In this paper, I argue for two claims, (i) that on a common conception of the second order property of being a mental disorder, some facts about mental disorders are the result of social constructions, and (ii) that the way facts about mental disorders are constructed differs from the received view on social construction. The difference is examined, a novel type of social construction is identified, and it is suggested that there are numerous other types of social facts that are constructed in a similar way.
It is a widely shared assumption among psychiatrists and philosophers of psychiatry alike that the socio-cultural environment in some sense or another shapes mental disorders, a claim that, today, also objectivists about mental disorder happily admit (see, for instance, Kendler et al. 2011). It has been pointed out that social environment or cultural formation may have an impact on access conditions to health care systems (U.S. Department of Health and Human Services, 2001) and on how patients express their experiences (American Psychiatric Association 2013, Section: Cultural Evaluation); recent data suggest that there is cultural variation among the ways in which people experience their conditions, and that the cultural background has an impact on how clinicians weigh the relevance of symptoms in the diagnostic process. Moreover, for at least some disorders, it has been suggested that cultural formation may even have an impact on the development of the disorder, not in the sense that cultural formation may cause severe trauma- or stressor-related disorders (as it might very well do), but in the sense that it “offers” certain ways of having a mental disorder (this is, in a nutshell, part of Hacking’s point (1995), a cognate of which has also been discussed by Piper and Merskey 2004). 
In yet another spirit, some psychiatrists and philosophers have argued that the evaluation of a condition as a disorder is a normative affair, and that the norms in question are often social norms. This view is widespread; as I will argue below, it can, by plausible standards, even be regarded as a standard view in current psychiatry. And the view has wide scope. The notion of a social norm is very inclusive, and as a consequence, violations of social norms not only include such obvious cases as, for instance, deterioration of personal hygiene in cases of schizophrenia, but also certain mannerisms or disorganized speech that may go together with forms of thought disorders.  Call the view that mental disorders require violations of social norms the Social Norm view about Disorder Evaluation (SNDE). Depending on the details of SNDE, this view may have far reaching consequences for psychiatry. Prima facie, SNDE sounds very much like a constructionist thesis – if SNDE is true, then facts about whether or not some condition is or counts as a mental disorder depend, at least partly, on a particular type of social facts, namely, social norms. It goes without saying: Constructionism about mental disorder may have a significant impact on theories regarding the relation between physiological and mental illness and health, the ethics of psychiatric intervention, and the practice of marking people as having a disorder  and, possibly, on the status of psychiatry as a science (Szasz 1972).  Moreover, the concept of a mental disorder may mask the social nature of the property of being a mental disorder, by presenting it as a non-social property – like, perhaps, some gender and race concepts do.
This paper is a piece in applied social ontology, the aim being to offer a better understanding of what SNDE consists in by uncovering some of its metaphysical presuppositions. The result will be a partial explication of the form of social constructionism underlying SNDE. In particular, I will argue that the form of social constructionism presupposed by SNDE differs significantly from standard constructionist claims, as, for instance, Searle’s claim that ‘money is money because [… people] regard it as money’ (Searle 2010, p. 17). Standard forms of constructionism hold that there is a tight connection between the content of the attitudes a social fact depends on and the nature of the dependent social fact. It appears that some such conception lies at the heart of philosophical theories about social construction: If facts about F’s are socially constructed, then some things are F’s because they are (commonly) regarded, treated, or recognized as F’s. This is the received view on social construction.
The case of mental disorder is different, though. Here, we have what one may call a form of indirect construction. If SNDE is true, some psychological conditions are mental disorders because they are the target of shared expectations or norms, whose content need not involve the concept of a mental disorder. In this respect, the construction of facts about mental disorder differs fundamentally from the construction of facts about money. Although the primary goal of the present paper is to uncover the social nature of facts about mental disorder as presupposed by SNDE, the results have consequences that transcend the boundaries of the philosophy of psychiatry, and may have an impact on social ontology in general. In passing, I will suggest that there are other categories of social facts that appear to be constructed in a similar way. Indirect construction appears to be a widespread phenomenon, so that the received view about the mechanisms of social construction needs to be refined.
Section 2 introduces the idea that if SNDE is true, then whether or not a psychological condition is a mental disorder partly depends on the violation of social norms, or the disposition to violate social norms (Section 2.1). The relevant notion of dependence is non-causal (2.2). Since social norms are constituted by shared expectations of a certain type, facts about the violation of a social norm depend, in part, on facts about shared expectations of a certain type (2.3). Section 3 explores the consequences of these observations: By the transitivity of dependence, whether or not a condition is or counts as a mental disorder depends, in part, on shared expectations (3.1). If SNDE is true, the case of mental disorder resembles paradigm cases of social construction in precisely this respect. Yet, dependence on shared attitudes is not sufficient for being socially constructed in the relevant sense. Rather, the objects a constructed social fact is about, such as coins or bills in the case of money, or a psychological condition in the case of mental disorder, needs to be the target of the shared attitudes on whose presence the constructed fact depends (3.2). I will explore the difference between facts about money and facts about mental disorder, and I will tentatively conclude that we are faced with a novel kind of social construction that deserves close attention (3.3), not only in the philosophy of psychiatry, but in social philosophy in general (3.4).
