s ?new thinking?, then it might succeed in signalling
to the others that it could be trusted.
Wendt realizes that ?[i]n most situations the best that can be expected is concentric
circles of identification, where actors identify to varying degrees with others
depending on who they are and what is at stake, while trying to meet their individual
needs as well?.
13
Still, he believes that while such a progressive transformation is not
inevitable, it is likely in fact to be irreversible?barring severe ?exogenous shocks?.
14
This is because, Wendt believes, ?[w]ith each ?higher? international culture states
acquire rights?to sovereignty in the Lockean case, freedom from violence and
security assistance in the Kantian?that they will be loathe to give up, whatever
institutions they may create in the future?.
15
What Wendt has here is a set of hypothetical causal sketches of collective identity
formation in interstate relations. His substantive contribution is to have suggested a
number of possible scenarios, based on various combinations and permutations of
these sketches, of the process through which the dominant international political
culture, states? conception of themselves, and their behaviours all become increas-
ingly less egoistic. It is not my purpose to engage in empirical testing of Wendt?s
substantive hypotheses. Suffice it to express my appreciation here that they point to
quite plausible transformative potentials of the contemporary international system,
and help undermine neorealism?s immutability thesis;
16
and further that Wendt?s
approach, based on structurationism and symbolic interactionism, offers a more
comprehensive picture of the ?evolution of co-operation? than the well-known
Axelrod version.
17
I am somewhat troubled that the inequality of power among
states and the cultural diversity in the contemporary world are absent, or largely
absent, from Wendt?s considerations, but it is not my aim here to try to formulate
more plausible empirical hypotheses than those contained in his work.
As I stated at the outset, the focus of this article is to expound and criticize the
complex philosophical argumentation by means of which Wendt locates his own
approach in the field. His basic contention in this respect is that, like neorealists
and neoliberals, he is on the side of science, but that, unlike them, he is not an
individualist. In connection with the first part of this contention, he expresses his
allegiance to scientific realism, which, as I shall try to argue in the next section,
would seem redundant?at least for his purpose of demonstrating the reality of the
state. In connection with the second part, he invokes the distinction between causal
26 Hidemi Suganami
13
Ibid., p. 364.
14
Ibid., p. 312.
15
Ibid.
16
See Andrew Linklater, ?The Achievements of Critical Theory?, in Steve Smith, Ken Booth and
Marysia Zalewski (eds.), International Theory: Positivism and Beyond (Cambridge: Cambridge
University Press, 1996), pp. 279?98.
17
See Robert Axelrod, The Evolution of Co-operation, with foreword by Richard Dawkins
(Harmondsworth: Penguin Books, 1990).
and constitutive relationships, on the basis of which, as I shall explain and scrutinize
in the subsequent section, Wendt thinks he can demonstrate the fundamental
inadequacy of the individualist ontology presupposed in neorealism and neo-
liberalism. There follows another section, in which I argue that, because of his own
concession to the individualist ontology, the difference between his position and the
other two he is criticizing is a matter of degree.
Scientific realism and the ontological status of the state
Concerning those entities?such as subatomic particles?which are not observable
directly by our senses but which scientists invoke to explain physical phenomena,
there are, as Wendt explains, two standard positions to which philosophers of
science subscribe: empiricism and scientific realism. Because these entities cannot be
directly observed, empiricists do not accept that they can be said to exist. By
contrast, scientific realists argue that it is reasonable to infer the existence of these
entities as the cause of certain observable effects, given that the theory invoking their
existence, independently of our knowledge-claims about them, is our most satis-
factory explanation for those effects. If every unobservable thing invoked in scientific
theory were nothing but a fiction, scientific realists say, the success of science would
be an unintelligible miracle.
18
My position is to refuse to be forced into choosing between these two alternatives.
