A tale of two cognitions : The Evolution of Social Constructivism in International Relations

Constructivism in International Relations (IR) is popular, but constructivists seem disappointed. Allegedly something has been lost. Such criticisms are misplaced. There was never a uniform Constructivism. Since constructivism is socially constructed, to argue that constructivism has evolved “wrongly” is odd. This paper explains the dissatisfaction with constructivism followed by a second reading of its evolution as a tale of two cognitions. These two cognitions distinguish genera in the constructivist “family”. A criticism against one genus based on the cognition of the other is unfair. A focus on cognitions and the use of genera helps in perceiving constructivism’s future evolution.


Introduction
T he contemporary state of social constructivism in International Relations (IR) is somewhat paradoxical 1 .Constructivism is well established and popular, but many constructivists seem unhappy.The common claim by dissatisfied constructivists is a variation on a theme about the mainstreaming of constructivism and how something important was lost in that process.Yet, as this article argues, such claims are misplaced.For one, there was never some uniform IR "Constructivism," which then lost something.Rather, constructivism has been characterized by plurality from the start.Second, dissatisfaction with contemporary mainstream constructivism risks being based on some notion of "real" constructivism, which in turn is difficult to square with an important constructivist tenet -social construction -which must apply also to social constructivism.It is odd to argue that the evolution Rev. Bras. Polít. Int., 60(1): e014, 2017 Peltonen 2 of constructivism has somehow been "wrong," because social constructivism is itself socially constructed, thus implying that there is no "right" or "wrong" way for it to evolve.Third, IR constructivism experienced evolutionary branching resulting in two "genera" in the constructivist "family".I distinguish the genera depending on the kind of cognition behind them.Because the history of IR constructivism is a tale of two cognitions, assessing one genus with the cognition used by another genus explains contemporary disappointment with constructivism, but it also highlights how such dissatisfaction is misplaced.
The brief remarks above contain two key concepts which I use in examining the history of constructivism: evolution and cognition.The former needs no further clarification besides reminding my readers that development is not a synonym for evolution.By cognition I generally refer to mental processes of knowing including perception, reasoning, and judgment.For the purposes in this paper, such a rough shorthand is preferable to the more typical vocabulary of epistemology and ontology, because, by talking about cognition, I can talk about them both together and simultaneously.By not using these loaded terms, I avoid old trenches, for instance, regarding which should take priority, which in turn would lead my discussion astray.Moreover, the concept of cognition allows me to show that mainstream constructivism might be inconsistent according to the cognition of more critical constructivists, but it is nevertheless consistent within its own cognition.Finally, cognition enables me to show that what is at stake is not something lost during the evolution of constructivism, but a branching out based on a different but legitimate process of knowing and reasoning.
To make my case, I follow a double reading of constructivism's history, in which I pay special attention to particularly influential works.The next main section provides the first reading in order to outline and explain recent dissatisfaction with constructivism.At the end of that section, I suggest that the discontent may be misplaced.The second main section provides a re-reading of the evolution of constructivism by making use of differences in cognition.The section begins by recalling that constructivism forms a kind of Wittgensteinian family that is socially constructed.As a second step, I argue that despite their significant differences, such early constructivists as Onuf and Kratochwil shared a kind of cognition, a way of reasoning about the world and ourselves in it.This differed from the contemporary mainstream cognition that influenced some constructivists, thus resulting in an evolutionary branching of constructivism.At the end of the paper conclusions are drawn from this re-reading of constructivist evolution in order to provide additional support for the argument that dissatisfaction with mainstream constructivism is misplaced.The conclusion also explains some of the benefits of using the vocabulary which I use in the double reading of constructivism's history, one being the anticipation of future evolution.