In order to avoid misunderstandings, let me explicitly state, first, that I will not discuss the truth of SNDE; I will be concerned with some of the implications of the assumption that SNDE is, by and large, correct. Moreover, I am concerned here with the second order property of being a mental disorder (or mental illness – I use these interchangeably).  Just like the second order property being a virtue is instantiated by the first order property wisdom, being a mental disorder is instantiated by first order properties like ADHD, schizophrenia, or dissociative identity disorder. As a consequence, the form of constructionism I will be concerned with here does not necessarily translate into constructionism about first order psychological conditions, such as ADHD, schizophrenia, etc. If, however, the notion of a mental disorder is built into concepts of ADHD, schizophrenia and the like (and not merely the term ‘disorder’ is, sometimes, built into the labels), constructionism about the second order property of being a mental disorder will generalize to at least some conceptions of first order psychological conditions. Finally, classifying a theory (SNDE) as constructionist should not lead to the impression that the allegedly constructed objects or facts are not real. A constructionist theory concerning talk about F’s is not ipso facto an error-theory concerning talk about F’s, although it may go together with an error-theory concerning the alleged inferential role of the concept of an F. Recall: The concept of a mental disorder may mask its social nature. If so, constructionism about the concept may be revisionary in this respect, without any commitment to the claim that mental disorder discourse is empty, or necessarily false. 
2 Mental Disorder and the Violation of Social Norms
2.1 SNDE within DSM-5
SNDE – the view that the violation of social norms is relevant for the question of whether a condition counts as a mental disorder – is widespread. In the tentative characterization of the notion of a mental disorder published in the introductory parts of the Diagnostic and statistical manual of mental disorders, 5th edition (henceforth: DSM-5), the authors state that “[a]n expectable or culturally approved response to a common stressor […] is not a mental disorder.” (American Psychiatric Association 2013, p. 20) It seems that thereby, the authors want to indicate that something is a mental disorder only if it involves responses to stressors that are not expectable or culturally approved. Although this reading is, of course, not implied by the phrase, it is supported by the following passage:
The boundaries between normality and pathology vary across cultures for specific types of behaviors. Thresholds of tolerance for specific symptoms or behaviors differ across cultures, social settings, and families. Hence, the level at which an experience becomes problematic or pathological will differ. The judgment that a given behavior is abnormal and requires clinical attention depends on cultural norms that are internalized by the individual and applied by others around them, including family members and clinicians. (American Psychiatric Association 2013, p. 14)
Note that this should not be construed as a purely epistemological point about access conditions,  a view that would have the odd result that some cultures might be more sensitive to the presence of a mental illness than others. The point is, rather, that not all possibly problematic experience is pathological, and that the level at which certain experiences (say, of anxiety) become pathological depends, in part, on the social context; passing the socially or culturally determined threshold is required for the presence of a mental disorder, not merely for the detection of the presence of a mental disorder. This idea has played a prominent role in the anti-psychiatrist movement (Szasz 1972), and the question of whether normative aspects involved in diagnostic procedures are social or can be cashed out in, say, descriptive or biological terminology has attracted considerable attention (Boorse 1975, Kendell 1975, Fulford 1999, Thornton 2007). Stier (2013) explicitly states that deviation from expectations will have an impact on whether a certain kind of behavior counts as clinically relevant; although officially defending a purely epistemological point, he seems to articulate the thought underlying the passage just quoted from DSM-5, when he writes:
In sum […] the respective cultural setup […] tends to dictate the boundary between the normal and the deviant on the basis of the expected values and virtues of its members. (Stier 2013, p. 28)
In the same context, Stier suggests that there is a “[normative] impact of society on the concept of mental disorder.” (Stier 2013, p. 28f.) The following appears to offer a concise partial explication of the underlying idea – an idea that is not the consensus among psychiatrists and philosophers, but that clearly marks an influential view: 
[A-1] Necessarily, if condition a is or counts as a mental disorder, then this depends, in part, on the fact that having a involves violation of (or involves the disposition to violate) social norms.
Of course, violation of social norms does not exhaust the dependence base for facts about mental disorder. Not anyone who violates, or is disposed to violate social norms exhibits the traits required for showing pathological behavior. [A-1] is a minimal characterization of SNDE we can work with. And [A-1] marks an important difference between those who believe the violation of social norms to be constitutive for the presence of a mental disorder and those who do not (cf. Boorse 1975, Kendell 1975). 
