Can family therapy, like Caesar's Gaul, be divided into three parts? What are the gains and what are the potential costs? How may this classification perform; what, in Foucaldian terms, will be its disciplinary power? The social constructionist critique tells us that language - far from being a neutral describer of our world - has the power to quietly enchant us into seeing in particular ways. It suggests that we project our socially constructed categories outside of ourselves, then forget to notice this human tampering and treat these categories as if they are real. Instead, it proposes that what is real can never be fully captured by words. Our language, in apparently illuminating some aspects of our world,  casts others into shadow. As an integral part of the discourses through which we live our lives it implicitly prescribes some actions and proscribes others. 

Make a mental note, for example, of all the family therapists you know, including yourself, who are happy to be described as "first order family therapists". I've never met one although I can think of some who would be happy enough to describe their work as either structural or strategic family therapy (or as being strongly influenced by these models). The term "first order family therapy" came into common use in the 1980s by those wishing to distance themselves from certain practices said to be associated with the early cybernetic models such as the use of power, control, normative theory etc. Leaving aside for a moment the descriptive merits of the term, my point is that "first order therapy" became a badge of shame; a (self-) disciplinary matter. It wasn't long before you could divide family therapists into the very many who were second order and those who grumbled quietly about epistobabble. Earlier still, the term "linear" helped, I think, to sort out the psychodynamic family therapists of the 1950's from the new Batesonian influenced therapists. To be "circular" in the 1960's  and 70's was to be on the side of the angels, but you had better watch your step - no backsliding. In such ways are the culture and defining boundaries of family therapy formed. 

Rudi Dallos and Amy Urry hope to persuade us not to be docile bodies (as Foucault puts it)  in the face of this tendency for each new family therapy movement to construct all previous developments as theoretically, technically or ethically deficient and, therefore, unworthy of further consideration. Instead, their rhetoric puts the dismissal of early family therapy ideas on a par with dumping key members of one's family or, at least, being so reckless that they float off after bath time. (Not, you'd think, a matter of indifference to a psychotherapy that values family relationships highly.)  I take their motives to be twofold: first, to legitimate personal syntheses or eclecticisms which, as they suggest, can be shameful practices in the face of internalised discourses which whisper criticisms at oneself such as "expert" and "linear"; and, secondly, to attempt to hold together the field of family therapy in the face of strong pressures to split along a developing fault line between social constructionist (postmodern) influenced therapies and the rest. As an eclecticist myself, of sorts, who finds much value in the postmodern critique, I could hardly support these motives more strongly. 

I think, though, that the field of family therapy is now so vast and complex that there can be no very good simple classification that does not, in the process of simplification, conceal or subdue many of its nuances thereby hiding from us and our clients some of its potential resources. However, it may be possible to produce some good-enough simple classifications to help family therapy feel like a more coherent place and the Dallos and Urry model may be one - perhaps only time will tell. Therefore, in raising doubts about the classification and in drawing attention to that which - through not fitting - is squeezed out, I am not intending to do damage to their framework. I believe that there can be no classification that would not need to be similarly critiqued. One analogy is that of a piece of tapestry in which the attractive coherence of  the side facing outwards always requires an incoherent obverse side; a place, too, where messy strands get tucked out of sight. I hope that exploring round the back of this framework will guard against too much complacency as well as bringing some of the resources it conceals back into the light. 
notes 
I will be seeking to deconstruct across each of the three category divisions and, for this purpose, I have invented "zero order" as a temporary term to stand for pre-Batesonian family therapy ideas. 
 

The sharp eyed reader will have noticed that while Dallos and Urry use the term "family systems therapy" to denote the territory that needs to be framed by their classification, I have used the older term "family therapy" since I am in favour of a definition that spreads a wider net. Hoffman in a recent paper (Gergen et al, 1996) also uses "family therapy" to describe the field. What is going on? Hoffman does not have an inclusive motive; she is interested in a definition that moves family therapy away from systems and cybernetics. Her imagery for this move is no longer the cybernetic circle but a river of flowing and endlessly shifting narrative (Hoffman, 1990). The present authors argue that social constructionism, with its emphasis on context and interaction, need not be located away from cybernetics and I agree with them. However, making cybernetics a rallying point will leave out what Broderick and Schrader (1991) refer to as the "founding decade" of family therapy (1952-1961) in which there was a considerable development in psychoanalytic thinking - from individuals to relationships. In addition, a cybernetic map provides no obvious place for contextual therapy, gestalt family therapy or behavioural family therapy all of which seem to be sliding into obscurity - at least in the UK. 
 

Dallos and Urry point out several continuities between first and second order cybernetic practices and this helps to deconstruct the dominant discourse which emphasises sharp differences between the two. But, since the authors also retain the terms "first order" and "second order" let us examine this given discourse to see if further deconstruction is possible. 

First order family therapies (structural, strategic and early Milan) are functionalist, concerned with normative theory, are relatively disinterested in meaning systems, take an objective view of the family system and proceed through the exercise of power and control on the part of the therapist. Right? Well, yes - except for all the times when they don't. On which side of the first/second order divide would you place the following statements? 