On the one hand, I consider that reducing the criterion of existence solely to direct
observability is misguided. It unduly narrows the range of things that can legiti-
mately be said to exist, and, in particular, I see no good reason why the verb ?to
exist? cannot be extended to cover a host of other things, ?the English language?, ?the
proof of Fermat?s last theorem?, ?the United Nations?, just to list a few candidates as
they come to mind. In any case, the idea of direct observability is itself quite prob-
lematic. I am also persuaded that the success of science could not be just a miracle,
and am willing to entertain the possibility that at least some of the directly
unobservable things invoked in scientific theory may be ?really present?, that is,
existent independently of our knowledge-claims about them. On the other hand, I
believe science to be fallible, and assume there to be no guarantee that those
unobservable theoretical entities, which science invokes in their currently best
explanation of nature, will not be rejected by science at some future point.
19
Of
course, all our scientific assertions can be construed realistically, that is, as involving
ontological commitment on the part of those who make such assertions. But
scientific assertions can be said to depict the world as it is independently of our
knowledge-claims only if they are not wrong, and we can never be sure which, if any,
scientific assertions are of this sort. In the end, I wish to remain agnostic as regards
the real (that is, knowledge-independent) existence of theoretical entities invoked in
the currently best explanation of nature. In fact, I do not understand why it is
necessary to go beyond simply acknowledging that a particular entity is invoked in
On Wendt?s philosophy: a critique 27
18
See STIP, ch. 2.
19
See, however, W. H. Newton-Smith, The Rationality of Science (London: Routledge, 1981), which
combines scientific realism with a sophisticated analysis of verisimilitude.
the currently best scientific explanation, and that implicit in the idea that this entity
explains nature is the supposition that it exists independently of our knowledge.
What I find harder to grasp is why it is necessary to invoke the doctrine of
scientific realism, which does not strike me as problem-free, in order to hold a
sensible view that states, though unobservable by our senses, are really existent. Yet
this is the line Wendt advances in his book when he discusses the reality of the state.
The following key passage needs to be quoted in full:
If on June 21, 1941 we had attributed to ?the German state? the intention to invade the Soviet
Union the next day, we would have correctly predicted the behavior of millions of individuals
on the 22nd. Without that attribution it would have been difficult, even impossible, to predict
and make sense of what was going on. The challenge for nominalists [or those who reject
scientific realism] is to explain why this is the case. If the concept of state agency is merely a
useful fiction, why is it so useful as to seem almost indispensable?
The realist has a ready answer: because it refers to a real but unobservable structure.
Drawing on the Ultimate Argument for the reality of unobservables discussed in chapter 2,
the realist could argue that it would be a ?miracle? if a concept that predicted observable
behavior so well did not refer to something real. Like quarks, capitalism, and preferences, we
know that states are real because their structure generates a pattern of observable effects, as
anyone who denies their reality will quickly find out. If John refuses to pay taxes on the
grounds that the US state is merely a fiction, then he is likely to experience consequences just
as real as he does when he stubs his toe on a table. The reasoning here is abductive: positing a
structure that is capable of intentional action is ?an inference to the best explanation? for the
patterns of behavior that we observe (chapter 2, pp. 62?3). In the realist view, any system,
whether biological or corporate, whose behavior can be predicted in this way counts as an
intentional agent.
20
Note that in the above passage Wendt does not distinguish between whether the
state, though unobservable, is real, and whether the state, though seemingly different
from a real person, still counts as an intentional agent. To deal with the latter first, I
am clearly a nominalist on this issue, as I believe that the idea of a state as an
intentional agent cannot be anything other than a useful fiction. Why this fiction is
?so useful as to seem almost indispensable? is simply that, in those cases where the
fiction appears so useful, the states concerned are highly effective. Of course, not all
states are as effective as was Nazi Germany at a moment of its highest effectiveness,
that is, when Hitler?s decision to invade another country was being implemented, as
in Wendt?s example; but, clearly, the US state is effective enough in collecting taxes,
and many other areas, to make John, again in Wendt?s example, suffer the con-
sequences of his non-payment. In the circumstances, refusing to pay taxes on the
ground that the US state is merely a fiction would make John not just a nominalist,
but a fool. John suffering the consequences does not show nominalism as such to be
an erroneous doctrine; rather it demonstrates his idiosyncratic stupidity. For if he
had been a sufficiently prudent nominalist, he would still have noted the usefulness
of going along with the fiction that the United States, like a real person, was an
intentional agent, and would have paid his taxes.
