1b
Shou-An Ariel Chang and Arielle Baskin-Sommers
Personality disorders (PDs) are among the costliest psychiatric conditions (Lenzenweger, Lane, Loranger, & Kessler, 2007; Soeteman, Roijen, Verheul, & Busschbach, 2008). They are chronic and pervasive conditions that have a severe personal, social, and financial impact. As Chan and colleagues (this volume) outline in “Neuroimaging in Personality Disorders,” the use of neuroimaging seeks to pinpoint mechanisms underlying several forms of PDs (e.g., schizotypal personality disorder [SPD], borderline personality disorder [BPD], and antisocial personality disorder [ASPD]). Neuroimaging methods, such as magnetic resonance imaging (MRI) or diffusor tensor imaging, identify structural and functional abnormalities associated with these PDs. Some neural abnormalities appear common across disorders, and others appear unique to each disorder.
Substantial progress has been made in characterizing neural abnormalities related to PDs through the use of neuroimaging methods. The chapter from Chan et al. (this volume) on “Neuroimaging in Personality Disorders” nicely lays out the current empirical foundation of neuroimaging in PDs. While not an explicit focus in their chapter, their review does unearth two issues plaguing neuroimaging research on PDs. First, in the studies covered in their review, most rely on a region of interest (ROI) approach that is limited in several important ways. Second, the tasks used in many of the studies covered in their review are too broad to specify a precise mechanism. Together, these two issues limit progress in understanding the variety of potential neural mechanisms that underlie PDs and the ability to link those mechanisms to pathological behavior. To unlock the full potential of neuroimaging methods to increase our clinical understanding of PDs, researchers must implement more advanced methods and use precise tasks that contextualize the behavior of those affected by PDs. Thus, the goal of this commentary is to provide a deeper dive into each of these limitations touched on less explicitly by Chan et al. (this volume), building towards suggestions for how to advance the neuroscience of PDs.
The continued reliance on a region of interest (ROI) approach (i.e., extracting signals from pre-specified regions; Poldrack, 2007) is a significant issue that limits our identification of broader dysfunctions that may cut across regions or the interplay between regions that yield complex behavioral dysfunctions. There are advantages to using an ROI approach, such as statistical control, theory testing, and functional exploration (Poldrack, 2007); however, a key criticism of the ROI approach is that many tasks evoke activation beyond ROIs, making it difficult to interpret the functional specificity of any particular ROI (Friston, Rotshtein, Geng, Sterzer, & Henson, 2006).
The limitations of the ROI approach for research on PDs is highlighted in a recent study cited in Chan et al. (this volume) by Holtmann and colleagues (2013). In this study, BPD patients and healthy controls (HC) completed a modified Eriksen Flanker during which task-irrelevant neutral and fearful faces were presented. Holtmann et al. (2013) identified several a priori ROIs, including the amygdala and dorsal anterior cingulate cortex, to compare between BPD patients and HC. Examination of a key contrast, fear versus neutral faces, revealed that the only ROI differentiating BPD patients and HC was the dorsal anterior cingulate.
First, in terms of a priori ROIs, one could question the inconsistency between this study and several other studies noting amygdala activation abnormalities in BPD patients versus HC (van Zutphen, Siep, Jacob, Goebel, & Arntz, 2015). However, the validity of ROIs is dependent on their context-sensitivity, making task demands an important factor to consider when identifying ROIs. Studies showing amygdala activation differences between BPD patients and HC often are in the context of direct viewing of emotional content, and not when this content is used as an irrelevant distractor, as was done in the Holtmann et al. (2013) study. Therefore, identification of ROIs must be based on context in order to appropriately constrain interpretation of the presence or absence of any effect.
Second, several regions, including the precuneus, distinguished BPD patients versus HC (see Holtmann et al., 2013, table 8 for full list). The precuneus has been implicated in self-related mental representations (Cavanna & Trimble, 2006). Combined with the dorsal anterior cingulate finding, it is possible that compared to HC, BPD patients recruit more neural resources to engage in self-referential evaluative processing of emotion content, regardless of task relevance, and that dorsal anterior cingulate hyperactivation is not simply a reflection of enhanced “executive functions” used to “compensate” for the processing of emotion content. Consideration of patterns of activation beyond ROIs is essential in order to accurately capture and interpret the complexity of neural abnormalities as they relate to PDs. To be clear, an ROI approach is not inherently problematic. However, clear interpretation of the patterns of ROI activation and their meaning is very difficult, especially when tasks are used that fail to discriminate mechanisms.
Beyond issues specific to ROI, neuroimaging studies in PDs also are hindered by the fact that many rely on tasks that measure functions broadly, and lack the necessary precision needed to identify a specific mechanism. Often tasks such as viewing emotional images or Go/No-Go are used to assess emotional and executive function, respectively, in research on PDs. These tasks engage broad processes, but tend to lack the specificity in design (e.g., condition manipulations) to identify the exact mechanisms underlying dysfunction. For example, face processing studies involve recording brain activity while participants evaluate faces on a particular dimension (e.g., emotional expression, gender, trustworthiness). Many mechanisms underlie face processing, including attending to the image, integrating visual information, and potentially selecting or inhibiting a response to the face (Jehna et al., 2011). Therefore, dysfunction in face processing could be due to narrowed attention to a specific part of the face (e.g., only looking at the mouth), an inability to process the face as a whole (e.g., difficulty integrating parts of the face), and/or overarousal, amongst other possibilities.