Dissatisfaction with constructivism
By the turn of the millennium, Constructivism had replaced Marxism in the IR "trinity" (Walt 1998, 38) 2 .For some, constructivism's rise was a "success story" (Guzzini 2000); for 2 As an example, consider that Fritsch (2011) dems (neo)realist, liberal, and constructivist contributions to the role of technology in global affairs, but he ignores Marxist insights to technology and alienation, although they are important themes, as is discussed by e.g.Wendling (2009).Note, though, that including constructivism in such a trinity often leads to the misunderstanding that constructivism is a substantial IR theory similar to, for instance,.structural realism (see Kurowska and Kratochwil 2012, 87).others it was "trendy" (Checkel 2004).Social constructivism's journey in IR, from an early position of "dissent" (George and Campbell 1990) to "normal science" (Onuf 2016), has been marked with a number of key contributions, of which two stand out in a particular fashion.Adler (1997) and Wendt (2010) consolidated constructivism and opened it to a wider audience by staking a "middle ground".They emphasized constructivism as via media.The aim was to build bridges between constructivism and other approaches, to overcome the abyss between rational and reflective (epistemological) positions.This so-called mainstreaming of constructivism enabled this perspective to solidify its presence and to move from a position of "dissent" to "normal science"3 .
Yet, recently many have argued that success has come with a price.To explain, I need to take a brief look at the past evolution of constructivism4 .I pay special attention to the publications, which others also have identified as crucial in mainstreaming constructivism.

Early constructivism
In the beginning there was no "constructivism" in IR.Rather, many diverse approaches were grouped under two general headings: rational and reflective (Keohane 1988).The latter included what became known as constructivism.While no single individual alone can be credited for the emergence of social constructivism in IR, Nicholas Onuf (1989) coined the term for our field, and he is often considered as one of the "fathers" of IR constructivism (e.g.Pouliot 2004, 323;Griffiths et al. 2009).The other "parents" of IR constructivism can be found, for instance, in Wendt's (2010, 3-4) classification of early IR constructivism.According to him, IR constructivism in the 1980s could be divided into three "streams," where John Ruggie and Friedrich Kratochwil represented modernist constructivism, Richard Ashley and R.B.J. Walker postmodernist constructivism, and Spike Peterson and Ann Tickner feminist constructivism5 .Putting it differently, it was clear from the beginning of the IR constructivist story that no single approach had a monopoly on how social constructivism was understood in IR -something to keep in mind 6 .
Thus, the IR constructivist family tree has had many roots and (intertwined) stems from the start, and there is no reason why things should now be any different.Here, though, I draw briefly from two influential constructivist books that preceded the consolidation of constructivism.Onuf (1989) coins the term "constructivism" for IR in his World of Our Making and reconstructs the field through an examination of rules and rule, both in social theory and in IR.For the constructivism he represents, in the beginning there was and is the deed.One does not begin with facts, because facts are A tale of two cognitions Rev. Bras. Polít. Int., 60(1): e014, 2017 Peltonen 4 not aspects of objective knowledge in the sense that they somehow display an inherent correspondence to reality.Rather, facts, whether social or natural ones, are constructions.The reason why science has a history is exactly because no fact is immune to deconstruction (Fuchs 1992).Statements become facts through the "deletion of modalities" (Latour and Woolgar 1986) that may take several forms.When it becomes too costly to deconstruct them, they become taken for granted or "black boxes" (Fuchs 1992, 46).Thus, because facts are a poor starting position, Onuf begins with deeds: "people and societies construct, or constitute, each other" (Onuf 1989, 36, original emphasis).
Onuf 's constructivism bypasses the choice between the world and the word.Epistemological realism would emphasize the world and its existence independently of the mind.The word refers to how the mind orders the world.Onuf 's constructivism recognizes the distinctions between the social and the material realities, and the importance of both, but, in his words, they "contaminate" each other while neither is privileged (Onuf 1989, 40).To exemplify, instead of following Ashley (1988) in seeing that one is either outside or within history, Onuf (1989, 43) asserts that we are "always within our constructions, even as we choose to stand apart from them".Here, his emphasis on the deed tries to circumvent any privileging over the word or the world, since a deed is "intelligible only as jointly a social construction and natural event, produced by mind yet phenomenal in its own right" (Onuf 1989, 43).
In Rules, Norms and Decisions (1989) 7 , Kratochwil shows how rules and norms constitute practices that in turn enable meaning; they operate as "guidance devices" (Kratochwil 1991, 7-14).Rules and norms constitute and give meaning to what a given situation is as well as how one ought to understand particular acts or choices (Kratochwil 1991, 98-100).For him, particularly important are the roles which language and discourse have in deciding cases that have no logically compelling solutions.Part of Kratochwil's insistence is that things do not have a meaning in themselves.There is no "essence" to be discovered or to be communicated.Rather, the meaning of things is constructed within their contexts, and since communication or language are the means of both relating to the world and to others in meaning-construction, one ought to pay attention to the various functions that language serves.A hierarchy of rules and norms is of no help because, like in Onuf 's argument, there are no firm foundations, because we cannot determine a priori what is relevant in all contexts.