Now, it may be asked, how could a state, an effective one, get its effectiveness if its
status as an intentional agent were a mere fiction? To this, my answer is simple. A
state is effective, and hence our understanding of it as an intentional person is a
28 Hidemi Suganami
20
STIP, p. 216.
useful fiction, to the extent that individuals treat it as an important component of
the world that they inhabit. But does this not mean, after all, that as a result of the
individuals treating a state in this way, the state comes to exist in the world? I would
accept this, but only in the sense that states?effectively existing ones?belong to our
social world, and I would add that this social world of ours is ideational in the
relevant respects.
The most important thing to notice about ?the ontological status of the state? is
that the state belongs to the realm of institutional facts, and as with any social
institution, the existence of any particular state is necessarily a historical one.
Grounded in historically continuous collective consciousness, a state (an effective
one) presents itself as a historical, societal, given?or reality?to all individual
persons, inside and outside the state, even though it would be true to say that the
state would not exist, or continue to do so, if no individual ever held seriously the
belief that it existed.
21
What is striking about Wendt?s discussion of this subject is
that he seems to be in agreement with all these propositions,
22
yet much exercised
about the supposedly metaphysical nature of the idea that unobservable structures
can have effects.
But this should not have troubled him. What cannot go unnoticed at this point is
that Wendt?s own concept of causation allows him to say that ?[n]orms are causal
insofar as they regulate behavior? and ?[r]easons are causes to the extent that they
provide motivation and energy for action?.
23
Clearly, neither ?norms? nor ?reasons?
are the sorts of things that Wendt would wish to call observable structures. So,
according to Wendt?s idea of ?causation?, unobservables can and do ?cause?. And I
do not think that either of these beliefs?that norms can be causal or that reasons
can be causes?commits one necessarily to the version or aspect of scientific realism
according to which it is reasonable to believe in the objective existence of an
unobservable entity invoked in the currently best scientific explanation. It seems to
me possible to remain agnostic on this issue and subscribe to the notion of causality,
according to which following a norm or acting for a reason counts as a species of
what can legitimately be called a causal process. In saying this I commit myself to
thinking that ?norms? and ?reasons? are the sorts of things that can meaningfully be
said to ?exist? ?independently of our explanation? of the action concerned, but I think
this is sustainable while remaining agnostic about the objective existence of an
unobservable entity invoked in the currently best scientific explanation of nature.
It should be added here that in his book Wendt rejects the covering-law model of
scientific explanation in favour of the view that scientific explanation consists in
presenting a plausible causal mechanism underlying the phenomenon in question.
24
This view is also associated with, or treated as an aspect of, scientific realism. I in
fact accept this view so far as ?causal mechanisms? are understood to refer to
one kind of generative potentialities which it would seem reasonable to suppose are
embedded in the world in order for us to make sense of what happens in it. This version
or aspect of scientific realism is consonant with the idea that the transformative
potentials towards a less egoistic international political culture are not just imagined,
but really present in the world of international politics. But it seems to be only in the
On Wendt?s philosophy: a critique 29
21
See John R. Searle, The Construction of Social Reality (Harmondsworth: Penguin Books, 1995).
22
See STIP, pp. 24, 215?221.
23
STIP,p.82.
24
See STIP, pp. 81?2.
context of discussing the real existence of the state that Wendt makes an explicit use
of scientific realism, and he does so, in my view, unnecessarily?for all that he needs
to do in this context is to point to the causal potentials of collectively held ideas,
to which the discussion of the reasonableness or otherwise of the belief in the
knowledge-independent existence of certain subatomic particles would seem to me
irrelevant.
