University of Groningen A Dynamical Approach to Psychological Resilience Hill, Yannick

(1)

University of Groningen

A Dynamical Approach to Psychological Resilience

Hill, Yannick

DOI:

10.33612/diss.144252644

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.

Document Version

Publisher's PDF, also known as Version of record

Publication date: 2020

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

Hill, Y. (2020). A Dynamical Approach to Psychological Resilience. University of Groningen. https://doi.org/10.33612/diss.144252644

Copyright

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

(2)

548915-L-bw-Hill 548915-L-bw-Hill 548915-L-bw-Hill 548915-L-bw-Hill Processed on: 30-9-2020 Processed on: 30-9-2020 Processed on: 30-9-2020

Processed on: 30-9-2020 PDF page: 125PDF page: 125PDF page: 125PDF page: 125

(3)

(4)

7

This thesis started by outlining that resilience is marked by returning to the previous state after being perturbed by a stressor. Although scholars tend to agree that resilience is a process characterized by “bouncing back” to the previous level of functioning, there is currently conceptual ambiguity in the psychological literature. Combining the insights from the different approaches to studying resilience in humans, we concluded that a coherent theoretical approach needs to capture the dynamics of the person-environment interaction, explain how these dynamics can change over time with repeated stressors, and differentiate the resilience process from related, but different concepts (e.g., plasticity). Therefore, we proposed a dynamical systems approach, which offers a parsimonious theoretical framework tailored to studying resilience. The original formulation of the dynamical systems approach to studying resilience triggered an academic discussion with various experts from the field (Chapter 2). The insights derived from this discussion contributed to our definition of resilience as a dynamical process and delivered inspirations for the empirical studies presented in this thesis.

Because mapping out how the process of resilience unfolds is relatively new to psycho-logy, we first investigated how resilience and the psychological factors that are assumed to predict resilience (i.e., protective factors) change over time (Chapter 3). After establishing that resilience and the protective factors show signs of non-ergodicity, which is typical for dynamical systems, several studies were conducted based on the methodological toolbox offered by this approach (Chapters 4-6). Thus, every study in this thesis represents a new step in studying resilience in humans with a dynamical systems approach. Because of the novel tools used in these studies, it is important to note that the presented findings are to be interpreted with proper caution. To date, very few comparable studies exist that provide insights into the dynamical patterns of resilience in humans. Nevertheless, the studies presented in this thesis may provide an important stepping stone for future research on resilience in psychology.

1 | A Dynamical Approach to Psychological

Resilience

To recap, the behavior of a dynamical system emerges from the continuous interactions between its underlying components (e.g., Van Orden et al., 2003; Den Hartigh et al., 2017). In terms of resilience, these components refer to psychological and physiological components within a person as well as the environment in which he or she is embedded in. Thus, the dynamical systems approach is specifically built on the dynamic person-environment interaction. The change that a system undergoes with repeated stressors can be explained with the third key property of dynamical systems, so-called attractor

(5)

128

Chapter 7

dynamics (Nowak & Vallacher, 1998; Vallacher & Nowak, 1997; Vallacher et al., 2015). Put simply, an attractor describes a stable pattern of recurring cognitions and behaviors (see for a graphic illustration Chapter 2, Figure 3). When a system is exposed to repeated stressors, these attractor states can change so that a minor stressor that would usually be overcome can cause a drastic change in the system’s state (Dai et al., 2012; Scheffer et al., 2009, 2012; Van de Leemput et al., 2014). This means that the dynamic interplay between the person and the environment from which resilience emerges may change over time.

Although the academic discussion presented in Chapter 2 pointed out that psychological processes usually refer to the causal chain of different psychological components (see Bryan et al., 2018), a dynamical systems approach focuses on the temporal patterns that provide insights into the underlying dynamics of a system. This means that the temporal patterns that can be identified when mapping a system over time, rather than the exact composition of its underlying components at a given moment in time, can provide insight into its current (and potentially future) state(s). As argued in Chapter 1, mapping out how the adaptational process unfolds over time may help to distinguish resilience from related constructs. Thus, all the presented empirical studies in this thesis mapped out changes in the relevant variables over time. Specifically, Chapter 3 extends the notion of studying temporal processes by emphasizing the importance of distinguishing statistical findings obtained from averaged group-level processes and findings obtained on an individual level. This points to the fact that dynamical processes need to be analyzed for each individual first, before the findings are summarized for a sample (i.e., non-ergodicity, Den Hartigh, Hill et al. 2018; Fisher et al., 2018; Glazier & Mehdizadeh, 2019; Liu et al., 2006; Molenaar & Campbell, 2009).

