<p>The network theory of psychopathology posits that mental disorders are stable states of symptom activation arising from causally interconnected symptoms, where individuals with more strongly connected symptom networks are at a higher risk of developing a mental disorder. Researchers have turned to idiographic network estimation to assess this theoretical position, yet it remains unclear whether the dynamics of these models align with the network theory. In this paper, we use stability landscapes to systematically map the parameters of idiographic network models onto the network theory. Specifically, we examine how the dynamics implied by the Ising model (with <InlineEquation ID="IEq1"> <EquationSource Format="TEX">$\{0, 1\}$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mo stretchy="false">{</mo> <mn>0</mn> <mo>,</mo> <mn>1</mn> <mo stretchy="false">}</mo> </mrow> </math></EquationSource> </InlineEquation> and <InlineEquation ID="IEq2"> <EquationSource Format="TEX">$\{-1, 1\}$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mo stretchy="false">{</mo> <mo>-</mo> <mn>1</mn> <mo>,</mo> <mn>1</mn> <mo stretchy="false">}</mo> </mrow> </math></EquationSource> </InlineEquation> encodings) and the Graphical Vector Autoregressive (GVAR) model relate to symptom severity and variability. Our results show that only for a subset of parameter values in the <InlineEquation ID="IEq3"> <EquationSource Format="TEX">$\{0, 1\}$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mo stretchy="false">{</mo> <mn>0</mn> <mo>,</mo> <mn>1</mn> <mo stretchy="false">}</mo> </mrow> </math></EquationSource> </InlineEquation> Ising model, higher connectivity is directly related to higher symptom severity, thereby aligning with the network theory. The <InlineEquation ID="IEq4"> <EquationSource Format="TEX">$\{-1, 1\}$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mo stretchy="false">{</mo> <mo>-</mo> <mn>1</mn> <mo>,</mo> <mn>1</mn> <mo stretchy="false">}</mo> </mrow> </math></EquationSource> </InlineEquation> Ising model only partially aligns with the network theory, as increased connectivity is related to both high and low symptom severity. In contrast, connectivity in the GVAR model is independent of symptom severity but instead reflects symptom variability over time. These findings demonstrate that the theoretical implications of network connectivity depend strongly on the chosen statistical model and its parameterization. Together, our results emphasize the need for systematic investigations that link theory to statistical models, and present stability landscapes as a useful tool to do so.</p>

Mapping the dynamics of idiographic network models to the network theory of psychopathology

期刊论文
出版模式：

开放获取
发表日期：

2026年4月21日

查看全文

Ria H. A. Hoekstra,
Jill de Ron,
Sacha Epskamp,
Donald J. Robinaugh,
Denny Borsboom

摘要

The network theory of psychopathology posits that mental disorders are stable states of symptom activation arising from causally interconnected symptoms, where individuals with more strongly connected symptom networks are at a higher risk of developing a mental disorder. Researchers have turned to idiographic network estimation to assess this theoretical position, yet it remains unclear whether the dynamics of these models align with the network theory. In this paper, we use stability landscapes to systematically map the parameters of idiographic network models onto the network theory. Specifically, we examine how the dynamics implied by the Ising model (with $\{0, 1\}$ ${0, 1}$ and $\{-1, 1\}$ ${- 1, 1}$ encodings) and the Graphical Vector Autoregressive (GVAR) model relate to symptom severity and variability. Our results show that only for a subset of parameter values in the $\{0, 1\}$ ${0, 1}$ Ising model, higher connectivity is directly related to higher symptom severity, thereby aligning with the network theory. The $\{-1, 1\}$ ${- 1, 1}$ Ising model only partially aligns with the network theory, as increased connectivity is related to both high and low symptom severity. In contrast, connectivity in the GVAR model is independent of symptom severity but instead reflects symptom variability over time. These findings demonstrate that the theoretical implications of network connectivity depend strongly on the chosen statistical model and its parameterization. Together, our results emphasize the need for systematic investigations that link theory to statistical models, and present stability landscapes as a useful tool to do so.