25
Causation, constitution, individualism and holism
One does not have to uphold the philosophical doctrine of scientific realism to
believe that international relations should be studied scientifically. But Wendt
professes to do the latter,
26
and his subscription to scientific realism can be seen as
part and parcel of this scientific orientation on his part. By proclaiming to be on the
side of science, Wendt has positioned himself within mainstream American
International Relations, at the core of which lie neorealism and neoliberalism. But it
is also Wendt?s wish to transcend these two closely interrelated positions, and he
pictures himself as doing so through his rejection of the individualist ontology
presupposed in these positions. My view is that while Wendt?s approach, based on
structurationism and symbolic interactionism, does produce hypotheses that go
beyond those stemming from neorealism and neoliberalism, the way in which he
tries to distinguish his position categorically from these two contains a number of
problems. In this section, I hope to demonstrate this step by step.
Wendt?s reasoning
27
The gist of Wendt?s complex reasoning, aimed at undermining neorealism and
neoliberalism in favour of his own position, is as follows. These doctrines are both
fundamentally unsatisfactory because they are based on individualism. Indivi-
dualism is in turn inadequate because it assumes that the agents?s properties?or
30 Hidemi Suganami
25
Smith also briefly remarks that Wendt?s scientific realism ?seems largely to drop out in the second half
of the book when he analyses international politics, where symbolic interactionism underlies the
argument?. Smith, ?Wendt?s World? (see n. 4 above), p. 163. In his comments on an earlier draft of this
article, Wendt alerted me to the fact that his ?concern to grant the reality of unobservables pertains
not only to the state, but also (and if not more so) to the deep structure of the states ?system??, and
questioned if my argument about the state applies there also. If by ?the states system?, Wendt means
its ?structure?, then, to the extent that his ?structure? is ?cultural?, and ?culture? is a system of collective
beliefs, my argument does seem to apply there as well. See also Kratochwil?s impressively erudite
engagement with Wendt?s book, in which he casts serious doubt on Wendt?s understanding of
scientific practice, and his rendition and employment of scientific realism, and characterizes Wendt?s
combination of scientific realism and constructivism as a failed marriage. Kratochwil, ?Constructing a
New Orthodoxy?? (See n. 4 above). My discussion here has focused more narrowly on the internal
logic of Wendt?s own argumentation regarding the need to embrace particular aspects of scientific
realism in order to make it possible to entertain the thought that states and the states system are
causally efficacious.
26
See STIP,p.39.
27
The reconstruction of Wendt?s reasoning here is based on my reading of his entire book in which
relevant aspects of his thinking are scattered. In particular, see STIP, chs. 1, 2, 4, 6, 8, but especially
pp. 77?88 and 165?84.
identities and interests?are logically or conceptually independent of the social
structure. In fact, the identities and interests of the agents are not independent of the
social structure logically or conceptually. This is because there is a constitutive
relationship between them. In a causal relationship, the cause and the effect are
logically or conceptually independent of each other; but, by contrast, a constitutive
relationship entails the presence of a logical or conceptual linkage between that
which does the constituting and that which is thereby constituted. Thus, the presence
of the structure?s constitutive impact on the agents? properties means that the agents
are not logically or conceptually independent of the structure, and this, according to
Wendt, reveals the inadequacy of individualism, and with it, the fundamental
unsatisfactoriness of the two doctrines concerned. Central to this line of thinking is
the distinction between causation and constitution, based on the belief that a logical
or conceptual linkage is absent from the former, while present in the latter.
I sympathize with Wendt?s desire to undermine the two dominant paradigms of
IR at a stroke. It would be exceedingly satisfactory for a metatheorist of a particular
discipline if she or he could come up with anything like a single conceptual
distinction, the neglect of which can be shown to have helped false doctrines to
flourish, and Wendt?s strategy here is of that sort. However, I am not persuaded that
it works well and want to show why. For this purpose, I wish first to analyse the
nature of ?causation? and ?constitution? in some depth as I find Wendt?s own ren-
dition somewhat lacking in precision.
Causation and constitution
In his discussion of causation, Wendt does not clearly separate an analysis of causal
relations from that of causal statements. The former requires us to speculate on what
goes on in the world, independently of us, when what we call, or rightly recognize as,
?causing? goes on; it asks for a rationally defensible interpretation of what ?causa-
tion? as such is (to which one possible answer of course is that it is ?nothing??for
?causation?, or ?causal necessitation?, is all in the mind, according to this ?causal
irrealist? view). The latter, by contrast, requires us to explicate our understanding of
what it means to say that something causes or caused something else.
