A COMPUTATIONAL MODEL of PANIC Advancing The

A COMPUTATIONAL MODEL of PANIC Advancing The

Running head: A COMPUTATIONAL MODEL OF PANIC Advancing the Network Theory of Mental Disorders: A Computational Model of Panic Disorder Donald J. Robinaugh1, Jonas M. B. Haslbeck2, Lourens J. Waldorp2, Jolanda J. Kossakowski,2 Eiko I. Fried3, Alexander J. Millner4, Richard J. McNally4, Egbert H. van Nes5, Marten Scheffer5, Kenneth S. Kendler6, Denny Borsboom2 1Massachusetts General Hospital, Harvard Medical School 2University of Amsterdam 3Leiden University 4Harvard University 5Wageningen University 6Virginia Commonwealth University May 22nd, 2019 Author Note Correspondence concerning this article should be addressed to Donald J. Robinaugh, Department of Psychiatry, Massachusetts General Hospital, 1 Bowdoin, MA 02114. E-mail: [email protected]. This manuscript was supported by a National Institute of Mental Health Career Development Award (1K23MH113805-01A1) awarded to D. Robinaugh and an ERC Consolidator Grant no. 647209 awarded to D. Borsboom. A COMPUTATIONAL MODEL OF PANIC 2 Abstract The network theory of psychopathology posits that mental disorders are complex systems of mutually reinforcing symptoms. This overarching framework has proven highly generative but does not specify precisely how any specific mental disorder operates as such a system. We address this gap in the literature by developing a network theory of Panic Disorder and formalizing that theory as a computational model. We first review prior psychological theory and research on Panic Disorder in order to identify its core components as well as the plausible causal relations among those components. We then construct and evaluate a computational model of Panic Disorder as a non-linear dynamical system. We show that this model can explain a great deal, including individual differences in the propensity to experience panic attacks, key phenomenological characteristics of those attacks, the onset of Panic Disorder, and the efficacy of cognitive behavioral therapy. We also show that the model identifies significant gaps in our understanding of Panic Disorder and propose a theory-driven research agenda for Panic Disorder that follows from our evaluation of the model. We conclude by discussing the implications of the model for how we understand and investigate mental disorders as complex systems. A COMPUTATIONAL MODEL OF PANIC 3 Advancing the Network Theory of Mental Disorders: A Computational Model of Panic Disorder The network theory of mental disorders posits that symptoms cohere, in part, because of causal relations among the symptoms themselves (Borsboom, 2017). From this perspective, mental disorders are analogous to an ecosystem. They do not appear as a coherent whole because of a shared underlying essence, but because of the web of causal interactions among the features of the disorder (Kendler, Zachar, & Craver, 2011). The notion that there are etiologically important causal interactions among symptoms has prompted the development of new methods for assessing the structure of relationships among symptoms (Epskamp, Maris, Waldorp, & Borsboom, 2016; Epskamp, Rhemtulla, & Borsboom, 2017; Marsman, Maris, Bechger, & Glas, 2015) and a host of empirical studies applying those methods across numerous psychiatric disorders (for an overview, see Fried et al., 2017). Moreover, the core idea that there are causal relations among symptoms has expanded into an overarching theory of mental disorders, how they develop, and how they remit (Borsboom, 2017). However, network theory remains abstract. It provides a conceptual framework for thinking about mental disorders but does not posit specific relationships among symptoms. Empirical network studies provide information about these relationships, but are not rich enough on their own to fully inform a network theory, as this requires a substantively interpreted model: a model that does not merely statistically associate variables, but rather specifies the mechanisms through which variables influence one another. Consequently, the network approach has produced statistical models that suggest putative network structures, but no theories that posit precisely how any given mental disorder operates as a complex system of interacting symptoms. We aim to address this gap in the literature by developing such a theory for Panic Disorder. Panic Disorder is a suitable starting point for several reasons. First, theories are A COMPUTATIONAL MODEL OF PANIC 4 about phenomena (Bogen & Woodward, 1988; Haig, 2005) and panic attacks are a robust phenomenon: a "stable, recurrent, and general feature of the world,” (Haig, 2005, p. 374). Experiences resembling panic appear in medical consultation reports as far back as the mid- 18th century (Coste & Granger, 2014) and have been described consistently in the medical literature since the late 