Before we continue the discussion of [A-1], let me briefly reflect on the status of DSM-5, and its relevance for philosophical accounts of conceptions of mental disorder. I intend to take the tentative characterization of the concept as proposed in DSM-5 at face value, in order to explore some of the metaphysical commitments that go together with this conception of a mental disorder. One might think that thereby, I somehow misconstrue the intention of DSM-5. After all, it is a manual that is supposed to guide clinical practice, and the characterization of a mental disorder just quoted may be regarded as purely operational in nature; it is supposed to articulate an idea one can work with in clinical practice, but which falls short of a meaning revealing definition, or scientific explication. Drawing metaphysical conclusions about commitments from a purely operational characterization may seem misplaced. Let me illustrate: Assume that I ask you to prepare a couple of quail eggs. You ask what a quail egg is. I give an operational characterization: ‘The only thing in my refrigerator that is blue-and-brown speckled and looks slightly artificial.’ One should not draw conclusions about the metaphysics of quail eggs from this characterization. But it will help you find the eggs. So, is this all there is to the DSM-5 characterization of the concept or property of being a mental disorder?
I believe a purely operational interpretation of the DSM-5 characterization to be problematic. First, it is unclear how reference to the violation of social norms would be of any help in a purely operational characterization of the concept of a mental disorder. Why should the violation of social norms (a criterion that is highly contextual) successfully and reliably track mental disorders unless the violation of a social norm is somehow required for having a mental disorder? This would be a rather miraculous connection. Second, radical biologism is not the dominant view in psychiatry (any more); and I think it is fair to assume that DSM-5, in its characterization of the concept of a mental disorder, articulates a widely shared assumption. So, even if the official purpose of DSM-5 supports an operational interpretation, it will, at the same time, point to a concept or theory of mental disorder that does play a vital role in current psychiatry. One may want to suggest that DSM-5 articulates a rather common view. By these lights, I think we are justified to take the DSM-5 characterization at face value. A commitment to the so-described concept goes together with a commitment to the metaphysical implications I will discuss in what follows. Let us thus turn back to [A-1]. Two points need to be addressed. We need a sufficiently precise understanding of both, ‘dependence’-talk (2.2) and of the notion of a social norm (2.3).
2.2 Two Readings of Dependence – Causal vs. Non-Causal
We need to distinguish carefully between two interpretations of ‘dependence’-talk in [A-1]. There is, first, a simple causal interpretation, which is not intended here. Assume that a subject develops a minor anxiety about, say, speaking in public in front of strangers.  Assume that this form of anxiety, and the behavioral patterns that go together with it, are conceived as socially awkward, or that they violate some shared expectation. Showing the relevant behavior (say, some form of avoidance behavior, or a specific type of deviant behavior shown while delivering a speech) constitutes a norm violation; people may react to the norm violation, thereby enforcing the anxiety in the subject – to a degree that it becomes pathological. In this case, actual social feedback in response to norm violations may trigger a feeling of shame, which, in turn, may cause the person to experience distress, which, in turn, may cause further deviations from socially expected behavior to a degree which makes the condition pathological. In this fictional case, violation of social norms plays a causal role for the development of a disorder. 
If you commit to [A-1], in the intended sense, you do not thereby commit to the claim that there was, first, norm-violation which then causally led to the development of a mental disorder. This is already suggested by the formulation that “the fact that a condition is or counts as a mental disorder depends on the fact that having it involves” a norm-violation, which, taken literally, does not admit of a straightforward causal interpretation (unless we take the relata of causation to be facts). There are several ways in which you can cash out the intended interpretation of [A-1], depending on your preferred views on explanatory or metaphysical dependence, truth-making, or grounding.  For instance, one may suggest that the truth that a person has a mental disorder is grounded in a truth about the violation of social norms. Or, alternatively, one may want to claim that the instantiation of the property of having a mental disorder metaphysically depends on the occurrence of norm-violations. We need not go into the details here. For our present purposes, suffice it to note that there are at least two interpretations of [A-1], a causal and a non-causal one, that the causal interpretation is not the intended interpretation, and that whatever the correct explication of the intended interpretation is, it will render [A-1] true without any implications concerning a possible causal connection between norm-violation and having a mental disorder.  An understanding along these lines is sufficiently precise for the goals of the present paper.
[A-1] offers a partial explication of SNDE, i.e. the view that whether or not a condition counts as a mental disorder depends on social norms. But what is a social norm?
2.3 Social Norms – A Minimal Characterization
All we need in the present context is a minimal characterization of the concept of a social norm. For those who are not familiar with the debate, let me briefly introduce one influential interpretation (simplifying a lot). The key idea is explicit in Elster’s characterization of a social norm, or, perhaps better: in his characterization of the conditions that determine whether or not a normative proposition is a social norm in a context: 
A social norm is simply a shared expectation that others will react to a given behavior in a way that is painful for oneself. (Elster 2009, p. 196)
On Elster’s view, it is a social norm that one should always wear black clothes at a funeral (his example) in a particular social environment because in this environment, people expect others to react in a particular way if one were to violate this norm. And what is it to violate a social norm? Dressing as a clown at a funeral violates a social norm (maybe amongst other norms) in the sense that we all, or most of us, or a significant number of us, expect that others would react in a way that would be painful for ourselves if we were to dress like a clown at a funeral. Elster stresses ostracism as the typical response to the violation of a social norm, and the feeling of shame as the typical feeling caused by (expected) ostracism (Elster 1989, 2009). He contrasts social norms with other norms in terms of the motivations to comply and the typical sanctions associated with these norms.