Kratochwil's constructivism emphasizes a way of reasoning rather than a logical proof.Logic alone does not help, because we cannot predict all contexts and circumstances to which a given rule or norm might apply, and there can be equally "good" solutions in many, if not in most contexts, but in practice we must nevertheless act and decide.Like judges must come to a judgment even in "tough" cases, so must we decide over competing, incompatible choices in tough situations, not to mention to first understand in what situation we find ourselves.Rules and norms (and values to the extent that they inform attitudes) provide the basis for such decisions, but they must be accompanied by a style of reasoning familiar to law: social scientists should be more like judges.
Both World of Our Making and Rules, Norms and Decisions challenged the IR discipline without promising a grand theory of IR.They are social theories, not substantive theories of IR.Both argue that the world was not simply "out there" for us to observe (empiricism).Rather, the social world is one of artifice, continuously constructed through our own deeds, conceptualizations, perceptions, rules, and norms.Moreover, the material world was also not given.Certainly, the material world is "there" for both, but being "there" is meaningless by itself.We access and act in the world through our deeds and languages, which then convey meanings to others.Or as Michel (2009, 405) puts it, "we as humans have no possibility to encounter matter as matter independent of our concepts of the world and anything in it".It is of little (practical) interest to speculate about the material world independent of us, because we can only access and give meanings to it through different media.For both Kratochwil and Onuf, the meaning of the material world is not, -and cannot be -independent of our conceptualizations and perceptions of it.

Consolidating constructivism by building bridges
A number of IR scholars welcomed early constructivism to the field, and many important constructivist works followed (e.g.Wendt 1992;Klotz 1995;Biersteker and Weber 1996;Finnemore 1996;Katzenstein 1996;Finnemore and Sikkink 1998).Yet, constructivism remained more of a disciplinary curiosity than part of the mainstream until it was translated to a wider audience, especially by Adler and Wendt.In Seizing the Middle Ground, Adler (1997) presents how constructivism can be understood in a way that builds bridges to other IR approaches.For him, debates over constructivism were not "science versus literary interpretation or 'stories', but at stake were the nature of social science itself and, therefore, of the discipline of International Relations" (Adler 1997, 320).Some of these debates are misguided.Constructivism is not at the opposite end of a spectrum from realism or liberalism, because it is not a theory of politics, but a social theory.Rather, Adler argues, constructivism is the middle ground between structuralism and individualism, and between materialism and idealism (Adler 1997, 331, esp. figure 1).This is because constructivism focuses on individuals within social structures, and because it does not deny the material world.Instead, according to Adler, it emphasizes the role of ideas in understanding and relating to the material world.In Adler's sense, constructivism does not privilege any of the four "-isms", but rather combines their key insights about the social world.
In Social Theory of International Politics, Wendt (2010) outlines a via media constructivism, located between positivist and post-positivist approaches.One of his main concerns is ontology; we must all "do" ontology, "because in order to explain how the international system works [we] have to make metaphysical assumptions about what it is made of and how it is structured" (Wendt 2010, 370).For him, power and interest are not simply in the domain of materialism, because their meaning and content are constituted by ideas and culture.Regarding the meta-level, Rev. Bras.Polít.Int., 60(1): e014, 2017 Peltonen 6 though, Wendt rejects the claim that constructivism is incompatible with positivist epistemology."There is," for him, "nothing in the intellectual activity required to explain processes of social construction that is epistemologically different than the intellectual activity engaged in by natural scientists" (Wendt 2010, 372).Both social and natural sciences focus on explaining causal patterns and their effects.And even though there are two kinds of questions "necessary to the scientific enterprise, causal and constitutive", he claims that we can treat them as the same, more or less (Wendt 2010, 372-73).

