Hacking on the Looping Effects of Psychiatric Classifications: What Is an Interactive and Indifferent Kind?

Abstract

This paper examines Ian Hacking’s analysis of the looping effects of psychiatric classifications, focusing on his recent account of interactive and indifferent kinds. After explicating Hacking’s distinction between ‘interactive kinds’ (human kinds) and ‘indifferent kinds’ (natural kinds), I argue that Hacking cannot claim that there are ‘interactive and indifferent kinds,’ given the way that he introduces the interactive‐indifferent distinction. Hacking is also ambiguous on whether his notion of interactive and indifferent kinds is supposed to offer an account of classifications or objects of classification. I argue that these conceptual difficulties show that Hacking’s account of interactive and indifferent kinds cannot be based on—and should be clearly separated from—his distinction between interactive kinds and indifferent kinds. In clarifying Hacking’s account, I argue that interactive and indifferent kinds should be regarded as objects of classification (i.e., kinds of people) that can be identified with reference to a law‐like biological regularity and are aware of how they are classified. Schizophrenia and depression are discussed as examples. I subsequently offer reasons for resisting Hacking’s claim that the objects of classification in the human sciences—as a result of looping effects—are ‘moving targets’.

Acknowledgements

I owe thanks to Ian Hacking, Robert Richards, Trevor Pearce, Rachel Ponce, Cecelia Watson, Dana Rovang, Anjan Chakravartty, Gregory Radick, Andrew Bailey, Karyn Freedman, Karen Houle, James McAllister, and two anonymous referees of this journal for helpful comments and suggestions. An earlier draft of this paper was presented at the annual meeting of the Canadian Society for the History and Philosophy of Science (CSHPS) at York University in May 2006.

Notes

[1] In the 1960s and 1970s, anti‐psychiatrists and labeling theorists discussed many of the same issues addressed by Hacking’s account of looping. Other social scientists (e.g., see Howard 1985) have articulated these issues in terms of the reflexivity of human subjects, i.e., human subjects are ‘directed or turned back upon themselves’ insofar as they are aware of and will react to how they are categorized. Hacking (1986, 2004) has presented his account as revisiting themes from Foucault’s (2006) historical studies on madness.

[2] Several commentators (Bogen 1988; Cooper 2004, 78–80; Schmaus 1992, 169–171) have argued that Hacking fails to demonstrate that human kind classifications differ significantly from natural kind classifications, given that some of the latter (e.g., ‘marijuana’ or ‘dog’) are subject to similar feedback effects as the former (cf. Ereshefsky 2004, 913–916). When Hacking allows indirect feedback effects (as in the case of autism) to count as looping effects, he must concede that natural kind classifications may possess analogous indirect looping effects. In response to these critics, however, Hacking could maintain that since humans kinds are aware of the way that they are classified in a way that natural kinds are not, human kind classifications possess another level of feedback (as a result of direct looping effects) that are not characteristic of natural kind classifications. This response would save Hacking’s contention that human kind terms significantly differ from natural kind terms.

[3] In section 5 of this article, I suggest that autism, schizophrenia, and mental retardation are not the best examples of interactive and indifferent kinds insofar as such kinds may more typically refer to kinds of people (e.g., depressed adults or anxious adults) who—in contrast with Hacking’s examples—are directly aware of how they are classified. For the purposes here, it is important to note that interactive and indifferent kinds can be associated with direct feedback effects (in addition to the indirect feedback effects stressed by Hacking).

[4] Hacking does not discuss in detail any of the contemporary biological research on autistic disorder; research indicates that this particular form of autism is associated with abnormalities in the medial temporal lobe, the brain stem, and the cerebellum (Carlson 1999, ch. 16).

[5] Hacking’s (1999) semantic resolution is also meant to explain how something can be both real and socially constructed (119), and he presents it as ‘putting a theory of reference alongside social construction’ (122). For a more comprehensive and critical discussion of Hacking’s semantic resolution, see Murphy (2001).