28
Wendt does
not in fact say much about the first subject, but he suggests that causal processes are
mechanistic ones producing effects, somewhat analogously to the workings of
machines.
29
He gives more space to the second subject, and states as follows: ?In
saying that ?X causes Y?, we assume that (1) X and Y exist independent of each
other, (2) X precedes Y temporally, and (3) but for X, Y would not have occurred?.
30
My only objections to this analysis are (1) that he might have clarified whether
this is meant to be an analysis of a singular causal statement, regarding the occur-
rence of a particular event at a particular time, or that of a general one, regarding a
regular causal sequence; and (2) that he might have drawn attention to the danger of
using symbols like ?X? and ?Y? in this context, which is to obscure the fact that the
On Wendt?s philosophy: a critique 31
28
See J. L. Mackie, The Cement of the Universe: A Study of Causation (Oxford: Clarendon Press, 1974).
29
See STIP, pp. 81?2.
30
STIP,p.79.
events thereby represented are describable in many ways. The second point is parti-
cularly pertinent because, under some descriptions, the cause-event and the effect-
event can be seen to have a logical or conceptual relationship?as can be seen, for
example, if we are to say ?the heat hitting the reflector at the back caused it to be
projected forward?. This shows that Wendt?s independence requirement, stated in (1),
is not in fact reducible to the claim that there should not be any logical or concep-
tual connection whatsoever between the cause-event and the effect-event, or more
precisely, that there should not be any descriptions available of the two events, such
that, under these descriptions, the two are seen to be logically or conceptually linked.
This realization would lead to the view, to which I subscribe, that the independence
requirement should be interpreted to mean, innocuously, that the cause-event and
the effect-event are simply different events. It is always ?something else? that ?some-
thing? causes?it is one of the conditions of our readiness to attribute a causal
statement to a particular sequence of events that there is indeed a sequence of two
separate events. It is not precluded that these events, under suitable descriptions,
have a logical or conceptual link?as suggested by the reflector example above.
31
As for ?constitution?, Wendt suggests that there are a variety of constitutive
relationships,
32
but in his book he draws special attention to the holist variety with
respect to the social world whereby ?social kinds? are constituted by, or embedded in,
?discursive structures?. He offers some helpful examples: ??treaty violations? are
constituted by a discourse that defines promises, ?war? by a discourse that
legitimates state violence, ?terrorism? by a discourse that delegitimates non-state
violence?.
33
Such constitutive relationships are on the plane of socially accepted
ideas, and the presence of such a relationship can be appreciated when, and only
when, the logical or conceptual connection, accepted within the society, between the
items so connected is understood.
The discussion above points to a somewhat neater way of distinguishing a causal
from a constitutive relationship than I believe I found in Wendt?s analysis: whereas
the former is a mechanistic relation of production, the latter is a relation of logical
or conceptual enablement. For my part, I would add that causal relations also
obtain when agents perform intentional acts to bring about some consequences.
Wendt will not object to this, given his conception of an intentional, ?auto-genetic?,
agent, which I shall discuss later in this article. In short, a causal relationship has to
do with mechanistic or intentional bringing about of an effect-event, whereas a
constitutive relationship has to do with making something (or a particular
description of something) logically or conceptually possible.
34
Now, Wendt gives much prominence to a constitutive relationship in his dis-
cussion because he believes, and wants to show, that the agents? identities and
interests are constituted by the social structure, and because for his purpose it is
imperative, he thinks, that the relationship could be demonstrated to be of a
constitutive kind as he defines it. As he moves to a more substantive discussion of
32 Hidemi Suganami
31
See Mackie, The Cement of the Universe (see n. 28 above), p. 32; Donald Davidson, ?Actions, Reasons,
and Causes?, in his Essays on Actions and Events (Oxford: Clarendon Press, 1980), pp. 3?19, at p. 14.
32
See STIP, pp. 83ff.
33
STIP,p.84.