Dissatisfaction in something lost
All four, Adler, Kratochwil, Onuf, and Wendt agree that ideas matter, but the question is how to study them, and what kind of knowledge one may obtain.For all, the social world is one of artifice, and our ideas inform us regarding the meaning of things.Yet, where Kratochwil and Onuf challenged prevailing IR mainstream notions of science, Adler and Wendt sought a dialog with the various ontological and epistemological positions prevalent within the field.And it worked; constructivism became evermore popular.Simultaneously, though, constructivism was mainstreamed, because, by following Adler or Wendt, one could hold on to one's mainstream understanding of scientific explanation, particularly causal explanation8 , without needing to embrace the more radical alternative offered by Kratochwil and Onuf.It sufficed to include into one's research program the notion that ideas, norms, or identities matter, but one does not need to adopt new methodologies or ways of reasoning.And this, to many, is cause for dissatisfaction.
According to Hynek and Teti (2010), early IR constructivism was normalized, and this process of normalizing resulted in "purging" constructivism of its early critical potential, and in a "hijacking" of the notion of critical by approaches that are neo-utilitarian.In the process, they argue, constructivism lost its early critical potential.For McCourt (2011), the thing lost is reflexivity of the social construction of our knowledge about the socially constructed objects we study.Hofferberth and Weber (2015, 75) also argue that something important was lost in translating -mainstreaming -social constructivism to the wider IR audience, namely "the insight that norms are constantly renegotiated in social interaction", which has resulted in cultural-determinist research framework ill-equipped to study and explain change.This argument is supported by Kurowska (2015), Subotic (2015), and Zarakol ( 2015), all of whom, in their own ways, show what sensitivity to positioning the researcher within a socially constructed field means in practice.Generally speaking, then, there are variations on a theme of constructivism having lost something while becoming more popular.
To illustrate these variations, I turn to a recent categorization of IR constructivisms by Kurowska and Kratochwil (2012).An examination of how constructivisms have recently been Rev. Bras.Polít.Int., 60(1): e014, 2017 Peltonen 7 categorized is useful, because it reveals the reason for dissatisfaction.Other examples would be possible, because implied dissatisfaction has also been present in earlier categorizations of constructivisms.Consider, for example, a distinction between conventional and critical constructivisms (Hopf 1998), where the implication seems to be that conventional constructivisms are not critical9 .A more neutral divide has been between modest and radical constructivisms, expressed through a rough division between modern and postmodern constructivist approaches (Price and Reus-Smit 1998), but also here arises an implication regarding "good" constructivism depending, of course, on one's view on postmodernism.Yet, I briefly examine the distinctions made by Kurowska and Kratochwil, because the labels they have already chosen are illuminating: mainstream and consistent constructivism.
For Kurowska and Kratochwil (2012, 88), mainstream constructivism is marked by a support for the scientific method due to its alleged guarantee of producing warranted knowledge.They argue that mainstream constructivism has "submitted to 'the illusion of science' … and re-opened the fundamental contradiction between intersubjective ontology of the social world and the positivist epistemology for studying it".In contrast, for them, consistent constructivism rejects the possibility that scientific realism is compatible with a consistent constructivist perspective.Where mainstream constructivists believe that there are such things as hard data, which can settle (scientific) questions, consistent constructivists argue that even so-called hard data are "based on conceptual choices and do not speak for themselves" (Kurowska and Kratochwil 2012, 88).As Kurowska and Kratochwil emphasize, unlike mainstream constructivism, consistent constructivism argues that we never test against "the world," "reality," or "the things themselves," but only against other theories or conceptions about the world or reality.For consistent constructivists, then, "truth is not a property of the 'world out there' but is always relative to the system of meaning within which particular actors are embedded" (Kurowska and Kratochwil 2012, 90).This means that "the observing subject and object cannot be separated but are mutually implicated" (Kurowska and Kratochwil 2012, 95), which contradicts the mainstream constructivist belief that a researcher can distance herself from the objects being studied 10 .
In sum, Kurowska and Kratochwil's labels express the overall dissatisfaction with the state of IR constructivism.The popular versions of constructivism, the mainstream, are inconsistent.Reversely the implication seems to be that non-mainstream versions of constructivism are consistent.

Misplaced dissatisfaction?
I sympathize with the discontent constructivists; they have identified important concerns in the state of IR constructivism.Yet, some of the dissatisfaction may be misplaced.
A tale of two cognitions Rev. Bras.Polít.Int., 60(1): e014, 2017 Peltonen 8 For one, there could easily be a marginalized version of constructivism, which is inconsistent, and at least to that extent, the labels "mainstream" and "consistent" constructivism are less than ideal.Thus, while some categories are necessary for us to structure and make sense of the world and ourselves in it, labeling is often counterproductive and inaccurate (see e.g.Fierke 2003).
