We greatly enjoyed reading Kalisch and colleagues' proposed conceptual framework for studying the neurobiology of resilience. As quantitatively oriented researchers, we strongly concur with their argument that quantitative measures are likely a better reflection of the underlying neurobiology of resilience and psychopathology compared with traditional psychiatric diagnostic categories. We also appreciated their emphasis on studying dysfunctions over disorders; however, there were also several aspects of the model that we believe would benefit from further refinement. In particular, we had concerns about the authors' definition of resilience as “an empirically observable phenomenon, namely that someone does not develop lasting mental health problems although he or she is subject to adversity” (sect. 2, para. 2).
Although this definition is laudable because of its simplicity, it is also an extremely narrow view of resilience. In contrast, Windle defines resilience as:
The process of effectively negotiating, adapting to, or managing significant sources of stress or trauma. Assets and resources within the individual, their life and environment facilitate this capacity for adaptation and “bouncing back” in the face of adversity. Across the life course, the experience of resilience will vary. (Windle Reference Windle2011)
Whereas Windle (Reference Windle2011) explicitly excludes mental health from her definition of resilience, the authors' definition is entirely dependent on the absence of mental health.
We believe that this overreliance on mental health to define resilience is problematic. First, it is unclear to us how studying a construct defined entirely by another construct can yield new findings. For example, the authors propose that a general self-report measure of psychopathology would be an ideal measure of resilience within their framework. If a measure of psychopathology is used to define resilience, however, are we not still just studying psychopathology? Any association between a given predictor and “resilience” defined in this manner will have an association of identical magnitude in the opposite direction with “psychopathology” using the very same measure. Hence, it is not clear how quantitatively defined resilience differs from quantitatively defined psychopathology in the proposed model.
Another notable difference between the two definitions is that Windle's (Reference Windle2011) is focused on adaptation to stress and also considers assets and resources that might be available to the individual. In contrast, Kalisch et al.'s definition does not seem to consider the very significant role that assets and resources play in resilience, and it fails to recognize the well-replicated finding that people with more resources are better able to adapt to stressful conditions (e.g., Hobfoll Reference Hobfoll2002). The latter point is critical for researchers interested in studying the neurobiology of resilience, as it is very likely that resources (e.g., social support, level of education, finances) and many other environmental factors moderate the influence of neurobiological factors (e.g., genetic and epigenetic factors) on resilience to stressful life events (e.g., Koenen et al. Reference Koenen, Nugent and Amstadter2008; Moffitt et al. Reference Moffitt, Caspi and Rutter2005).
We also were surprised by the authors' argument that “resilience-conducive traits” (i.e., traditional measures of resilience and hardiness, such as the Connor-Davidson Resilience Scale (CD-RISC); Connor & Davidson Reference Connor and Davidson2003) should not be used as measures of resilience within their model. We disagree with the authors on this point. We think that the inclusion of traditional measures of resilience is an ideal way of broadening the definition of resilience beyond the rather simplistic “absence of psychopathology” approach that the authors seem to advocate. An additional advantage of including these types of measures in studies of resilience is that it enables researchers to consider whether the manner in which self-reported resilience is defined (e.g., “absence of psychopathology” vs. “ability to bounce back”) influences their findings.
A third advantage of broadening the resilience construct to include traditional measures of resilience is that it would enable researchers to construct latent factors that include both “trait” and “state” resilience characteristics. To illustrate this point, we conducted a confirmatory factor analysis (CFA) on the CD-RISC, the Davidson Trauma Scale (Davidson et al. Reference Davidson, Book, Colket, Tupler, Roth, David, Hertzberg, Mellman, Beckham, Smith, Davison, Katz and Feldman1997), and the Symptom Checklist-90 Global Severity Index (Derogatis et al. Reference Derogatis, Lipman and Covi1973) in a sample of 2,339 U.S. veterans. The CFA model exhibited exceptionally good fit to the data, χ
2(3) = 3875.020, RMSEA = 0.00, CFI = 1.00, and factor loadings were high across measures ≥0.65, providing strong support for this type of latent variable modeling approach to quantitatively define resilience. We believe that this approach could be further extended to include self-report measures of functional impairment (e.g., Üstün Reference Üstün2010) and quality of life (e.g., Burckhardt & Anderson Reference Burckhardt and Anderson2003), as well as measures derived from behavioral, physiological, and neuroimaging paradigms.