On this view (and I take it that this is the consensus among those who work on social norms), social norms depend on attitudes of members of the group in which the social norm exists, or applies, or in fact regulates behavior. In contrast to the debate on the law, no form of naturalism about social norms has survived when philosophy matured; loosely speaking, social norm positivism is common ground. In contrast with the law, of course, social norms are not regarded as being grounded in contracts or performatives; they are supposed to be grounded in shared expectations or attitudes, which may be only implicit. Now, a violation of a social norm requires that there be social norms and, hence, certain expectations or attitudes. Whether or not an action counts, in a context, as a violation of a social norm depends on the expectations that constitute, or realize the norm, or have the relevant normative proposition as their content. Hence, the following minimal condition seems hardly contentious:
[A-2] Necessarily, if some action b constitutes the violation of a social norm, then this depends, in part, on facts about shared attitudes.
Similarly, for dispositions: If a disposition is a disposition to perform actions that count as violations of a social norm, then the fact that the disposition is a disposition to perform actions that count as violations of a social norm, depends on facts about shared expectations or attitudes. Note again that this notion of dependence is non-causal. Although shared expectations may cause a violation of these expectations (remember the time when you were a teenager!), the point here is one concerning non-causal dependence.
Before we turn to the connection between [A-1], [A-2] and constructionism about mental disorder, let me flag one potential problem standard conceptions of social norms seem to face. It seems hardly contentious that different social norms target different groups; what may be considered non-deviant for younger persons may be regarded as deviant for the elderly; what seems acceptable for adults need not be regarded as acceptable for teenagers. This is at odds with the extremely egalitarian conception of social norms that seems to dominate the debate within the philosophy of the social sciences.  It is tacitly assumed that social norms do not discriminate among groups within a society – those whose attitudes ground the norm are also the possible targets of the norm. Yet, there appear to be social norms that are group specific (and not merely the outcome of an implicit bias): The shared expectations regarding children’s behavior differ vastly from expectations regarding adults, not always to the children’s benefits. And it should be clear that it is not the children’s attitudes that ground these norms. Also, social norms regarding members of certain race- or gender-categories may vastly differ, and again, the set of individuals who are expected to conform is not identical to the set of individuals whose attitudes ground the norm, although, of course, these sets may overlap. And especially in the context of psychiatric evaluation, group-specific expectations may play an important role, as sexuality related disorders illustrate. 
Here, I am merely concerned with the social nature of a general conception of mental disorder. [A-2], in connection with [A-1], is sufficient to support this point. And [A-2] will come out true, independent of whether we construe social norms as egalitarian or group specific. In this respect, possible flaws of egalitarian conceptions of social norms are peripheral to the main line of the argument. Yet, it is important to note that if the violation of social norms is essential to the classification of a psychological condition as a mental disorder, future research, especially on the ethical ramifications of constructionism, should take into account that social norms may be group specific, so that prima facie, perceived or actual group-membership may have an impact on the likelihood of being classified as suffering from a mental disorder, due to the contingencies of group-specific expectations.
We have collected the pieces, now let us put them together: How do we move from [A-1] and [A-2], and, hence, from SNDE to constructionism about mental disorder?
3 From Norm-Violation to Social Constructionism
3.1 SNDE, Attitude-Dependence, and Constructionism
Constructionism about mental disorder maintains that:
[C] Facts about whether or not a psychological condition is or counts as a mental disorder are social facts.
Let me repeat: [C] concerns the second order property of being a mental disorder; it does not concern first order properties such as being an alcoholic, being schizophrenic, or suffering from PTSD. You can subscribe to [C] without being committed to any particular view concerning the nature of the various disorders there are. DSM-5 and similar manuals list criteria for various disorders that are independent of social facts.  These criteria concern the first order properties that are supposed to realize the second order property of being a mental disorder. Now, is [C] true? Consider again [A-1] and [A-2]:
[A-1] Necessarily, if condition a is or counts as a mental disorder, then this depends, in part, on the fact that having a involves violation of (or involves the disposition to violate) social norms.
[A-2] Necessarily, if some action b constitutes the violation of a social norm, then this depends, in part, on facts about shared attitudes.
From these, and the assumption that dependence is transitive, it follows that facts about whether or not a psychological condition counts as a mental disorder depend, amongst other things, on facts about shared attitudes (or expectations). This is, in itself, a significant result, if one believes the second order property of being a mental disorder to be a natural, or objective property. But clearly, those who adhere to SNDE do not believe that it is an objective property, in the sense that it is independent of shared expectations or attitudes. The intermediate result that facts about mental disorder depend on facts about shared attitudes will hardly surprise them. How do we move from the dependence claim to [C]?