Second, in this particular case of contemporary IR constructivism, there is a risk of judging all versions on the basis of a single set of criteria, and that may be inappropriate.Mainstream constructivism appears inconsistent to Kurowska and Kratochwil due to what they consider as legitimate and valid according to their cognition (see below).Mainstream constructivism could be consistent, or could be applied consistently, but according to its own cognition.
Here, one is reminded of a challenge early constructivism encountered, that constructivism, among other "reflective" approaches, should share the meta-theoretical standards of "rational" approaches (Keohane 1988)  11 .Such a requirement was unfair to early constructivism, because it would have effectively transformed a reflective approach into a rational approach; while the whole point of the earlier "dissent" (George and Campbell 1990) was to demonstrate that there are other legitimate standards, and that rational standards were not necessarily always the best.This time, though, the challenge seems to be reversed and made within the constructivist family: mainstream constructivisms are called to adhere to the same standards as non-mainstream constructivisms.This risks being an unfair requirement, because it would effectively transform them to something they are not.
The arguments that some forms of social constructivism lack something and that something was lost along the way, combined with the noticeable dissatisfaction about the current state of IR constructivism, seem to arise from a notion that there is some "proper" kind of constructivism.Instead, IR constructivism has evolved in its own ways, and much like in biology, it makes little sense to say that a particular evolutionary path was "wrong".For consistent constructivists to be consistent, they ought to accept that IR constructivism has evolved, even if not in the way that they would have preferred.Otherwise, they risk being inconsistent constructivists themselves by forgetting an important constructivist insight: social constructivism is also socially constructed.Therefore, it is subject to social re-construction and change, which is what has happened.

Another look at the evolution of IR constructivism
Given my argument about the misguided dissatisfaction regarding the current state of IR constructivism, another look at the evolution of social constructivism is important.In this re-reading, I first emphasize the social construction of constructivism by characterizing constructivisms as a Wittgensteinian "family", which allows calling diverse approaches constructivist, even if no single characteristic is shared by all.As a second step, I tell the evolution of constructivism as a tale of Rev. Bras.Polít.Int., 60(1): e014, 2017 Peltonen 9 two cognitions.The two cognitions distinguish two genera within the constructivist family12 .As evident, I draw (very) loosely on some concepts from biological taxonomy, but only for heuristic purposes in order to highlight how the evolution of social constructivism can be told without holding on to some notion of what social constructivism ought to be (and being disappointed in those forms that do not correspond with it).

A socially constructed social constructivist family
There "is indeed not one constructivist theory, but a range of constructivist positions" (Risse and Wiener 1999, 776), and there is a number of ways to order this range of constructivisms, as I briefly noted above.Within such broad categories lie subdivisions of constructivist positions, ranging from realist to feminist constructivisms and beyond, with some having representatives, for example, in both modern and postmodern "camps."The difficulty of ordering constructivisms is well illustrated in the complex "genealogical topography of constructivism" by Hynek and Teti (2010, 178).
Part of the reason for the wide variety of constructivist approaches is a lack of consensus about whether constructivism in IR is an ontology, methodology, or epistemology (Barkin 2003, 325).For instance, Risse and Wiener (1999, 778 emphasis added) argue that constructivism, while not a substantive theory of IR like (neo)realisms or (neo)liberalisms, is "a meta-theoretical approach offering an ontology which differs from, say, rational choice".In contrast, Barkin (2003;2010) has argued that realist theory is compatible with constructivist methodologies, and that constructivist epistemology is compatible with classical realist theory 13 .
Given the wide variety of constructivisms and disagreement over what it is even at the meta-level, one is tempted to ask what do they all have in common?Traditional categorization requires that all of these different approaches must share at least one characteristic in order to qualify as constructivist.
