Therapeutic Chaos

• Main Authors: Guido Strunk, Anna Lichtwarck-Aschoff
• Format: Article
• Language: English
• Published: Lund University Library 2019-12-01
• Series: Journal for Person-Oriented Research
• Subjects: non-linear dynamic system theories; process research; client-therapist interaction; complexity; self-organization; critical fluctuations
• Online Access: https://www.person-research.org/journal/files/5_2/filer/99.pdf
• ISSN: 2002-0244; 2003-0177

Abstract: The conventional view on interventions as mechanistically causing interchangeable clients to get better has come under attack. Group-based and linear approaches fall short in adequately describing the idiosyncratic and dynamic nature of treatment processes. Non-linear dynamic system theories in contrast hold great potential to better conceptualize and understand the generalities and idiosyncrasies of psychotherapeutic change processes. The aim of this study was to examine whether we can detect markers of complex dynamical systems behavior in two single-case therapies. All sessions from both therapies were coded with sequential plan analysis using a 10s sampling frequency. The coding system incorporates verbal and non-verbal behaviors and allows for the representation of contextualized interactive behaviors. The high sampling frequency results in long time series, which allowed us to apply non-linear analysis techniques. We found strong support for complex behavior and the existence of a butterfly effect, i.e., a relatively short prediction horizon in which reliable predictions about the system’s future behavior could be made. Further, critical fluctuations as a marker for phase-transitions were detected that were accompanied with different interactional patterns in both therapies. Finally, there was strong support for self-organized pattern formation, with a few interactional patterns dominating the interaction. Considering that we are intervening on complex dynamical systems means that we have to (1) acknowledge the principal individuality of change processes, (2) accept the fundamental limitations of the mechanistic input-output model of treatment effects and (3) appreciate the impossibility of long-term predictions of treatment responses.