19th century (Berrios, 1999; Leroux, 1889). Accordingly, these attacks are a suitable phenomenon about which to develop a theory. Second, Panic Disorder symptoms are structurally inter-connected in the network of symptoms from the Diagnostic and Statistical Manual (DSM; Boschloo et al., 2015), suggesting that these symptoms commonly co-occur and, thus, represent precisely the type of phenomenon that network theory seeks to explain. Third, theorists have posited causal relations among Panic Disorder symptoms (e.g., a mutually reinforcing relationship between panic attacks and avoidance behavior; Goldstein & Chambless, 1978). Indeed, some of these relationships are embedded in the disorder’s diagnostic criteria (e.g., to meet the diagnostic criterion, avoidance behavior must be related to panic attacks; Borsboom, 2008). Fourth, there is strong body of research on the etiology, phenomenology, and epidemiology of panic attacks (Barlow, 1988; McNally, 1994). Moreover, there are well-established interventions that treat Panic Disorder (e.g., cognitive behavioral therapy; Barlow, 1997) and reliable ways to induce panic attacks (e.g., biological challenges; Gorman, Liebowitz, Fyer, & Klein, 1987). This work provides criteria by which to evaluate our theory, as any theory failing to account for these empirical findings will be found wanting. The prior literature is thus sufficiently rich to both inform theory development and to provide a basis for theory evaluation. Our development of Panic Disorder theory will proceed as follows. In Section 1, we review theory and research on Panic Disorder, identifying its essential components and the posited functional relations among them. In Section 2, we integrate prior work and propose a theoretical model of Panic Disorder as a complex system, formalizing the relationships A COMPUTATIONAL MODEL OF PANIC 5 among individual symptoms of Panic Disorder in a mathematical model. In Section 3, we implement this model in R, a freely available software environment for statistical computing (R Core Team, 2014). Computational modeling is an effective tool for theory development because it allows us to simulate the model’s behavior and assess what the theory can and cannot explain (Epstein, 2008). We will show that the model can explain a great deal, including core phenomenological qualities of panic attacks, individual differences in the vulnerability to panic attacks, the onset of Panic Disorder following an initial panic attack, and the efficacy of cognitive behavioral therapy for Panic Disorder. We will also show that explicating Panic Disorder theory in this way reveals significant gaps in our understanding. Moreover, the model fails to explain some key features of Panic Disorder. These shortcomings suggest further theory development is needed. In Section 4, we propose a theory-driven research agenda for Panic Disorder that follows from our evaluation of the model. We give particular focus to the need for further theory development and illustrate how such development could proceed using the model proposed here as well as a previously proposed mathematical model of panic attacks (Fukano & Gunji, 2012). Finally, in Section 5, we discuss the implications of the model for our understanding of mental disorders. Section 1: A Survey of Panic Disorder Theory and Phenomenology The symptoms identified in diagnostic manuals provide a tractable starting point for identifying the components of a mental disorder’s causal system (Borsboom, 2017). Current diagnostic criteria for Panic Disorder are rooted in work by Sigmund Freud from the late 19th century (Frances et al., 1993). In his work on anxiety neurosis, Freud described angstfallen: sudden attacks of anxiety characterized by “ideas of the extinction of life… or of a threat of madness” accompanied by intense somatic symptoms (Freud, 1962, pp. 93-94). The key features of his description can be traced from an early precursor to the DSM (American Medico-Psychological Association, 1918), through the DSM-II (American Psychiatric A COMPUTATIONAL MODEL OF PANIC 6 Association, 1968), to the first diagnostic criteria for anxiety neurosis (Feighner et al., 1972), where these attacks were required to include apprehension, fearfulness, or a sense of impending doom accompanied by at least four of the following six somatic symptoms: dyspnea, heart palpitations, chest pain, choking or smothering sensations, dizziness, and paresthesias. In the DSM-III, these attacks were separated into the newly created Panic Disorder diagnosis (Kendler, 2017; McNally, 1994), but the criteria used by Feighner and colleagues were largely retained, defining panic attacks by the sudden and unexpected onset of two symptoms: somatic symptoms and fear (American Psychiatric

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    76 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us