A final issue concerns the measurement of stressful life events. The authors provide a nice summary of the many difficulties associated with the assessment of both traumatic stress and daily hassles. We agree with many of their points, as we have spent a great amount of time trying to quantify the impact of one particular type of stressful experience (i.e., combat exposure) on returning veterans' mental health (e.g., Kimbrel et al. Reference Kimbrel, Evans, Patel, Wilson, Meyer, Gulliver and Morissette2014). Although we recognize the difficulties in quantifying trauma exposure that researchers face, we also believe that there is currently a great need to quantify stress exposure across a wide range of different populations if we truly wish to improve our understanding of genetic and epigenetic influences on resilience. For example, efforts are currently under way to begin conducting meta- and mega-analyses of genome-wide association study (GWAS) data through the posttraumatic stress disorder (PTSD) working group of the Psychiatric Genomics Consortium (Koenen et al. Reference Koenen, Duncan, Liberzon and Ressler2013); however, a key challenge currently facing researchers interested in combining GWAS datasets of PTSD is how to characterize trauma load, trauma type, and the timing of the trauma across different trauma populations (Koenen et al. Reference Koenen, Duncan, Liberzon and Ressler2013). Both traumatic stress and stressful life events are robust predictors of a wide range of psychopathology. Given that, it is clear that the development of universal measures of stressful events that can accurately quantify traumatic load for a wide range of stressful experiences across different populations is a crucial next step in advancing our understanding of the neurobiology of resilience.
You have
Access
We greatly enjoyed reading Kalisch and colleagues' proposed conceptual framework for studying the neurobiology of resilience. As quantitatively oriented researchers, we strongly concur with their argument that quantitative measures are likely a better reflection of the underlying neurobiology of resilience and psychopathology compared with traditional psychiatric diagnostic categories. We also appreciated their emphasis on studying dysfunctions over disorders; however, there were also several aspects of the model that we believe would benefit from further refinement. In particular, we had concerns about the authors' definition of resilience as “an empirically observable phenomenon, namely that someone does not develop lasting mental health problems although he or she is subject to adversity” (sect. 2, para. 2).
Although this definition is laudable because of its simplicity, it is also an extremely narrow view of resilience. In contrast, Windle defines resilience as:
The process of effectively negotiating, adapting to, or managing significant sources of stress or trauma. Assets and resources within the individual, their life and environment facilitate this capacity for adaptation and “bouncing back” in the face of adversity. Across the life course, the experience of resilience will vary. (Windle Reference Windle2011)
Whereas Windle (Reference Windle2011) explicitly excludes mental health from her definition of resilience, the authors' definition is entirely dependent on the absence of mental health.
We believe that this overreliance on mental health to define resilience is problematic. First, it is unclear to us how studying a construct defined entirely by another construct can yield new findings. For example, the authors propose that a general self-report measure of psychopathology would be an ideal measure of resilience within their framework. If a measure of psychopathology is used to define resilience, however, are we not still just studying psychopathology? Any association between a given predictor and “resilience” defined in this manner will have an association of identical magnitude in the opposite direction with “psychopathology” using the very same measure. Hence, it is not clear how quantitatively defined resilience differs from quantitatively defined psychopathology in the proposed model.
Another notable difference between the two definitions is that Windle's (Reference Windle2011) is focused on adaptation to stress and also considers assets and resources that might be available to the individual. In contrast, Kalisch et al.'s definition does not seem to consider the very significant role that assets and resources play in resilience, and it fails to recognize the well-replicated finding that people with more resources are better able to adapt to stressful conditions (e.g., Hobfoll Reference Hobfoll2002). The latter point is critical for researchers interested in studying the neurobiology of resilience, as it is very likely that resources (e.g., social support, level of education, finances) and many other environmental factors moderate the influence of neurobiological factors (e.g., genetic and epigenetic factors) on resilience to stressful life events (e.g., Koenen et al. Reference Koenen, Nugent and Amstadter2008; Moffitt et al. Reference Moffitt, Caspi and Rutter2005).