Yet, this question is difficult to answer.Consider Pouliot's (2004) argument that social facts are the "essence" of constructivism.He argues that social facts "constitute the building blocs of socially constructed reality" (Pouliot 2004, 329).They "provide constructivists with 'foundations of reality'" (Pouliot 2004, 332).For him, "social facts constitute the only 'foundations of reality' upon which constructivists can build knowledge about global politics and social life in general" (Pouliot 2004, 320).While Pouliot is correct in this argument, a problem arises if one subscribes to traditional categorization: constructivists are not the only ones to whom social facts provide the foundations of reality.Social facts are the foundation of reality also to non-constructivist social scientists and to natural scientists (see e.g.Knorr-Cetina 1981).Physics uses tools that rely on social facts and social constructions (see e.g.Ziman 1991;Fuchs 1992;Searle 2011;Kuhn 2012).Even the distinction between social and natural sciences is a socially constructed fact.Social facts A tale of two cognitions Rev. Bras. Polít. Int., 60(1): e014, 2017 Peltonen 10 provide the foundations of reality to all.Thus, while "social facts" is a "correct" answer to the question about what is shared by different constructivisms, it leaves many unhappy because the same could be said about any group of approaches.
And yet, even if I do not agree with Pouliot, I would not deny him the constructivist title if he chooses to apply it to his work.This is because the initial question is misleading.
The question asks what all constructivisms have in common, and thus it pushes us to look for at least one shared characteristic.Yet, not all constructivisms agree with such a requirement, because a number of constructivist approaches have followed, for instance, later Wittgenstein and let go of the idea of essences, the correspondence theory of truth, and exclusionary logic in categorization.In Wittgenstein's (2003 esp. §65-66) famous example, the "proceedings we call games" do not all have a common denominator, an essence, and yet they are related to each other.Wittgenstein (2003 §67) calls such relations family resemblances.This also applies to constructivism, and the various approaches in IR named constructivist need not share an essence with all others in order to count as constructivist.Instead of essences, we have family resemblances, fuzzy boundaries, and possibly non-exclusionary categorization, where one "thing" can legitimately fall under multiple categories or descriptions (already argued by e.g.Kratochwil 1989).Thus, IR constructivist approaches form a Wittgensteinian family 14 .
In sum, an important social constructivist tenet is that things are not "out there," waiting to be discovered, and this must apply also to constructivism itself.Already the diversity of IR constructivisms testifies that constructivism depends on us, the research community (or an epistemic community, Haas 1992; Adler 2005), and our intersubjective understandings.Because no authority exists which could "lay down the law" and define social constructivism once and for all by fiat, and because "there is no constructivist orthodoxy" (Jackson 2011, 4), the contemporary situation is one in which researchers construct intersubjective meanings and understandings of constructivisms -plural.Social constructivism is socially constructed, just like any other theory or approach, but there are multiple ways in which it can be constructed, deconstructed, and reconstructed15 .And this is exactly what has happened as IR constructivism has evolved over the years.

A tale of two cognitions
The constructivist evolution can be seen as a tale of two genera, each distinguished by their cognitions.Instead of the usual story of some more or less uniform understanding of constructivism in its early days, which was then mainstreamed and lost something in the process, recall my earlier point that the roots of the IR constructivist family tree were always diverse.From the beginning there were many constructivisms, (see e.g. the topography in Hynek and Teti Rev. Bras. Polít. Int., 60(1): e014, 2017 Peltonen 11 2010, 178).Despite their differences, what such early constructivists as Kratochwil and Onuf shared, however, were particular notions of legitimate processes of knowing.Some might call it a shared worldview, and use the usual vocabulary of epistemology and ontology, or perhaps something else.I prefer to use the term cognition.
I use cognition as a shorthand for the mental processes of knowing, which includes perception, reasoning, and judgment.I prefer it over epistemology and ontology, because cognition allows me to take both of them together and simultaneously.Moreover, by not using these loaded terms, I avoid old trenches 16 .Furthermore, with the term cognition I can highlight, for example, that what has been at stake is not whether norms exist or whether they matter, but what the legitimate processes of knowing are and how they matter.Furthermore, cognition underlines the differences in perception and reasoning between various constructivisms without implying that some versions are not "really" constructivist, or that they are inconsistent or something similar.What some might call inconsistent constructivism, I call constructivism based on a different cognition which is consistent according to that cognition.Finally, my focus on cognition may help in seeing how many of the (meta-) theoretical and conceptual arguments and debates between constructivists and others (and among constructivists) have been about or over the legitimate mental processes of knowing.At times it has been explicit 17 , at other times implicit 18 .