"But therapists who have been trained in interpersonal channels of communication know that the act of observation influences the material observed, so that they are always dealing with approximates and probable realities." (p80) 

"The family is not a static entity. It is in the process of continuous change, as are its social contexts. To look at human beings apart from change and time is solely a construct of language." (p20) 

"Fortunately for family therapy, therapists have not been able to develop diagnostic categories for families that can pigeonhole some family forms as normal and others as deviant; with any luck, we never will develop them." (p263) 

All are quotes from those "first order therapists" Minuchin and Fishman (1981) who, judged against the given criteria for first and second order cybernetics, seem to be involved in some epistemological cross-dressing. Of course one should also argue that first order is what first order does: that what happens in the session may be more relevant than what happens on the page. But this was the very point of Golann (1988) who could see the advantage of first order power and control at least being out in the open compared with what he saw as the possibility of a more covert use of therapist power in the second order therapies. Maturana's dictum that there can be "no instructive interaction" was widely used to support the shift to second order practice but it could more logically have been used to exonerate first order therapists from accusations of acting powerfully. Its implications are that therapists cannot create change through power and control no matter how hard they try; the exercise of power is not ethically wrong, according to Maturana, just impossible. 

It is arguable, too, that the biggest impetus for a shift to second order family therapy came from a very un-cybernetic feminism which had simply had enough of circular causality. The degree to which family therapists tend to think of the shift towards a more collaborative stance as a constructivist development may be the same amount that the field is not quite willing to recognise the vigorous impact of feminism's critique of cybernetics. (E.g. Bograd, 1984; Hare-Mustin, 1978.) 

Finally in this section, space only allows me to briefly mention that Minuchin seems to have little interest in functionalism; nor much in the sociological theory of structural functionalism. He does seem to take some inspiration from the then fashionable structuralism (the notion inspired by Saussure that systems have hidden structures like a language) but he has no theory of the homeostatic value of symptoms in the way that strategic and early Milan family therapists do. For him, they arise in response to the system being stressed and have no other significance. 

The authors interchange throughout the paper the terms "1st, 2nd, 3rd order cybernetics", "1st, 2nd, 3rd cybernetics" and "1st, 2nd, 3rd phase". They mostly use these terms to signify phases of development of the discourse of family therapy. However, they seek legitimacy for the nomenclature by suggesting that our current phase of interest in social constructionist and narrative ideas can be seen as an extension to first and second order cybernetics. I now plan to critique the use of the term "third order" to stand for social constructionism and narrative and will argue that it acts to conceal relational versions of constructivism. The reader will need to forgive me for starting with a potentially tedious journey back in time to dust the cobwebs off some early nomenclatures in cybernetics. 

"Cybernetics"1 is a term which encompasses communication theory, self-regulation and control theories which, especially in the 1940s and 50s, were thought to be equally applicable to machines, organisms and social structures. It was Bateson (1972) who particularly saw the relevance of cybernetics for understanding the complexity of human relationships. The terms "1st and 2nd order" have a more distinct origin, being derived from a special kind of discontinuous relationship between variables in mathematics called step functions. A good example of a step function change is a shift of gear in a car. In 1st gear a certain range of speed is possible but 2nd gear allows a higher range of car speed for the same engine speed and so on. "First and second order change" was the earliest use of step function terms in cybernetics; the former referring to a trivial change in one part of  a system, the latter to a more profound change in the rules or structure of the system itself. (Ashby, 1952) There was initially much confusion when the terms 1st and 2nd order came to be also used to refer respectively to systems independent of an observer and observer-dependent systems (von Foerster, 1981).2 

When a therapist says (1) "this is a cross-generational coalition" she is first order, when the same therapist steps back into herself and says (2)"I am constructing what I "see" as a cross-generational coalition" she is second order. (Unless she has already been frightened off from thinking structurally by the disciplinary power of second order discourse: quite a different issue from second order cybernetics itself.)  But what would another step be like which would justify the term 3rd order? Well, extending the logic of the nomenclature it might look something like this:  (3) "The part of "I" that is constructing what "I see" as a cross-generational coalition is constructed through participation in the discourse of structural family therapy."  This would seem to be quite consistent with social constructionism but I now want to argue that both (2) and (3) could just as easily lie within some constructivisms and, therefore, within 2nd order cybernetics. 

The authors consistently and, I believe, wrongly assume that constructivism (second order cybernetics) is only concerned with intra-psychic processes of the isolated individual and, following Hoffman (1990), consider that social constructionism is distinct and relatively discontinuous - in other words, justifying a step function shift. In her 1990 paper, Hoffman used some powerfully bleak images to represent constructivism: ".. I did not like the idea that people were stuck in a biological isolation booth. If you took this idea to an extreme, you could say that therapists and clients were like people in bathyspheres trying to communicate underwater." [My emphasis.] Much of the field followed her into the apparently much more .. well .. sociable notion of social constructionism. Her envisaged "Family Therapy: Part 2" would leave behind in "Part 1" both mechanical (cybernetic) and biological (constructivist) metaphors in favour of the literary metaphors of narrative, deconstruction etc. So it has largely proved to be, and it is this potential rift that Dallos and Urry's classification is attempting to bridge. 

