Focusing Inward: A Timely Yet Daunting Challenge for Clinical Psychological Science

Amir and Bernstein (this issue) propose a dynamical model of internally-directed cognition aimed at explaining the complex interactions between current goals, negative affect, and attentional selection in working memory. They connect the literature on internal attention, working memory, affect, rumination, and mind wandering to propose a formal mathematical model of internally-directed cognition. In this paper, they do not just provide a window on how people become stuck in loops of negative thinking, but they also provide a nice example of how clinical psychological science can move toward more formal theoretical models. In taking such an exciting step, we believe that this work also encounters some of the challenges faced by formal models of maladaptive cognition. Below we discuss some of these issues, not in order to criticize the current work, but to open a discussion, which we feel is paramount as the field of clinical psychology moves in the direction of developing formal theoretical models. In brief, the three main issues are: (1) the proposed model does not build on the existing cognitive models; (2) the model increases rather than decreases the complexity of the phenomenon; (3) there are no standard/alternative frameworks to compare the A2T model to, and it is not clear which kind of data or experiments could corroborate or falsify the model. New models should build on the existing formal models of cognitive processes. The reproducibility crisis in psychology (Open Science Collaboration, 2015; Simmons, Nelson, & Simonsohn, 2011) has led to significant changes in the way we conduct research, which include preregistration and better statistical methodology (Benjamin et al., 2018; Nosek, Ebersole, DeHaven, & Mellor, 2018). In the slipstream of this movement, a reinvigorated discussion has been opened on the role and current status of theory in psychology (e.g., Fried, 2020; Grahek, Schaller, & Tackett, 2021; Haslbeck, Ryan, Robinaugh, Waldorp, & Borsboom, 2021). Clearly, clinical psychological science has no shortage of rather vague, descriptive theories that are difficult to test and disprove. Many areas of psychology are moving in the direction of developing stronger theories, which could guide experimentation and increase the overall rigor of psychological science. In this context, clinical psychology is faced with the task of creating formal mathematical models of important phenomena, including the ones which the A2T model tackles. This effort, often referred to as computational psychiatry (Huys, Maia, & Frank, 2016; Montague, Dolan, Friston, & Dayan, 2012), is showing a lot of promise. The crucial part of this effort is to develop computational models that are relevant for understanding psychopathology, but also have direct links with the existing formal models from cognitive science. In this way, clinical psychology can build on the existing models, and extend them in order to better understand psychopathology. Such efforts are already present in many domains including decision-making (Huys, Daw, & Dayan, 2015), learning (Brown et al., 2021), working memory (Collins, Albrecht, Waltz, Gold, & Frank, 2017), and cognitive control (Dillon et al., 2015; Grahek, Shenhav, Musslick, Krebs, & Koster, 2019). This is where the A2T model will require the most work in its further development. While the authors mention some of the models that map onto the components of their model (Hazy et al., 2007), all of these components are modeled at a very high level right now. While this is a necessary and a great first step, the model would benefit from incorporating the architecture of existing models of attentional control, working memory, and affect. Unless this is done, we are missing the opportunity to do cumulative science and develop an integrated understanding of clinical processes occurring within a normative framework. In past years, this has been raised in the area of cognitive control in depression, where theory in clinical psychological science developed without attention to the basic cognitive-experimental science on cognitive control (Grahek, Everaert, Krebs, & Koster, 2018).