To recall, George and Campbell (1990) called the wider intellectual debates of the 1980s "dissent".It was a call for "thinking space".Such thinking space was called for due to the prevalent mode of "scientific" cognition that seemed to ignore other legitimate modes of knowing.Such early IR constructivists as Kratochwil (1989) and Onuf (1989) were writing against established empiricist-(neo)positivist IR in order to make intellectual space for constructivism, which entailed a different mode of cognition from the contemporary mainstream (see also Lapid 1989).
The way in which both Kratochwil and Onuf fought for the widening of the contemporary intellectual space was by resorting to norms and rules 19 .Part of the argument was illustrative; they showed what the study of norms would enable or allow to see beyond what the prevalent, mainstream cognition enabled.For example Kratochwil's (1991, 5) justification for his book is to "gain a better picture of why actors in the international as well as in the domestic arena have to resort to norms".Contemporary regime theory debates formed the backdrop for his argument, and his call was for a different kind of thinking and reasoning than what was prevalent: norms must be understood as more than just intervening variables, if we are to perceive and apprehend the perceptions of the actors and acts we study -and by extension how norms affect our perceptions of their perceptions (a kind of double hermeneutic, see e.g.Giddens 1976).Also, Onuf argued for a different cognitive mode than the contemporary mainstream.His point is that we cannot 16 See, for example, the exchange between Kratochwil (2007a,b) and Wight (2007a,b) on the Journal of International Relations and Development.17 See, for example, Kratochwil (1989) whose main argument is exactly about the way we should reason.
18 Constructivism "is about human consciousness and its role in international life" (Ruggie 1998, 856 emphasis added).
Both Kratochwil and Onuf use a cognition that denies the usefulness of separating things in the world "out there" into neat entities in order to study their (mechanistic) causal relations.Instead, their proposals outlined a joined mode of cognition that understands its own situatedness in and as part of a social world and uses reasoning and judgment rather than inference or (logical) proof by necessity to make its case.
Some of this kind of cognition is evident in the so-called mainstream constructivism, of which I use Wendt's (2010) via media constructivism here as an example.Although their approaches differ in many respects, all three, Kratochwil, Onuf, and Wendt, share at least one central argument.The argument presented in their books is that we gain better or more useful knowledge and insight by perceiving things like culture, of which we are part, as being in motion but under constant reproduction (Wendt 2010, 188), rather than as static until motion or action is caused by some force.All three focus on a cognition that emphasizes knowing through an awareness of the co-constitution of agents and structure.Or, to quote Wendt (2010, 372), "state cognition depends on states systemic culture".
A difference, however, is seen in Wendt, and in a number of other contemporary authors 20 , because he diverges from the joined mode of cognition of Kratochwil and Onuf.He uses a cognition and language more familiar to the wider, mainly North American audience.Instead of the language and the ways of reasoning used by the "dissenters", he is in line with the contemporary understandings of legitimate and scientific ways of reasoning and argumentation.
The contemporary mainstream cognition, and the kind of reasoning process it deemed legitimate, can be illustrated by one of its authoritative sources: King et al. (1994, 37).According to King et al. "science … and interpretation are not fundamentally different endeavors aimed at divergent goals".This was used as a justification, for instance, by a highly influential volume on what became an important example of constructivist security research (Katzenstein 1996).
The important detail to note is that even though "science" and "interpretation" were not fundamentally different, or aimed at different goals in the reasoning deemed legitimate by the mainstream, these two were kept separate.This was contrary to what Kratochwil and Onuf had told us.They had argued that we should understand "science" and "interpretation" together, because ultimately they could not be separated from each other.This mainstream cognition, separate but equal, made it possible to reason that the "dissent" was not as radical as some may have initially thought.Here, one need not share Hynek and Teti's (2010) vocabulary of "purges" and "hijackings".It suffices to recall Wendt's (2010, 372-73) 20 See, for example, Katzenstein (1996) and Finnemore (1998).On how constructivism's empirical contributions fit within the (neo) positivist social science of IR, see Dessler (1999).
Rev. Bras.Polít.Int., 60(1): e014, 2017 Peltonen 13 argument that there was epistemologically nothing different between explaining social construction and what natural scientists study.Another example is how the article by Kratochwil and Ruggie (1986) had been a shock to the field, but the separate but equal cognition was used to explain that such approaches were not as radical as they seemed and to reason that they "converge substantially with those advocated by mainstream scholars" (Jepperson et al. 1996, 68).A third example of such efforts is the argument by Wendt, mentioned earlier, that causal and constitutive questions can be treated the same.