[6] There are other relevant inconsistencies in Hacking’s treatment of interactive and indifferent kinds worth mentioning. For example, Hacking (1999, 118–119) clearly has objects of classification (rather than classifications) in mind when he briefly presents interactive and indifferent kinds in diathesis‐stress terminology. Hacking suggests that the ‘predisposing cause’ (diathesis) of a condition can be understood as its indifferent part, while its ‘occasioning cause’ (stress) can be understood as its interactive part. Here, both the indifferent part (biological cause) and interactive part (social cause) of interactive and indifferent kinds refer to things in nature, rather than classifications. It is also worth mentioning that in this formulation ‘interactive’ does not even refer to looping effects, since occasioning causes refer to traumatic life events (e.g., the death of a spouse), and nothing to do with the looping effects of scientific classifications.

[7] Hacking has brought my attention to a forthcoming paper (Hacking 2007) that is not liable to the criticism articulated in the previous section. In his forthcoming paper, ‘Kinds of People: Moving Targets’, Hacking distinguishes five different elements of the human sciences that interact with one another: (a) classifications; (b) people; (c) institutions; (d) knowledge; and (e) experts. While I have criticized Hacking for failing to distinguish between classifications and people (or ‘objects of classification’), his forthcoming analysis avoids this difficulty. Accordingly, my criticism of Hacking in the previous section is limited to his presentation in The Social Construction of What? (Hacking 1999). Hacking’s latest analysis, however, is liable to the criticism (that I raise in section 6 of this paper) that he has not persuasively shown that objects of knowledge in the human sciences are ‘on the move’ or unstable.

[8] For a general discussion of pathway research, see Thagard (2003).

[9] Since the dopamine hypothesis was suggested—mainly on the basis of research on the older ‘typical’ anti‐psychotic drugs that work by blocking D₂ receptors—other neurotransmitters (e.g., serotonin) and brain areas have been implicated in schizophrenia (see Carlson 1999, 440–443; McKim 2000, 274–276).

[10] All of the antidepressant drugs mentioned here are monoamine agonists, which function to increase the activity of the monoamine neurotransmitters: dopamine (DA), norepinephrine (NE), and serotonin (5‐HT). Most antidepressants selectively alter NE and 5‐HT, while not effecting DA. For a more comprehensive discussion of the various antidepressant drugs and neurobiological theories of depression, see Carlson (1999, 451–454), McKim (2000, 289–292), and Millon et al. (1999, 167–175).

[11] As an ideal, I assume that natural kinds in psychiatry are discrete, non‐arbitrarily bounded classes of abnormal behavior that share a specific etiology (e.g., a genetic cause or a physiological cause); to adopt a phrase from Putnam (1975), natural kinds possess the same ‘general hidden structure’ (235, emphasis in the original). I further assume that the mental disorders listed in DSM IV (TR) (Diagnostic and Statistical Manual, 4th ed., text revision) (APA 2000) take on a variety of structural forms, and that only a few approach this ideal of natural kinds as a matter of degree (see Haslam 2002). To be clear, my analysis assumes that there are discrete mental disorders that are structurally similar with respect to etiology. I have listed paranoid schizophrenia (and, more tentatively, major depression) as candidates that approach this ideal; other disorders that I would consider are Down syndrome, panic disorder, and bipolar disorder. For a more comprehensive discussion of the issue of mental disorders and natural kinds, see Zachar (2000), Haslam (2002), Cooper (2005, ch. 2), and Murphy (2006, ch. 9). For discussion of Hacking’s own position on natural kinds, see Hacking (1990, 1991a, 1991b), Boyd (1991), and Cooper (2004, 74–78).

[12] My claim here rests on an oversimplified assumption regarding the ways in which looping effects operate. While my analysis suggests that looping effects express themselves as cultural differences, it is possible that some of the uniformities seen across cultures reflect particularly widespread shared forms of interactivity (e.g., global norms for expressing anxiety). Hence, looping effects can also manifest themselves as trans‐cultural similarities. For the purposes of this paper, I would say that that the uniformity of a condition across cultures can typically be understood as a manifestation of (stable) biological aspects of that condition.

[13] This issue complicated by the issue of ‘biolooping’ (Hacking 1999, 109–110). Biolooping is a phenomenon wherein changes in mental states result in changes in physiological states. For the purposes here, I assume that biolooping will not effect the typical physiological patterns associated with conditions such as schizophrenia or depression.

Additional information

Notes on contributors

Jonathan Y. Tsou

Jonathan Y. Tsou is a Ph.D. candidate in the Committee on Conceptual and Historical Studies of Science, University of Chicago.