A key insight that can be derived from mapping an individual system’s behavior over time is being able to identify not only if a system is more or less likely to demonstrate resilience in response to a given stressor, but also when its resilience is changing. The above-mentioned notion of attractor dynamics has already been used in other biological systems to predict resilience losses in response to repeated stressors that can signal drastic transitions (e.g., Scheffer et al., 2009, 2012). Hereby, temporal patterns indicating a slowing in the resilience process (i.e., critical slowing down) have been identified as early warning signals of resilience losses. Identifying these periods of resilience losses may help to apply preventive interventions. Hence, we utilized a new measure to determine resilience in response to a stressor to assess resilience losses during a motor task in Chapter 4. However, contrary to our expectations, the resilience of the participants in terms of movement accuracy improved in response to repeated stressors. In terms of dynamical systems theory, this means that instead of weakening the current attractor state that the participants resided in regarding their movement accuracy, the current

(6)

7

attractor became even stronger. This is contrary to the findings of research on resilience losses in other dynamical systems, such as eco-systems. This may indicate that repeated stressors do not necessarily have a negative impact on human motor performance, but may actually be utilized to improve it (Davids et al., 2008).

In Chapter 5, the research on early warning signals was extended. Specifically, based on the request by Galli and Pagano (2018) to demonstrate how the dynamical systems approach may be applied to team performance, we assessed early warning signals of resilience losses in dyadic coordinative performance. Hereby, another novel measure was introduced that is closely connected to an inherent property of resilience: complexity (see Chapter 2). Complexity can be derived from temporal patterns in the collective variable produced by a system (Delignières et al., 2004; Goldberger et al., 2002; Manor et al., 2010). A series of studies from various domains has shown that adaptive biological systems show a combination of flexibility and stability when tracked over time (see Figure 19). This temporal pattern has been found in healthy motor systems (Davids et al., 2003; Den Hartigh et al., 2015; Kiefer & Myer, 2015), cardiovascular systems (Goldberger et al., 2002) and psychological systems (e.g., self-esteem, De Ruiter et al., 2015). The combination of flexibility and stability enables an athlete to adapt to stressors (flexibility) while simultaneously maintaining global functioning (stability). A deviation from this state to either extreme (i.e., either extreme stability or extreme flexibility) would indicate a loss of the adaptability and therefore diminished resilience in the system (Kiefer et al., 2018; Pincus et al., 2014, 2018, 2019; Pincus & Metten, 2010). For example, in line with the notion of metastability pointed out by Kiefer and colleagues (2018) in the discussion of Chapter 2, an overly stable athlete would reside within a rigid behavioral pattern and thus could not adapt to stressors, while an overly flexible athlete could not maintain necessary functioning when adapting to stressors. Accordingly, we found that negative performance transitions were marked by instabilities of complexity in the team’s coordinative performance, whereas positive transitions were associated with stable complexity. This reinforces the findings of Chapter 4 that repeated stressors do not necessarily have a negative effect on motor performance, but may be implemented strategically to trigger facilitative responses. Furthermore, the fact that we did not find a group-level association between the teams showing positive or negative performance transitions with the stressors that were induced points towards the idiosyncrasy of the resilience process outlined in Chapter 3.

(7)

130

Chapter 7

Figure 19 | Three simulated time-series of a collective variable. Graph B represents a time-series that demonstrates an optimal mix of stability and flexibility. Graph A (loss of stability) and C (loss of flexibility) show patterns that deviate from the optimal mixture.