From the of those using the separate but equal cognition, constructivism did not appear all that radical.For instance, in terms of methodology, an acknowledgment of norms and the social construction of identities did not mean a departure from "normal science" (Jepperson et al. 1996, 65).This meant adherence to the way of reasoning that science must advance through empirically testable propositions, whereas the joined cognition emphasized that there are other ways of knowing, and, anyway, tests do not explain themselves.Yet, because the kind of constructivism with the separate but equal cognition became more popular, for instance, constructivist norm research turned towards emphasizing structure, which in turn fit neo-positivist methodology in IR (Hofferberth and Weber 2015, 81), thus creating a mutually reinforcing cycle21 .Important to note is, though, that what is at stake was not whether norms (or identities) matter.Constructivisms following both cognitions agreed that norms (and identities) matter.Rather, at stake was which processes of knowing are legitimate for studying how norms or identities matter.
To put it in a nutshell and with a different vocabulary, IR constructivism has been diverse from the beginning and it has been, and is, a kind of Wittgensteinian family.Some within this family use a different cognition and ways of reasoning, than others.Rather than a purge or a mainstreaming, the constructivist family has experienced evolution and an evolutionary branching due to the different cognitions used by different authors 22 .This means that the constructivist family split into two genera: one continued to apply the joined cognition, the other used the separate but equal cognition, which happened to be the contemporary mainstream cognition.The latter circumstance denotes that constructivism's evolutionary branching was probable, because the separate but equal cognition was so prevalent.
Furthermore, the evolutionary branching of constructivism supports my earlier argument that calling some constructivisms inconsistent is odd.Given that the constructivist family split into two genera, calling one genus (separate but equal constructivism), or particular "species" (individual contributions, to continue the vocabulary of biology) within it inconsistent according to the characteristics of another genus (joined constructivism) would be the real inconsistency.Putting it differently, constructivisms subscribing to the separate but equal cognition are consistent according to the criteria that characterizes this genus, and constructivisms subscribing to the joined cognition are consistent according to the criteria of that genus.

Conclusion
The paradox of IR constructivism being a success, but also a disappointment for many constructivists can be understood through constructivism's evolutionary path.While the disappointment is understandable, it is misplaced.
From the start, IR constructivism was not uniform; there never was "The Constructivism", but constructivisms.Among themselves, early constructivists, such as Onuf and Kratochwil, were quite different in their approaches and expressions.They shared, though, a kind of cognition that differed from the contemporary mainstream.A key insight of this "joined" cognition, or process of knowing and reasoning, was the inseparability of science and interpretation, which led them to emphasize historicity, context, contingency, and reflexivity.This cognition gained legitimacy thanks to the work of early constructivists (and others).Yet, with hindsight, and given that the contemporary dominant cognition was different, an evolutionary branching was probable, although at an uncertain time.The deepening of constructivist legitimacy required adaptation to the mainstream cognition, to its processes of reasoning and knowing, if for no other reason than to be able to communicate with the mainstream and to explain social constructivism in a language that was familiar to it (and vice versa).This cognition I called separate but equal, because it considers science and interpretation as equal in value, but, nevertheless, keeps them separate.In turn, this leads its adherents to believe in the centrality of the scientific method in the production of warranted knowledge.In a nutshell, the history of IR constructivism is an evolutionary tale of two cognitions, of two genera, within the Wittgensteinian constructivist family.
To the extent that there are (at least) two main genera of constructivism, contemporary disappointment in constructivism needs to self-reflect whether it is applying the characteristics of its own genus to the other one, and being unfair in doing so.It seems misplaced to be disappointed with some constructivisms for having adapted to the prevalent mainstream cognition, thus evolving into a new genus.Since the constructivist family split into two genera, mixing criteria of appraisal across them would be unfair.
The language of cognitions and genera helps in assessing future evolutions of IR constructivisms.If new ways of reasoning begin to influence constructivism, new genera are easy to add.It remains to be seen, for instance, how the practice turn and its way of reasoning influences constructivist evolution, perhaps more toward pragmatism (e.g.Friedrichs and Kratochwil 2009), or whether McCourt (2016) is correct in practice theory and relationalism forming the new constructivism.At least, by making distinctions in cognitions and genera, we can leave behind the unconstructive infighting over the "right" kind of constructivism.