In fact, SPD, BPD, and ASPD all are associated with difficulty in face processing tasks, particularly when it comes to evaluating emotional expressions (SPD; Dickey et al., 2011; ASPD; Dolan & Fullam, 2006; BPD; Domes et al., 2008; Schönenberg & Jusyte, 2014). However, the reason(s) why individuals with these disorders show difficulty on this type of task is likely different. By using broad face processing tasks, we are left in the dark about the cause of the difficulties in these tasks and are falling into the trap of conflating behavior (e.g., face processing dysfunction) with mechanism (i.e., the cause of the dysfunction). As an example of an approach that attempts to narrow down the mechanism underlying face processing dysfunction, an emotion recognition task could include a condition with uninstructed viewing of faces and a condition with an explicit attentional cueing to the eyes to see if the resulting difference in emotion recognition accuracy or brain activity is due to atypical attention rather than emotion processing more broadly. Disambiguating mechanism from behavior is not only important for understanding mechanisms within a specific personality disorder, but also for distinguishing among PDs. Phenotypically, many PDs share overlapping symptomology (e.g., interpersonal difficulties); and, in order to understand what is driving this symptomology within and across PDs, more precise tasks are needed.
Related to the issues of relying on ROIs and using broad tasks in the study of PDs, several methodological advancements are needed to further the state of research on PDs. First, researchers should combine neuroimaging with other measures, such as genetic data or behavioral measures, in order to create a profile or “fingerprint” of different PDs (Brazil, van Dongen, Maes, Mars, & Baskin-Sommers, 2018). For example, Chan et al. (this volume) outline a wealth of research evidencing abnormalities in structural, functional, and resting state brain activity in PDs. These measures could be combined with genetic and/or behavioral (e.g., task performance, neuropsychological assessments, real-world indicators) data in order to identify latent profiles classifying individuals based on common and unique associations among these variables (Uludağ & Roebroeck, 2014; Wolfers, Buitelaar, Beckmann, Franke, & Marquand, 2015; see Chan et al.’s BPD imaging-genetics example). This “fingerprinting” approach even could leverage the ROI approach by combining these region activations with other measures to better contextualize our understanding of the contribution of this activation. Building a profile of biological and behavioral dimensions and their interrelationships could prove to be especially fruitful in distinguishing PDs, as it is unlikely that one specific mechanism underlies each distinct personality disorder.
Second, researchers could implement more advanced methods, such as computational modeling and newer imaging methods to further clarify the mechanisms driving symptomology in PDs. Computational psychiatry aims to break down cognitive functions, such as attention or action selection into smaller cognitive operations. This approach necessitates the implementation of more precise tasks for the study of neural function in PDs. For instance, Behrens, Hunt, Woolrich, and Rushworth (2008) leverage models of reinforcement learning to show that individuals learn about reward probability from a social informant in the same manner as when learning from personal experience. When combined with functional MRI data, they show that distinct anatomical structures are involved in encoding learning from a social informant versus learning from personal experience. Of note, SPD, BPD, and ASPD all are associated with atypical reinforcement learning (BPD; Bornovalova, Lejuez, Daughters, Zachary Rosenthal, & Lynch, 2005; ASPD; Glenn & Yang, 2012; Gregory et al., 2015; Schuermann, Kathmann, Stiglmayr, Renneberg, & Endrass, 2011; SPD; Waltz, Frank, Robinson, & Gold, 2007). If applied to individuals with PDs, this method would allow for greater precision in relating neural abnormalities to behavior by delineating whether a deficit in reinforcement learning is broad or whether the deficit in reinforcement learning is specific to a particular domain (e.g., social, reward, punishment).
Another example of more advanced methods that could help refine our understanding of neural function in PDs is the use of graph theory. Broadly, graph theory moves beyond traditional connectivity analyses, which measure the association between two regions, to estimate how well and in what manner neural regions communicate with one another (Bullmore & Sporns, 2009). As noted by Chan et al. (this volume), all PDs show abnormal structural and functional connectivity. For example, in both SPD and BPD there are alternations in default mode network connectivity and in ASPD disruptions in fronto-parietal control network connectivity is evident. Graph theory could be advantageous for studying PDs, as it moves past quantifying the ability of regions to communicate and specifies the quality of that communication by estimating the amount of time or energy required to transfer information from one part of the network to any other part of the network. The application of graph theory in future research might be useful for clarifying in what ways information processing is disrupted in PDs.
Summary
The chapter by Chan et al. (this volume) provides a thorough review of the current state of neuroimaging research in PDs, and a base from which researchers can build to design precise tasks isolating underlying mechanisms of dysfunction, combine methods to create “fingerprints” of PDs, and apply more advanced neuroimaging methods and analyses to identify distinct and overlapping mechanisms underlying PDs symptomology. It is essential that researchers strive to link neural mechanisms to behavior in order to inform precision in assessment and the development of targeted treatments.