3.2 Attitude Dependence and Being Social
First, let us have a look at what an argument for [C] might look like. After all, ‘social fact’ is a philosophical term of art, and there may not be an ultimate answer to the question of whether some fact that p is really a social fact. Meanings for technical terms are up for grabs. Instead of showing that facts about mental disorders fall under some well-defined notion of a social fact, I will argue that there are significant similarities between paradigmatic social facts and facts about mental disorders. These similarities give reason to apply the term. But how far does the similarity go? As we have seen, it follows from [A-1] and [A-2] that
[C*] Facts about whether or not a psychological condition is or counts as a mental disorder depend, in part, on facts about shared attitudes.
In this, facts about whether or not a psychological condition counts as a mental disorder resemble stock examples of social facts, such as facts about money, borders, or dress-codes. So, one might think that we are already there: Facts about mental disorder resemble paradigmatic social facts, such as facts about money, in the relevant respect! Unfortunately, dependence on shared attitudes alone does not suffice to show that some fact is sufficiently similar to facts about paradigmatic social objects, like money or borders, to group them together. There are facts that depend, in whole or in part, on facts about shared attitudes that are clearly not social in the relevant sense. Before we turn to the details of the relation between facts about mental disorder and paradigmatic social facts, let us, in a first step, get clear about when dependence on shared attitudes suffices to classify a fact as social. The following sentence expresses a fact that depends on shared attitudes and is not social, in the relevant sense of ‘social’:
 The set of shared attitudes has at least one element.
Clearly,  differs from facts about money or borders; I assume that it is not socially constructed in the sense in which “social construction” is used in the philosophical debate. Yet, it depends on facts about shared attitudes. Hence, dependence on facts about shared attitudes alone does not make a fact a social fact, in the relevant sense. Where there may be doubts in the case of mental disorders, it appears that there is a clear intuitive difference between facts about paradigmatic social objects, such as money or borders, and the fact expressed by .  Yet, although dependence on facts about shared attitudes is not sufficient for being a social fact, in the intended sense of “being social”, the dependence of social facts on shared attitudes is not fully irrelevant for these facts’ status as social facts either.  At least, this is what most philosophers who deal with social reality would agree on. It seems that there must be something in the way social facts depend on the shared attitudes they depend on, some aspect of the connection, in virtue of which dependence on shared attitudes sometimes, but not always, amounts to social construction. An analysis of the difference between  and paradigmatic cases of social facts will shed light on the connection between social facts and shared attitudes in general, and thereby pave the way for an assessment of the details of the relation between SNDE and other forms of social constructionism.
Let us begin with a relatively superficial feature of the connection between social facts and the attitudes they depend on. Consider, for instance, Searle’s so-called standing declarations, or constitutive rules, which have the form ‘X counts as Y in C’ (Searle 1995, 2010). Searle says next to nothing about the actual functioning of the ‘counts as’ locution. But his use of the locution points to an interesting difference between stock examples of social facts and . Consider the following claim:
 Dollar bills issued by the Federal Reserve banks are money.
Now, interestingly, it appears that if  is true, then so is:
[2*] Bills issued by the Federal Reserve banks count as money.
And one may suspect that necessarily, if  is true, then this is so because [2*] is; we can account for why some things are money in terms of facts about what counts as money (in a context). Interestingly, no such transformation is possible in the case of . Whereas  may be true, it does not follow that:
[1*] The set of shared attitudes counts as having at least one element.
Before we turn to the question of what underlies the difference, let us have a quick look at transformations from ascriptions of being a mental disorder to ascriptions of counting as a mental disorder, (still on the assumption that SNDE is true). It seems that we can easily move from  to [3*]:
 Schizophrenia is a mental disorder.
[3*] Schizophrenia counts as a mental disorder.
If SNDE is correct, we have another indicator of the similarity between facts about being a mental disorder and paradigmatic social facts. But what explains the difference between  on the one hand, and  and  on the other?
Reflection on the ‘counts as’ locution may help: Facts about money and mental disorder depend on attitudes that target the objects of the constructed facts, i.e. coins, bills and psychological conditions. On Searle’s view, money is money because it counts as money, in a context, or because people regard it as money. Similarly, if SNDE is true, some conditions count as mental disorders because these conditions, dispositions to behave in a certain way, are the target of shared attitudes.  And for a condition to be a mental disorder, it has to be the target of shared attitudes (if SNDE is true). Not so in the case of  – the set of shared attitudes need not be the target of any attitude at all (shared or not), to have at least one element. I claim that this observation accounts for the relevant difference: Social construction involves targeting the object of the constructed facts. Dependence on shared attitudes alone is not sufficient. In this respect, facts about mental disorder and facts about money are similar. If SNDE is true, then the construction of facts about mental disorder is a kind of social construction. [C] is in the clear.