Because the two previous studies (Chapter 4 and 5) both indicated the potentially positive effect of stressors, we explicitly focused on how to stimulate growth from stressors in Chapter 6. Using the extension of the dynamical systems approach to studying growth presented by Kiefer and colleagues (2018), we again found that for some athletes increasing stress-loading can have beneficial effects. Furthermore, in line with the idiosyncratic patterns displayed by dynamical systems over time in Chapter 3, we found that athletes who reached similar performance levels displayed different adaptability to varying stress-loading. This means that stress can indeed have a negative, but also potentially positive effect on athletic performance. However, the optimal loading is highly individual and thus needs to be tailored to each individual athlete.

In summary, the dynamical systems approach represents a parsimonious framework for studying resilience, which is based on the dynamic person-environment interactions (Chapter 2). It offers a precise theoretical explanation how resilience in a person changes in unique patterns over time (Chapter 3). These temporal patterns can be analyzed to test not only if a person is more or less likely to demonstrate resilience than others, but may provide tailored indications of when a person is losing resilience and becomes sensitive to minor stressors (Chapter 4 and 5). Finally, the dynamical systems framework does

Collective variable Time Collective variable Time Collective variable Time Loss of stability Loss of flexibility A B C

(8)

7

not only offer a theoretical explanation of how people can grow from stressors, but also provides tools to identify the optimal stress-loadings that can be beneficial for athletic performance (Chapter 6). Thus, this framework connects the insights from the existing literature on psychological resilience, distinguishes resilience from related concepts, and offers new avenues for researchers and practitioners in the domain of sports for both preventing undesirable effects of stress as well as leveraging the potential positive effects of stressors on an empirical basis.

2 | Research Implications for Sports Psychology

Based on the insights from this thesis, future research may build on the findings and methods. For instance, the notion of early warning signals may be extended in several ways. First, the presented studies have assessed the occurrence of early warning signals of resilience losses in laboratory movement tasks. Although such initial insights are important to understand the basic principles, extending the results to sports may be possible. Therefore, future studies should examine resilience changes by applying either areas under the curve calculations or complexity assessments in sports performance. For example, similar to the movement sensitivity manipulation applied in Chapter 4, studies utilizing rowing ergometers may repeatedly manipulate the drag factor of the flywheel to study the dynamics of the resilience process and movement complexity (cf. Den Hartigh et al., 2015). This may provide insights into early warning signals of resilience losses in athletic performance.

Additionally, the current studies on early warning signals (Chapter 4 and 5) have implemented repeated stressors at fixed time-intervals. However, critical slowing down is marked by repeated stressors in a short timeframe (Scheffer et al., 2009). Therefore, the reason why we observed improvements of resilience in Chapter 4 and numerous positive transitions in coordinative performance in Chapter 5 may be due to the fact that the time interval between the stressors was not small enough. Future studies may thus explore varying time intervals in order to experimentally induce critical slowing down in order to obtain further insight into the temporal signatures that athletes display when losing resilience.

Another important extension regarding timescales is the assessment of early warning signals over an extended period beyond a single performance. According to Van de Leemput and colleagues (2014), such indicators of resilience losses may also be detected in fluctuations of daily measurements. In terms of sports, this means that daily assessment of athletic performance may be used to identify resilience losses not only within a single performance incident, but also on larger time-scales, such as weeks or months in response

(9)

132

Chapter 7

to stressors that occur on a daily basis. Combining insights of changes in resilience on both short and long timescales may also be examined in order to understand the interconnection of these timescales (cf. Den Hartigh et al., 2016). For example, repeated resilience losses on a daily basis may foster resilience losses on a larger timescale and vice versa. Untangling this dynamic relationship between different timescales may indicate how short-term and long-term events interact to change resilience.

In line with demonstrating that the dynamical systems approach can be scaled from individual performers to teams (Chapter 5), a next logical step would be to scale load-response profiles to team sports. In Chapter 6, we elaborate on how this technique can be applied to team sports, such as rowing. Identifying similar load-response profiles in teams may further the credibility of its application to expose athletes to safe levels of stress in order to facilitate their performance without risking injuries.

