Constructs and units of analysis

RDoC emphasises the investigation of mechanisms across multiple units of analysis, from molecules and neurocircuits to self-reports. Research centred on ‘intermediary’ psychological processes that link neurobiological mechanisms with clinical outcomes may therefore be well aligned with the NIMH’s translational aims. Given that excessive/maladaptive threat responsivity plays a central role across anxiety-related disorders, it might be particularly fruitful to anchor multimodal assessments around ecologically valid paradigms that elicit responses to acute and anticipated threats, thereby helping to elucidate the mechanisms underlying threat-related behaviour. ^{Reference Pine2}

Fear conditioning research offers an excellent example of such an approach. Fear conditioning paradigms use classical Pavlovian conditioning with aversive outcomes (e.g. electric shocks) to examine the acquisition and extinction of learned fear. Responses evoked by conditioned threats can be assessed across multiple units of analysis, including subjective ratings of fear, approach and avoidance behaviour, neural activity and physiological arousal. For example, conditioned fear has been consistently associated with central autonomic-interoceptive neural network activation. ^{Reference Fullana, Harrison, Soriano-Mas, Vervliet, Cardoner and Àvila-Parcet3} Importantly, fear conditioning research has advanced understanding and treatment of anxiety disorders by allowing scientists to model how excessive fear responses might develop and be reduced via exposure-based therapies. The translational nature of fear conditioning paradigms has also highlighted the importance of inhibitory learning, leading to new approaches for optimising exposure therapy. ^{Reference Craske, Treanor, Conway, Zbozinek and Vervliet4} Future RDoC-informed research that anchors multimodal assessments around ‘intermediary’ psychological processes may similarly hold promise for furthering the translational science of anxiety disorders.

Several methodological and quantitative approaches could improve the feasibility, utility and ecological validity of multimodal assessments. Ambulatory assessment, which combines active experience sampling through self-report with passive monitoring of behavioural and physiological outcomes using portable technologies, offers an effective method for integrating multiple units of analysis within a single study. For example, Cohen et al ^{Reference Cohen, Naslund, Lane, Bhan, Rozatkar and Mehta5} used digital phenotyping to predict mood and anxiety changes from one month of active (i.e. symptom surveys) and passive (i.e. geolocation, accelerometer and screen state) smartphone sensor data in a transdiagnostic sample, highlighting the utility of such an approach for identifying prognostic indicators. Ambulatory assessment and digital phenotyping may also help determine for whom and under which conditions treatments for anxiety-related disorders are most effective.

Additionally, quantitative approaches such as latent variable and network analyses will be crucial for examining the relationships among constructs and multimodal measures assumed to reflect a single construct involved in the aetiology and maintenance of anxiety disorders. ^{Reference Pine2} For example, Quah et al ^{Reference Quah, Jo, Geniesse, Uddin, Mumford and Barch6} applied a latent variable approach to whole-brain task-based functional magnetic resonance imaging data to test the hypothesised associations between RDoC constructs and patterns of neural activity. They compared models with theory-driven (i.e. RDoC-defined) and data-driven (i.e. empirical) factor structures to determine which model better reflected functional brain organisation. An empirically derived model, incorporating a task-general visual-functional domain and a refined cognitive systems domain, showed a better fit than the original RDoC framework across both training and validation datasets.

Importantly, these findings illustrate that rigid adherence to predefined RDoC domains assumes a definitive understanding of how the ‘joints of nature’ should be carved, potentially limiting the innovation that the framework was designed to encourage. Indeed, the RDoC framework may be most valuable if treated as an evolving framework that is routinely updated to best reflect novel findings about the nature of the domains and constructs underlying psychopathology. Such flexibility could help determine the added value of incorporating biological evidence into symptom-based syndromes in current classification systems for mental disorders. Such an iterative, data-driven approach to classifying mental disorders is also reflected in the global Precision Psychiatry movement. Further empirical tests of the assumptions embedded within RDoC regarding the coherence among measures of a single construct across different units of analysis (e.g. brain–behaviour relationships) and the validity of proposed biobehavioural dimensions will be essential for identifying reliable mechanisms that can inform the prevention and treatment of anxiety and related disorders.