Yet, there is an important difference between facts about mental disorder and facts about money (and related facts).
3.3 The Limits of the Similarity and Forms of Indirect Construction
Searle offers examples of what he takes to be observer-relativity, or what he later called ‘intentionality-relativity’ (Searle 2010, p. 17). An instance we are already familiar with is this: “Money is money because the actual participants in the institution regard it as money.” (Searle 2010, p. 17) There is, thus, a tight connection between the content of the attitude and the fact grounded in the attitude, i.e. regarding something as an F, and being an F. The content of the shared attitude (and, in a similar way, a standing declaration) transfers a social status that matches the content of the attitude. This is the simple case, and cases of this sort form the basic material for Searle’s theory of institutions. The underlying mechanism can be described as follows: The social fact that depends on the shared attitudes, or standing declarations, is (partly) constituted by the content of these attitudes or declarations.
In constructionist claims about gender categories, we find a similar structure. Consider, for instance, Haslanger’s characterization of being a woman. On her view, being a woman (in the social sense) is, roughly, to be systematically subordinated in some respect in virtue of being marked as the target of such treatment based on observed or imagined bodily features.  Again, the content of the attitude, namely: being a candidate for a specific type of treatment, matches the social fact grounded by this attitude; namely, having this very property. In a similar spirit, Haslanger claims that “[w]e all know that sagging pants are only cool, insofar as they are, by virtue of being viewed as such by an in-group” (Haslanger 2012, p. 459). I will turn back to this last example below, where I will suggest that Haslanger may in fact misconstrue the case. But the idea should be clear: Constructionist claims involve the idea that some things are F’s because people stand in some intentional relation to these things, where these things are represented as F’s. This is the received view on social construction.
Despite the vast differences between Searle’s characterization of institutions and Haslanger’s views on gender categories and coolness, they share the feature that social construction is, in a sense, supposed to be direct. The attitudes’ content is directly transferred to the fact the attitudes ground. So, by these lights, are facts about mental disorder just like facts about money or borders? After all, it seems that the content of those shared expectations in virtue of which some psychological conditions count as mental disorders need not involve the concept of a mental disorder. In this respect, the construction of facts about mental disorder differs from social construction in standard cases. For money to be money, people have to regard it as money. It is not the case that according to SNDE, for a psychological condition to be a mental disorder, people have to regard it as a mental disorder. As a consequence, social construction need not be direct; the constructed fact need not be part of the content of the corresponding shared attitudes.
Reflection on an analogous case may help to shed light on the basic structure of what I would like to call indirect construction: Repeated violation of the law makes one a habitual offender (ignore any explicit habitual offender laws, such as “three-strikes laws” in the US – let us stipulate that this is a term for people who frequently break the law). The status of being a habitual offender depends, in part, on the existence of laws. This status can be acquired only in a certain social setting – a setting where laws exist. Facts about being a habitual offender depend, in part, on the law. And if the law depends on shared attitudes, then so do facts about being a habitual offender. The law targets the behavior of the habitual offender. Nevertheless, it is not the case that habitual offenders are habitual offenders because people regard them (or the law explicitly classifies them) as habitual offenders. 
Similarly, it seems that even if SNDE is true, the way the content of shared attitudes contributes to facts about mental disorders is far more subtle than in the case of money. It is simply not true that by endorsing some version of SNDE, one is committed to the view that, for instance, alcoholism is a mental disorder because people regard it as a mental disorder. Rather, on this view, alcoholism is a mental disorder, in part, because people have some attitude towards alcoholism (or the behavior of the alcoholic). It is a mental disorder, partly because it involves the violation of a social norm. And it involves norm-violation because it is considered, in a sense, bad behavior, or because the behavior indicative of the condition does not comply with shared expectations dominant in the relevant group. There is a tight connection between the content of attitudes and the fact that alcoholism is a mental disorder: The content of the relevant attitudes, i.e. social norms or shared expectations, targets the behavior of the alcoholic, as well as the underlying psychological condition.SNDE assumes that this evaluation is, or should be, built into the scientific concept of a mental disorder. For some condition to be a mental disorder, it is required, though of course not sufficient, that the behavior indicative of the condition counts as inappropriate. There is an element of construction; but it is, so to speak, indirect. A requirement for being a mental disorder is, on this view, to be a psychological condition that disposes a subject to non-compliance with social norms, or shared expectations. The content of these expectations need not, in any way, involve the concept of a mental disorder.  In this, SNDE differs from other types of constructionist theories.