3 | Research Implications in Other Fields

As pointed out previously, the presented theoretical approach and the according findings from the different studies may also have implications for studying resilience in other domains within psychology. For example, there is a current trend in clinical psychology to study psychological disorder from a dynamical systems approach (e.g., Kuranova et al., 2020; Van de Leemput et al., 2014; Wichers et al., 2019). This means that similar to sport performance, psychopathology emerges from constant person-environment interactions that may change over time. Thus, in order to determine early warning signals of resilience loss in psychological health to prevent the onset of psychopathological patterns, two specific insights from this thesis may be applied to clinical research. First, the area under the curve calculation presented in Chapter 4 may lend itself to determining resilience in psychological health. Using diary studies that assess a person’s well-being several times a day, researchers may detect stressors that perturb the equilibrium state and detect how strong the perturbation was and how long the individual needs to return to their previous state (see Practical Implications). This way, changes in resilience that may indicate the onset of a sudden transition may be identified (cf. Van de Leemput et al., 2014).

Second, another technique that may help clinicians to detect early warning signals of resilience losses is the (cross-)recurrence quantification analysis presented in Chapter 5. Specifically, changes in a system’s complexity may indicate that a system loses resilience (Pincus et al., 2014, 2018, 2019; Pincus & Metten, 2010). Thus, windowed analyses of complexity can help clinicians to detect when such critical fluctuations in the complexity of the underlying system occur to implement interventions to prevent subsequent transitions from occurring.

(10)

7

The implications for the detection of early warning signals of resilience losses is not limited to sport and clinical psychology. In the domain of work and organizational psychology, these techniques may be applied to detect negative transitions in work performance to prevent burnouts in high demanding jobs (cf. Guastello, 1995). Furthermore, the implications may expand beyond psychology. The concept of critical slowing down has been applied largely to detecting transitions in eco-systems due to climate change and other environmental hazards (e.g., Clements et al., 2015; Scheffer et al., 2009, 2012). Conversely, area under the curve calculations and (cross-)recurrence quantification analyses may provide new avenues to explore for detecting early warning signals of resilience losses in larger biological systems as well.

Finally, the load-response profiles presented in Chapter 6 may also be implemented for identifying adaptation to varying loading different domains. For example, using virtual reality, workers may be exposed to different loadings (e.g., distractions or workload, cf. Hoendervanger et al., 2019) in order to detect whether people adapt differently to varying loadings and under what loadings people perform optimally. This may be especially important for recruitment of highly demanding jobs, such as air traffic controllers. However, as the load-response profiles have been derived directly from research on biological systems, they may also be applied to other living systems beyond psychology.

4 | Practical Implications

Although the presented approaches are primarily theoretical and methodological, they may also bare important implications for practice. One such implication can be derived from the idiosyncrasy of dynamical processes and the according non-ergodicity (Fisher et al., 2018; Liu et al., 2006). Given that group-level findings may not generalize to the individuals, the dynamic systems approach focuses on understanding resilience on an individual-level. This means that interventions need to be tailored to the specific needs and characteristics of a given person (cf. Walton, 2014). For example, whereas for one person performing under relatively low or high stress-loadings may improve resilience, another person may benefit from various stress-loadings (Hill et al., 2020; Kiefer et al., 2018). Thus, coaches and sports psychologists working with athletes may use research findings that are generated on a group-level as a starting point for working with a client, but need to make sure that the suggested links between different factors and resilience are actually applicable to the client.

Another practical implication can be derived from the research on early warning signals of resilience losses. Although establishing and analyzing time-series may appear difficult to implement for coaches and practitioners, technological advances can be utilized

(11)

134

Chapter 7

to make the data collection non-intrusive and automate the necessary analyses. For example, athletes may make use of wearables, like smart watches, to continuously track various parameters (e.g., heart rate, Blaauw et al., 2016). This data may be automatically be uploaded to a server, which can apply algorithms to immediately analyze the data and provide tailored feedback. Such feedback may include the times of the day during which heart rate seemed to have deviated from the normal pattern or when recovery periods after specific workouts seem to be increase. These measures combined with short psychological assessments can create informative temporal profiles that may be used to detecting early warning signals of resilience losses in athletes.