But five years previously Hoffman seems already to have had in mind a "Family Therapy: Part 2" but one which, in this much earlier version, was to include constructivism. It is clear from her 1985 paper that Hoffman understood very well that while some constructivisms (those of Kelly and von Glasersfeld)  were intra-psychic, others (those of Maturana, Varela and von Foerster) had well developed social (inter-psychic) aspects. Maturana has a theory of connotative language which, in some respects, is similar to Wittgenstein's theory of language games - which in turn has had much influence on social constructionists. Varela developed a theory of autonomous systems in which individuals are hardly isolated - "we live and breathe in dialogue and language." (Varela, 1979). And Hoffman (1985) herself quotes von Foerster (1981) approvingly who defines reality as "a consistent frame of reference for at least two observers". Immediately after the von Foerster quote, Hoffman summarises her view - derived entirely from constructivism since, at that stage, she had yet to discover social constructionism: "Our ideas about the world are shared ideas, consensually arrived at and mediated through givens like culture and language." 

But by the time of her 1990 paper, Hoffman's constructivism had turned into a straw man; a social isolate that few would miss. The discontinuous shift from constructivism to social constructionism is a shift of beliefs and practice in family therapy (discourse) rather than the theoretical paradigm shift from individual constructions (in constructivism) to social constructions that many in our field have come to believe. This myth of the isolated constructing individual has, I think, acted to conceal the relational forms of constructivism and encouraged the sudden and large scale loss of interest in constructivist thinking in family therapy. In summary, social constructionism shares a strong family resemblance with some of the constructivisms of second order cybernetics and, accordingly, I believe there is little case for the step function shift which would justify a "3rd order cybernetics". 

I can now elaborate an earlier point. In the psychoanalytic theory of splitting, a way is required to distance oneself from that which was previously connected.  The disappearance of relational constructivisms has helped many to believe that social constructionism and narrative represent an entirely new departure and this, in turn,  constructs the Part 1/Part 2 schism as given and natural. We can note how increasingly popular it has become for family therapy ideas from "Part 1" to be perceived by many contemporary family therapists as other and alien: museum pieces from a history of dubious therapist ethics. 

But let me clear that my motive is not to squeeze social constructionism into an overarching supra-theory of cybernetics: a one-size-fits-all metanarrative. Neither do I think this is the motive of Dallos and Urry, despite their three orders of cybernetics.  I like Lyotard's (1979) definition of postmodern as "incredulity toward metanarratives". Psychoanalytic, cybernetic, social constructionist/narrative and smaller movements in family therapy have asked, each in turn, for our exclusive and unswerving loyalty. By retaining our scepticism and refusing to commit to any one we take the interesting choice of remaining open to that which might be of use in them all. 

So let me conclude by strongly supporting once more the project of Dallos and Urry to find some kind of loose coherence in family therapy and to look with fresh eyes at the resources of the first forty years of its thinking and practice. I hope that this deconstruction of their nomenclature will, if anything, assist this process. We can be mobilised by their classifying model but there are hidden riches, too, in the less classifiable muddle that it stands upon. 
 

Ashby, W. (1952) Design for a Brain. London: Chapman and Hall. 

Bateson, G. (1972) Steps to an Ecology of Mind. New York: Ballantine. 

Bograd, M. (1984) Family systems approaches to wife battering: a feminist critique. American Journal of Orthopsychiatry, 54: 558-568. 

Broderick, C. B. and Schrader, S. S. (1991) The history of professional marriage and family therapy.  In A. Gurman and D. Kniskern (eds) The Handbook of Family Therapy, Volume II. New York: Brunner/Mazel. 

Gergen, K. J.,  Hoffman, L. and Anderson, H. (1996)  Is diagnosis a disaster?: a constructionist trialogue. <http://www.swarthmore.edu/SocSci/kgergen1/text5.html> (Chapter draft for F. Kaslow (Ed.) Relational Diagnosis, Wiley.) 

Golann, S. (1988) On second-order family therapy.  Family Process, 27: 51- 65. 
 

Hare-Mustin, R. (1978) A feminist approach to family therapy. Family Process, 17: 181-194. 

Hoffman, L. (1985) Beyond power and control: toward a "second-order" family systems therapy. Family Systems Medicine, 3: 381-396. 

Hoffman, L. (1990) Constructing realities: an art of lenses. Family Process, 29: 1-12. 

Lyotard, J. (1979) The Postmodern Condition: A Report on Knowledge. Manchester:Manchester University Press. 

Maruyama, M. (1960) Morphogenesis and morphostasis. Methods, 12: 251-296. 

Minuchin, S. and Fishman, H. C. (1981) Family Therapy Techniques. Cambridge, MA: Harvard. 

Varela, F. (1979) Principles of Biological Autonomy. New York: North Holland. 

von Foerster, H. (1981) Observing Systems. Seaside, CA: Intersystems. 

1 A term invented by the mathematician Norbert Wiener in 1948 from the Greek cy