3.4 Indirect Construction Beyond SNDE
Upon reflection, it turns out that the phenomenon of indirect construction is quite common. One may, for instance, speculate that facts about being hip, being cool, or being a nerd may, in some contexts, function similarly. Above, I mentioned Haslanger’s claim that “sagging pants are only cool, insofar as they are, by virtue of being viewed as such by an in-group.” But is this true? Perhaps, sagging pants, or people who wear them, are cool in virtue of there being some views about these pants in an in-group, but conceptualization or treatment as being cool is not required. Falling under the concept of coolness is not a matter of explicit or implicit ascription of the property. Instead, one may speculate that it is, in part, a matter of satisfying certain aesthetic expectations. Then, it may be false to claim that those who are cool are cool because they are regarded as being cool; one may discover that someone is cool, even if no one noticed before. I find this intuitive. And it may very well be the hope of many of the not so cool kids. Similarly, one may literally discover that certain types of behavior are cool; namely, that they satisfy aesthetic expectations of a certain sort. The connection between the content of the shared aesthetic expectations is not given in terms of being cool, and the content of these attitudes is different from those facts about being cool that depend on the presence of these attitudes.
The difference between cases of direct and indirect construction can then be explicated as follows. Let us begin with an inclusive condition on social construction, taking the relevant type of shared attitude to be that of regarding an object in a certain way, and ignoring the difference between regarding individuals or types of objects in that way. Then, the fact that x is F is constructed only if necessarily, if x is F, then there is some G such that (x is F because, in part, people regard x as being G). We can then introduce the distinction between direct and indirect construction as follows: We have a case of direct construction iff the concept of an F=the concept of a G; and we have a case of indirect construction when these concepts are distinct. 
In this context, it is worth stressing that facts involving indirect construction appear to be easily misjudged as natural facts. In contrast to the money-case, lack of a tight conceptual connection between the content of our attitudes towards the objects of indirect construction and the constructed facts themselves explains why these facts are so easily misjudged. Hence, it should not come as a surprise that facts about being a mental disorder (still on the assumption that SNDE is true), or, for that matter, being cool, or being a nerd present themselves as, and are easily mistaken for natural facts.
The goal of the present paper was to argue for two claims. First, endorsing SNDE commits one to constructionism about mental disorders. The connection between the social grounds of facts about mental disorder and these facts is subtle. Just like the status of being a habitualoffender depends, in part, on the existence of laws, and not on explicit or implicit ascriptions, or treatment as a habitual offender by other people, the status of being a mental disorder depends, in part, on social norms, and not on explicit or implicit ascriptions, or treatment as having a mental disorder by other people (if SNDE is true). The social construction of facts about mental disorder is indirect. The previous discussion thus indicates that social construction may go beyond the standard cases of money and borders, and the less obvious but no less interesting cases of gender and race categories. The leading paradigm in social ontology has been of the following sort: ‘Money is money because people regard it as money.’ Searle has based his entire theory of social construction on a family of cases that resembles the money-case. The mechanism of social construction in these cases is straightforward – an object or a person acquires a status by being represented as having this status. This type of social construction obviously resembles cases of performatives, as Searle has repeatedly stressed. When successful, a performative speech act changes reality so as to fit the content of the speech act itself, i.e. in accordance with how the world is represented by this act.
The case of indirect construction is vastly different. Roughly, language appears to contain concepts that group objects by virtue of standing in certain relations to social facts, without making the connection transparent. The concept of a mental disorder groups conditions in virtue of which people are disposed to violate a social norm. The concept of coolness groups people or habits by virtue of showing or being a type of behavior that satisfies a particular type of aesthetic expectations. No social status is explicitly conferred in these cases, and since the mechanism underlying this type of social construction does not require explicit or implicit representation of the relevant objects as having the property in question, one should not be surprised that coolness and having a mental disorder are commonly regarded as natural, rather than social properties. So, we may not only have detected a novel type of social construction. In fact, the mechanism underlying this type of social construction may offer the resources to explain the resistance to accept that some social facts are indeed social in nature.
I would like to thank two anonymous referees for their very helpful comments on an earlier draft of this paper, as well as participants in the philosophy colloquium at the University of Duisburg-Essen, where this paper has been discussed. I would also like to express my gratitude to the Volkswagen Foundation for their generous support of the project “A Study in Explanatory Power” and the German Research Foundation (DFG) for their generous support of the network “Social Functions.”