34
This is the most lucid I can make of the distinction between the two kinds of relationship that Wendt
invokes at key junctures in his argument. Wendt?s category of ?constitutive theorizing? is in fact wider
than is suggested by the way the distinction is explicated here. See STIP, pp. 83ff. But this does not
seem to me to affect the argument that follows.
international politics in the final chapters of his book, he identifies three cultures of
anarchy, or three discursive structures under which states may find themselves.
These, as we saw, are labelled Hobbesian, Lockean, and Kantian. It is his key
contention that these cultures not only causally influence the lives of states living
under them or coming into contact with them, but they each constitute for the states
their identities, or roles, distinctive of each culture. Thus, the Hobbesian culture is
said to constitute states as enemies, the Lockean culture as rivals, and the Kantian
culture as friends.
35
But I am not sure that Wendt?s argument works.
Mutual constitution or equivalence
My reason for doubting that, for example, the Hobbesian international culture can
be said to be constitutive of state identity of enmity is a logical one. I take it that in
this context ?the Hobbesian international culture? and ?state identity of enmity? are
shorthand expressions, respectively, for (1) ?the international system having the
Hobbesian culture (and neither of the two other cultures)? and (2) ?states having
enemyhood (and neither of the two other roles) as their identity?. We should recall,
however, that, according to Wendt, the constitutive process works in both directions
here.
36
I take this to mean that, in Wendt?s thinking, (1) is logically necessary for (2),
but (2) is also logically necessary for (1). This makes me think that (1) and (2) are
not in fact mutually constitutive, but different descriptions of one and the same
situation, just as ?A becoming the wife of B? and ?B becoming the husband of A? are
two different?one male-centric, the other female-centric?descriptions of one and
the same event, their marriage.
In suggesting that (1) and (2) are different descriptions of one and the same
situation, however, I am not making a generalized claim that structural and agential
properties are identical, or that one can be reduced to the other. But, in this
particular instance, I fail to see the difference between (1) and (2), which strike me as
two ways of making the same point?one at the system-level, the other at the unit-
level. The Hobbesian international culture just is states having enemyhood as their
identity, and states having enemyhood as their identity just is the Hobbesian
international culture.
Back to causal relations
It may be noted that, if the Hobbesian international culture and state identity of
enmity are one and the same thing, and cannot therefore be mutually constitutive,
neither can there be a causal relationship between them because, as we saw, such a
relationship demands that the cause and the effect be different events (or situations).
However, it is a reasonable causal hypothesis that the presence or dominance, in a
given international system, of the idea that under anarchy states are inevitably in the
On Wendt?s philosophy: a critique 33
35
See STIP, ch. 6.
36
Ibid., pp. 171, 180, 259, 260.
Hobbesian state of ?war of all against all? would tend to make states think of
themselves as enemies and behave in a hostile way towards each other, make it
difficult for them not to do these things, and encourage those entering the system to
follow the pattern. Would we have to abandon this reasonable causal hypothesis as a
consequence of rejecting Wendt?s constitutive thesis? 
Fortunately not. We can hold on to the definitional linkage between the
Hobbesian international culture and state identity of enmity, and yet, without
contradiction, subscribe to a causal hypothesis that the Hobbesian idea (that under
anarchy states are inevitably enemies all against all), if present or dominant in an
international system, tends to make states think of themselves as enemies and
behave in a hostile way towards each other, make it difficult for them not to do these
things, and encourage those entering the system to follow the pattern. This causal
hypothesis, it will be noticed, relates, as it should, two items which are not one and
the same thing: on the one hand, the presence or dominance, within a given system,
of a general idea about the consequences of anarchy; and on the other, the tendency
of the individual units to act out the scenario.
Over and above this plausible causal hypothesis, which Wendt himself accepts,
there is, I submit, only his definition of the Hobbesian culture as the culture of
enmity. It is entirely consistent with this assessment that, when he moves to flesh out
his allegedly constitutive thesis regarding the relationship between international
political cultures and state identities, he ends up offering a number of straight-
forwardly causal narratives.