We also were surprised by the authors' argument that “resilience-conducive traits” (i.e., traditional measures of resilience and hardiness, such as the Connor-Davidson Resilience Scale (CD-RISC); Connor & Davidson Reference Connor and Davidson2003) should not be used as measures of resilience within their model. We disagree with the authors on this point. We think that the inclusion of traditional measures of resilience is an ideal way of broadening the definition of resilience beyond the rather simplistic “absence of psychopathology” approach that the authors seem to advocate. An additional advantage of including these types of measures in studies of resilience is that it enables researchers to consider whether the manner in which self-reported resilience is defined (e.g., “absence of psychopathology” vs. “ability to bounce back”) influences their findings.
A third advantage of broadening the resilience construct to include traditional measures of resilience is that it would enable researchers to construct latent factors that include both “trait” and “state” resilience characteristics. To illustrate this point, we conducted a confirmatory factor analysis (CFA) on the CD-RISC, the Davidson Trauma Scale (Davidson et al. Reference Davidson, Book, Colket, Tupler, Roth, David, Hertzberg, Mellman, Beckham, Smith, Davison, Katz and Feldman1997), and the Symptom Checklist-90 Global Severity Index (Derogatis et al. Reference Derogatis, Lipman and Covi1973) in a sample of 2,339 U.S. veterans. The CFA model exhibited exceptionally good fit to the data, χ 2(3) = 3875.020, RMSEA = 0.00, CFI = 1.00, and factor loadings were high across measures ≥0.65, providing strong support for this type of latent variable modeling approach to quantitatively define resilience. We believe that this approach could be further extended to include self-report measures of functional impairment (e.g., Üstün Reference Üstün2010) and quality of life (e.g., Burckhardt & Anderson Reference Burckhardt and Anderson2003), as well as measures derived from behavioral, physiological, and neuroimaging paradigms.
A final issue concerns the measurement of stressful life events. The authors provide a nice summary of the many difficulties associated with the assessment of both traumatic stress and daily hassles. We agree with many of their points, as we have spent a great amount of time trying to quantify the impact of one particular type of stressful experience (i.e., combat exposure) on returning veterans' mental health (e.g., Kimbrel et al. Reference Kimbrel, Evans, Patel, Wilson, Meyer, Gulliver and Morissette2014). Although we recognize the difficulties in quantifying trauma exposure that researchers face, we also believe that there is currently a great need to quantify stress exposure across a wide range of different populations if we truly wish to improve our understanding of genetic and epigenetic influences on resilience. For example, efforts are currently under way to begin conducting meta- and mega-analyses of genome-wide association study (GWAS) data through the posttraumatic stress disorder (PTSD) working group of the Psychiatric Genomics Consortium (Koenen et al. Reference Koenen, Duncan, Liberzon and Ressler2013); however, a key challenge currently facing researchers interested in combining GWAS datasets of PTSD is how to characterize trauma load, trauma type, and the timing of the trauma across different trauma populations (Koenen et al. Reference Koenen, Duncan, Liberzon and Ressler2013). Both traumatic stress and stressful life events are robust predictors of a wide range of psychopathology. Given that, it is clear that the development of universal measures of stressful events that can accurately quantify traumatic load for a wide range of stressful experiences across different populations is a crucial next step in advancing our understanding of the neurobiology of resilience.
ACKNOWLEDGMENTS
This research was supported by a Career Development Award (1IK2 CX000525) to Dr. Kimbrel and a Research Career Scientist Award to Dr. Beckham from the Clinical Science Research and Development Service of the VA Office of Research and Development. This work also was supported by resources from the Durham VA Medical Center; the VA Mid-Atlantic Mental Illness Research, Education, and Clinical Center; and Duke University Medical Center. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the United States government.