American Psychiatric Association (2013): Diagnostic and Statistical Manual of Mental Disorders (5th ed.). Washington, DC: American Psychiatric Association. Search in Google Scholar
Beebee, Helen and Nigel Sabbarton-Leary (2010): “Are Psychiatric Kinds ‘Real’?”. In: European Journal of Analytic Philosophy 6, p. 11–27. Search in Google Scholar
Bicchieri, Cristina (2006): The Grammar of Society: the Nature and Dynamics of Social Norms. New York: Cambridge University Press. Search in Google Scholar
Boorse, Christopher (1975): “On the Distinction between Disease and Illness”. In: Philosophy and Public Affairs 5, p. 49–68. Search in Google Scholar
Correia, Fabrice and Benjamin Schnieder (Eds.) (2012): Metaphysical Grounding. Understanding the Structure of Reality. Cambridge: Cambridge University Press. Search in Google Scholar
Elster, Jon (1989): The Cement of Society. Cambridge: Cambridge University Press. Search in Google Scholar
Elster, Jon (2009): “Norms”. In: Peter Hedström and Peter Bearman (Eds): The Oxford Handbook of Analytical Sociology. Oxford and New York: Oxford Univ. Press, p. 195–217. Search in Google Scholar
Fulford, Bill (1999): “Analytic Philosophy, Brain Science and the Concept of Disorder”. In: Sidney Bloch, Paul Chodoff and Stephen A. Green (Eds.): Psychiatric Ethics. Oxford: Oxford University Press, p. 161–192. Search in Google Scholar
Hacking, Ian (1995): Rewriting the Soul. Princeton, NJ: Princeton University Press. Search in Google Scholar
Haslanger, Sally (2000): “Gender and Race: (What) Are They? (What) Do We Want Them To Be?”. In: Noûs 34, p. 31–55. Search in Google Scholar
Haslanger, Sally (2012): Resisting Reality. Social Construction and Social Critique. New York: Oxford University Press. Search in Google Scholar
Hoeltje, Miguel, Benjamin Schnieder and Alexander Steinberg (Eds.) (2013): Varieties of Dependence: Ontological Dependence, Grounding, Supervenience, Response-Dependence. München: Philosophia Verlag. Search in Google Scholar
Hoeltje, Miguel (2016): “Subject Dependence and Missing Explanations”. Unpublished manuscript. Search in Google Scholar
Kendell, Robert (1975): “The Concept of Disease and its Implications for Psychiatry”. In: British Journal of Psychiatry 127, p. 305–315. Search in Google Scholar
Kendler, Kenneth S., Peter Zachar and Carl Craver (2011): “What Kinds of Things are Psychiatric Disorders?”. In: Psychological Medicine 41, p. 1143–1150. Search in Google Scholar
Kirmayer, Laurence and Daina Crafa (2014): “What Kind of Science for Psychiatry?”. In: Frontiers in Human Neuroscience 8, Article 435, DOI: 10.3389/fnhum.2014.00435. Accessed on 12.9.2016. Search in Google Scholar
Koch, Heiner (2013): “Gemeinsame antipatriarchale Kämpfe und die Ontologie von Gender und Herrschaft”. In: Gabriele Jähnert, Karin Aleksander and Marianne Kriszio (Eds.): Kollektivität nach der Subjektkritik. Bielefeld: Transcript, p. 153–177. Search in Google Scholar
Piper, August and Harold Merskey (2004): “The Persistence of Folly: A Critical Examination of Dissociative Identity Disorder. Part I. The Excesses of an Improbable Concept”. In: Canadian Journal of Psychiatry 49, p. 592–600. Search in Google Scholar
Pickering, Neil (2006): The Metaphor of Mental Illness. Oxford: OUP. Search in Google Scholar
Searle, John (1995): The Construction of Social Reality. London: Penguin. Search in Google Scholar
Searle, John (2010): Making the Social World: The Structure of Human Civilization. Oxford: OUP. Search in Google Scholar
Stier, Marco (2013): “Normative Preconditions for the Assessment of Mental Disorder”. In: Frontiers in Psychology 4, Article 611. DOI: 10.3389/fpsyg.2013.00611. Accessed on 12.9.2016. Search in Google Scholar
Szasz, Thomas (1972): The Myth of Mental Illness. London: Paladin. Search in Google Scholar
Thornton, Tim (2007): Essential Philosophy of Psychiatry. Oxford: OUP. Search in Google Scholar
U.S. Department of Health and Human Services (Ed.) (2001): Mental Health: Culture, Race, and Ethnicity – A Supplement to Mental Health: A Report of the Surgeon General. Rockville, MD: U.S. Department of Health and Human Services, Substance Abuse and Mental Health Services Administration, Center for Mental Health Services. http://www.ncbi.nlm.nih.gov/books/NBK44243/. Accessed on 10 August 2016. Search in Google Scholar
van Riel, Raphael (2016): “What is Constructionism in Psychiatry? From Social Causes to Psychiatric Classification”. In: Frontiers in Psychiatry 7, Article 57, DOI: 10.3389/fpsyt.2016.00057. Accessed on 12.9.2016. Search in Google Scholar
Wakefield, Jerome C. (1992): “The concept of mental disorder: On the boundary between biological facts and social values”. In: American Psychologist 47, p. 373–388. Search in Google Scholar
Wakefield, Jerome C. (2007): “The Concept of Mental Disorder: Diagnostic Implications of the Harmful Dysfunction Analysis”. In: World Psychiatry 6, p. 149–156. Search in Google Scholar
[Correction added after online publication 21 October 2016: The reference to Hoeltje (2016), which was incorrectly quoted in the ahead-of-print publication, has been amended as an unpublished manuscript.]
©2016, Raphael van Riel, published by De Gruyter.
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.