37
Wendt?s concession to individualism
I have argued that Wendt?s strategy?to undermine neorealism and neoliberalism by
demonstrating individualism to be inadequate and to do so by pointing to the
constitutive relationship between the existing cultural structure of the international
system and the states? role or identity associated with the culture?contains a
number of moves that cannot go unchallenged. I believe at least some of my
challenges would persuade the reader to think again about the efficacy of Wendt?s
strategy. But I now want to draw attention to two other things. First, Wendt is quite
persuasive in his portrayal of how in the course of interaction states may be caused
to alter not only their behaviour but also their identity.
38
Since this is a major
contribution to our understanding of the mechanisms of the evolution of co-opera-
tion between states (neglected by the neorealists, and only partially understood
within the neoliberal paradigm), Wendt does not have to drive himself quite so hard
to cause further damage to the two theoretical positions which his own in substance
supersedes. Second, his drive in this regard to criticize the individualist basis of
neorealism and neoliberalism is somewhat misguided because, notwithstanding his
34 Hidemi Suganami
37
Ibid., pp. 274ff.
38
Ibid., ch. 7.
stress on holism, he makes very important concessions to the individualist ontology
himself. In this section, I want to say a few words about this last point.
Wendt on individualism and holism
In a clearly formulated passage, Wendt contrasts individualism and holism as
follows:
The individualist hypothesis is in effect that all identities are personal identities, all interests
personal interests, all behaviors meaningful because of personal beliefs. Nothing in or about
the actor or his behavior logically or conceptually presupposes other actors or culture. The
holist hypothesis is that culture constitutes role identities and their corresponding interests
and practices. Regardless of the thoughts in one?s head, one cannot be a certain kind of
agent, or engage in certain practices, unless these are recognized by others. If holists are right
then it will be impossible to reduce society to independently existing idiolects, as required by
the individualist view that thought is logically prior to society. Individualist approaches to
social inquiry may still be useful for some questions, but will be inherently incomplete insofar
as they presuppose irreducible societal facts. If the holist is right, in other words, we will have
to revise our conventional view of intentional agency, which is rooted in individualism, if not
jettison it altogether.
39
While being critical of individualism, Wendt is equally unwilling to subscribe to
radical holism, according to which ?intentionality is merely an effect of discourse?,
40
or ?motive and intentionality [far from being causes of behaviour] actually refer to
the public criteria by which we make behavior intelligible, by which we make ascrip-
tions of motive?.
41
What Wendt prefers is a moderate holism, which means the same
as a moderate individualism. It is ?not that culture does not help constitute the
meaning of an agent?s desires and beliefs, but that agents have a role to play in social
explanation which cannot be reduced to culture?.
42
But in coming to terms with the
need to accommodate individualism, Wendt makes remarks which are worthy of
attention. According to him:
[It is possible to distinguish] between individuality per se and the social terms of individuality.
The former refers to those properties of an agent?s constitution that are self-organizing and
thus not intrinsically dependent on a social context. Some of these properties are material:
individuals live in genetically constituted bodies that do not presuppose other bodies, and
have minds in virtue of independent brains. Others are cognitive: agents exist partly in virtue
of their own thoughts, which they can continue to have even if they are marooned on a desert
island. Both kinds of properties are essential to intentional agency, and, even if they are
caused by society, they exist independent of them. They give the Self an ?auto-genetic?
quality, and are the basis for what Mead called the ?I?, an agent?s sense of itself as a distinct
locus of thought, choice, and activity. Without this self-constituting substrate, culture would
On Wendt?s philosophy: a critique 35
39
Ibid., p. 178.
40
Ibid.
41
Ibid., p. 179; Wendt?s italics.
42
Ibid., p. 181.
have no raw material to exert its constitutive effects upon, nor could agents resist those
effects. The intuitions that sustain individualism are rooted in this aspect of individuality.
43
Wendt?s Cartesianism
What is striking in the above passage is Wendt?s apparent subscription to the view
that we have a sense of ourselves as a distinct locus of thought, choice, and activity,
and that we have a presocietal individuality which enables us to be intentional
agents. Unlike rocks, he says, human beings are intentional agents that ?exist partly
in virtue of their own thoughts?.