References
Behrens, T. E., Hunt, L. T., Woolrich, M. W., & Rushworth, M. F. (2008). Associative learning of social value. Nature, 456(7219), 245–249.
Bornovalova, M. A., Lejuez, C. W., Daughters, S. B., Zachary Rosenthal, M., & Lynch, T. R. (2005). Impulsivity as a common process across borderline personality and substance use disorders. Clinical Psychology Review, 25(6), 790–812.
Brazil, I. A., van Dongen, J. D. M., Maes, J. H. R., Mars, R. B., & Baskin-Sommers, A. R. (2018). Classification and treatment of antisocial individuals: From behavior to biocognition. Neuroscience & Biobehavioral Reviews, 91, 259–277.
Bullmore, E., & Sporns, O. (2009). Complex brain networks: Graph theoretical analysis of structural and functional systems. Nature Reviews Neuroscience, 10(3), 186–198.
Cavanna, A. E., & Trimble, M. R. (2006). The precuneus: A review of its functional anatomy and behavioural correlates. Brain, 129(3), 564–583.
Dickey, C. C., Panych, L. P., Voglmaier, M. M., Niznikiewicz, M. A., Terry, D. P., Murphy, C., … McCarley, R. W. (2011). Facial emotion recognition and facial affect display in schizotypal personality disorder. Schizophrenia Research, 131(1), 242–249.
Dolan, M., & Fullam, R. (2006). Face affect recognition deficits in personality-disordered offenders: Association with psychopathy. Psychological Medicine, 36(11), 1563–1569.
Domes, G., Czieschnek, D., Weidler, F., Berger, C., Fast, K., & Herpertz, S. C. (2008). Recognition of facial affect in borderline personality disorder. Journal of Personality Disorders, 22(2), 135–147.
Friston, K. J., Rotshtein, P., Geng, J. J., Sterzer, P., & Henson, R. N. (2006). A critique of functional localisers. NeuroImage, 30(4), 1077–1087.
Glenn, A. L., & Yang, Y. (2012). The potential role of the striatum in antisocial behavior and psychopathy. Biological Psychiatry, 72(10), 817–822.
Gregory, S., Blair, R. J., ffytche, D., Simmons, A., Kumari, V., Hodgins, S., & Blackwood, N. (2015). Punishment and psychopathy: A case-control functional MRI investigation of reinforcement learning in violent antisocial personality disordered men. The Lancet Psychiatry, 2(2), 153–160.
Holtmann, J., Herbort, M. C., Wustenberg, T., Soch, J., Richter, S., Walter, H., … Schott, B. H. (2013). Trait anxiety modulates fronto-limbic processing of emotional interference in borderline personality disorder. Frontiers in Human Neuroscience, 7, 54.
Jehna, M., Neuper, C., Ischebeck, A., Loitfelder, M., Ropele, S., Langkammer, C., … Enzinger, C. (2011). The functional correlates of face perception and recognition of emotional facial expressions as evidenced by fMRI. Brain Research, 1393, 73–83.
Lenzenweger, M. F., Lane, M. C., Loranger, A. W., & Kessler, R. C. (2007). DSM-IV personality disorders in the National Comorbidity Survey Replication. Biological Psychiatry, 62(6), 553–564.
Poldrack, R. A. (2007). Region of interest analysis for fMRI. Social Cognitive and Affective Neuroscience, 2(1), 67–70.
Schönenberg, M., & Jusyte, A. (2014). Investigation of the hostile attribution bias toward ambiguous facial cues in antisocial violent offenders. European Archives of Psychiatry and Clinical Neuroscience, 264(1), 61–69.
Schuermann, B., Kathmann, N., Stiglmayr, C., Renneberg, B., & Endrass, T. (2011). Impaired decision making and feedback evaluation in borderline personality disorder. Psychological Medicine, 41(9), 1917–1927.
Soeteman, D. I., Roijen, L. H.-v., Verheul, R., & Busschbach, J. J. (2008). The economic burden of personality disorders in mental health care. Journal of Clinical Psychiatry, 69(2), 259–265.
Uludağ, K., & Roebroeck, A. (2014). General overview on the merits of multimodal neuroimaging data fusion. NeuroImage, 102, 3–10.
van Zutphen, L., Siep, N., Jacob, G. A., Goebel, R., & Arntz, A. (2015). Emotional sensitivity, emotion regulation and impulsivity in borderline personality disorder: A critical review of fMRI studies. Neuroscience & Biobehavioral Reviews, 51, 64–76.
Waltz, J. A., Frank, M. J., Robinson, B. M., & Gold, J. M. (2007). Selective reinforcement learning deficits in schizophrenia support predictions from computational models of striatal-cortical dysfunction. Biological Psychiatry, 62(7), 756–764.
Wolfers, T., Buitelaar, J. K., Beckmann, C. F., Franke, B., & Marquand, A. F. (2015). From estimating activation locality to predicting disorder: A review of pattern recognition for neuroimaging-based psychiatric diagnostics. Neuroscience & Biobehavioral Reviews, 57, 328–349.