Dimensionality and multidimensionality

RDoC proposes that constructs underlying symptoms of psychopathology should be studied across the full spectrum of functioning, from ‘normal’ to ‘abnormal’. ^{Reference Cuthbert1} However, case–control designs continue to dominate mechanistic research, potentially misrepresenting the dimensional nature of psychopathology while emphasising small, heterogeneous group differences of limited clinical utility. ^{Reference Pine2} Accordingly, dimensional approaches should be adopted more widely when investigating mechanisms associated with specific types of anxiety. Furthermore, given the high rates of comorbidity both within anxiety disorders and across other psychiatric disorders, multidimensional approaches, in which mechanisms are examined in relation to multiple clinical spectra (e.g. anxiety and depression), may be particularly valuable for identifying the unique, shared and interactive mechanisms underlying anxiety-related disorders. ^{Reference Kircanski, LeMoult, Ordaz and Gotlib7} Such approaches may also inform transdiagnostic interventions (e.g. the Unified Protocol), thereby advancing the NIMH’s translational mission more efficiently. Greater emphasis on recruiting samples spanning the full range of functioning, particularly across commonly co-occurring symptom dimensions (e.g. anxious apprehension and anxious arousal), is therefore needed to advance the translational aims of RDoC in anxiety-related disorders.

Experimental methods

RDoC aims to identify neurobehavioural dysfunctions that underlie psychopathology, ultimately to inform novel diagnostic, preventive and intervention targets. Therefore, in the context of anxiety disorders, establishing causal relationships between specific dysfunctions and anxiety-related outcomes is essential, making experimental methods well suited to these aims. ^{Reference Cuthbert1} Within experimental psychopathology studies and randomised controlled trial (RCT) designs, researchers are well positioned to elucidate mechanisms underlying changes in clinical outcomes through (multi)mediator analyses and to identify prescriptive indicators influencing treatment response through moderation analyses.

Chen et al’s ^{Reference Chen, Lv, Hao, Gan, Yu and Zheng8} RCTs testing the effects of attentional bias modification (ABM) and transcranial direct current stimulation (tDCS) on anxiety exemplify the advantages of such an approach. By randomising participants to receive one, both or neither intervention, incorporating sham control conditions and assessing constructs at multiple timepoints, these studies not only demonstrated the efficacy of ABM and tDCS in reducing anxiety but also identified moderators (i.e. baseline patterns of attention) and mediators (i.e. enhanced or attenuated engagement with threats) of treatment response. By enabling researchers to determine both how and for whom interventions are effective, experimental approaches can advance mechanistic models of anxiety and support progress towards personalised medicine.

Environmental and developmental factors

Although less commonly the focus of RDoC-driven research, examining interactions between biobehavioural dysfunctions, environmental factors and developmental trajectories is emphasised in the RDoC initiative. ^{Reference Cuthbert1} Indeed, limited findings from early genetic association and genome-wide association studies in anxiety disorders underscored the limitations of focusing solely on the genetic architecture of anxiety-related psychopathology and prompted a shift towards examining gene-by-environment (GxE) interactions. ^{Reference Sharma, Powers, Bradley and Ressler9} GxE research has shown that the effects of gene variants (e.g. the 5-HTTLPR polymorphism) on anxiety often depend on environmental exposures such as trauma, and that the developmental timing of these exposures also influences anxiety risk. For example, evidence suggests that early- rather than later-life stress potentiates hypothalamic–pituitary–adrenal axis activation, increasing risk of stress-related disorders. ^{Reference Sharma, Powers, Bradley and Ressler9}

Future RDoC-driven research may benefit from the lessons of the GxE movement by explicitly incorporating environmental and developmental factors into research questions. More broadly, the RDoC framework organises research into six broad domains of functioning, yet these domains continue to be investigated largely in isolation from one another and from their environmental and developmental context. Progress may instead depend on understanding how these domains interact. Adopting a multisystem, interaction-focused approach could generate research that provides a more nuanced and clinically useful understanding of the development, prevention and treatment of anxiety disorders.

In conclusion, RDoC has left a lasting mark on the science of anxiety disorders. To maximise its clinical impact, future RDoC-informed research should incorporate and empirically validate multimodal assessments, recruit samples that represent the full multidimensional spectra of functioning, capitalise on the strengths of experimental methods and explicitly examine interactions with environmental and developmental factors.