Finally, the methodology we applied to create the load-response profiles in Chapter 6 can easily be applied by coaches in various sports. The two key ingredients to establishing load-response profiles is the systematic variation of loading and observing/measuring the behavioral response of interest. It should be noted, however, that the maximum load that can be applied to athletes may not be the same across all individuals. This means that the specific loading should either be tailored to the individual (e.g., the time they can complete running a certain distance) or the maximum loading that an athlete is exposed to should be controlled for. After observing the behavioral response to the varying loading factors, the resulting profiles can be inspected to detect what loadings trigger the best behavioral response and how an athlete adapts to different loadings in general (i.e., phenotypic plasticity, Kiefer et al., 2018).

5 | A Cautionary Note

Although resilience, in general, has a positive connotation given that it describes a process by which individuals adapt to stressors, it may represent a problematic response in certain contexts (Hassler & Kohler, 2014). There are two particular scenarios in which resilience as defined by the return to the previous level of functioning following a perturbation (Hill et al., 2018b) may actually hinder positive development of individuals. First, in order to teach new motor skills to a person, a coach may design exercises that change a particular behavioral pattern. For example, when trying adapt a golf swing to make it more accurate, the old motor pattern needs to be overwritten and adapted. Therefore, exercises may contain stressors aimed at making changes to the behavioral (e.g., wrist rotation) and structural (e.g., muscular) pattern of the trainee. However, a resilient movement system may be difficult to change as the individual would return to their previously adopted pattern following the intentionally induced perturbations to the movement system. Thus, in this context, resilience represents an undesirable return to the previous amount of accuracy and scores. That is to say that resilience may cause a system to become trapped in certain recurring patterns that may not be desirable.

(12)

7

Therefore, promoting the development a large array of functional motor solutions to a task in response to stressors may actually have more adaptive consequences than focusing on the longevity of a state that is re-entered following a stressor (cf. Edelman & Gally, 2001; Latash, 2012).

A related scenario in which resilience may be ill-advised is the pursuit of unobtainable goals (Wrosch et al., 2003, 2007). As discussed previously, stressors inform an organism that structural and behavioral changes need to occur in order to functionally engage with the environment (e.g., Kiefer et al., 2018, Taleb, 2012). Throughout our experiments the level of functioning was defined in regards to a specific task performance. Importantly, in these experiments, the required performance level for the tasks was obtainable for each participant. However, some goals like becoming Olympic champion, may be unobtainable for individuals in some achievement contexts. Therefore, when repeated setbacks are experienced, always returning to the goal (and its associated behavioral and cognitive pattern) may hinder the development of other skills and excelling in other domains. In such a context, it is advised to disengage from pursuing this unobtainable goal and not display resilience, but divert the resources to other achievement domains (cf. Wrosch et al., 2003, 2007). In summary, although resilience is a desirable process in many contexts, it should not be considered as the ideal default response in every situation, but tailored to each individual and their circumstances.

6 | Conclusion

Throughout this thesis, I have explored a number of new avenues to studying resilience as a dynamic process. Mapping out how individuals ‘bounce back” from stressors over time can help developing precise measure of resilience and differentiating it from related concepts, such as immunity (i.e., resistance to stress) and plasticity (i.e., growing from stress). For example, we have shown that resilience and its underlying components follow idiosyncratic change in individuals which may indicate that the individual should be the focus of analyses on resilience, rather than the group. Furthermore, we explored new analysis strategies ranging from simple surface area calculations to nonlinear time-series analysis techniques to understand how resilience changes in response to stressors. These techniques may provide input for future studies dedicated to identify resilience losses that precede negative transitions in order to develop well-timed interventions in various domains. Finally, the potential positive effect of stressors on human performance may be further explored with load-response profiles, which can be leveraged to optimize human performance across various achievement domains, such as tailoring workload in the workplace or educational institutions. Thus, as proposed at the start of the thesis, the dynamical systems approach to studying resilience in humans indeed provides

(13)

136

Chapter 7

a clear conceptualization, as well as the tools to studying this process. Because the insights derived from this approach may be translated to other fields within and outside of psychology, I hope it helps offering a unified approach to studying resilience that is currently absent in psychology.

(14)

(15)