44
The point here could be that the actual content of
our thoughts are socially constituted and produced, but that, nevertheless, such
thoughts are ours in the very restricted sense that we are conscious that the thinking
is going on inside our head. But that would not distinguish Wendt from radical
holists, for whom, when we think we are doing the thinking, it is in fact the culture
that is doing it; but, as we saw, Wendt professes not to subscribe to such a radical
position. It seems to follow from this consideration that here Wendt is saying that we
are indeed capable of thinking the thoughts not constituted by our culture. In short,
agents are not constituted by the discursive social structure through and through. It
is in conformity with this line of thinking that Wendt also accepts that there are
certain pre-given human traits which are part of human nature,
45
and, in a parallel
fashion, he talks of certain objective national interests inherent in statehood.
46
It
seems to me, however, that if Wendt is willing to make this degree of concession to
individualism, then his effort to undermine neorealism and neoliberalism because of
their rootedness in individualism begins to lose its edge. The difference between
Wendt, on the one hand, and Waltz and Keohane, on the other, is a matter of
degree. For Wendt the difference is no doubt highly significant; for the more radical
holists it is less so.
47
36 Hidemi Suganami
43
Ibid., p. 181?2. The penultimate sentence here, incidentally, is intriguing. If a constitutive relationship
is, as Wendt says, one of logical or conceptual enablement, not like a billiard ball hitting another, it
makes no sense to talk of the constituting element exerting effects upon the element thereby to be
constituted, or the latter resisting the effects exerted by the former. ?Exertion? and ?resistance? may be
appropriate metaphors in some causal relationships, but are not possible in logical or conceptual
relationships. Wendt?s reference here to ?exertion? and ?resistance? would seem to reveal that, at least
part of the time, he is not distinguishing the two kinds of relationships clearly. Likewise, Smith says
that he is ?troubled by his [Wendt?s] notion of causal and constitutive theories?, that, in his view,
?Wendt?s view of constitutive theorizing is very different to the dominant use of the term
?constitutive? in the social sciences, where it tends to be contrasted to explanatory theory?, and he
considers ?Wendt as seeing constitutive theory as a form of causal theory?. Smith, ?Wendt?s World?
(see n. 4 above), pp. 156, 157.
44
STIP, p. 181; see also p. 225.
45
Ibid., pp. 131?3.
46
Ibid., pp. 233?8.
47
The individualist basis of symbolic interactionism, followed by Wendt, has been noted by Palan,
?A World of Their Making? (see n. 4 above).
Concluding remarks
As Friedrich Kratochwil has observed in his review of Wendt?s book, ?[a] work of
such scope and complexity is likely to engender a series of questions that arise from
ambiguities in the argument, or out of mistakes in logic or fact?.
48
As Steve Smith
has noted, however, Wendt?s is a ?sophisticated book, one that has clearly been
drafted and redrafted so as to refine the argument and anticipate many of the likely
objections ? Criticism of the book is not an easy task?.
49
My critical engagement with this book has focused on the way it characterizes
itself, rather than upon its substantive contentions. As far as the latter goes, it is my
view that Wendt?s exploration of transformative potentials in the contemporary
international system places him above Waltz and Axelrod. But in the course of my
detailed engagement with the self-characterization of the book, I hope to have
eliminated some ambiguities in the argument and pointed out some mistakes in its
logic.
In particular, I have argued that Wendt?s invocation of scientific realism is
unnecessary for the purpose of ascertaining the reality of the state; that his
discussion of causal and constitutive relationships could be tightened; that his
demonstration of the specifically constitutive relationship between a given inter-
national political culture and the corresponding identity of the states is uncon-
vincing; and that his castigation of individualism as the main source of the
inadequacies of neorealism and neoliberalism is marred by important concessions he
makes to it.
On Wendt?s philosophy: a critique 37
48
Kratochwil, ?Constructing a New Orthodoxy?? (See n. 4 above), p. 87.
49
Smith, ?Wendt?s World?, p. 151.