Minds, Models and Mechanisms:
Current Trends in Philosophy of Psychiatry
Three Epistemic Roles of Computational Modelling in Psychiatric Therapy
Computational modelling plays various epistemic roles in contemporary psychiatric therapies for alcohol use disorder. Focusing on alcohol-avoidance training, this paper clarifies how computational modelling increases the precision of tests of existing therapies, helps researchers to identify clinically relevant sub-groups of patients and to discover some of the causal factors of psychiatric disorders. To the extent computational modelling successfully plays these roles, we can ask whether and in what sense computational modelling can offer adequate explanations of psychiatric phenomena that are autonomous from neurobiological mechanisms.
An Enactive Approach to Chronic Pain: Addressing Conceptual, Methodological, and Therapeutic Challenges
Pain has made its way into various debates of philosophy of mind as the prime example of an experience with a particular feeling of what it is like to undergo such phenomenon. Surprisingly, most philosophers ignore the phenomenal-existential perspective of those suffering (Svenaeus, 2015) and the implications of their approaches to the management of and interaction with patients with chronic pain. In clinical contexts, multidimensional and therapeutically orientated models, such as biopsychosocial models, have become more popular, but still leave much to be desired (Stilwell & Harman, 2019). Too often such models are used in reductionist, linear, and fragmented ways. In particular, they face three challenges. (1) Conceptual challenge: how can we understand a subject in pain as being at the same time a biological organism, a member of social-cultural communities, and an individual who experiences and meaningfully interprets themselves and their environment? (2) Methodological challenge: how can we integrate insights from multiple empirical domains which employ different methods (e.g., qualitative interviews, brain imaging techniques, behavioral studies) while targeting the same phenomenon? (3) Therapeutic challenge: how can we best treat pain when the biological, social, and psychological aspects involved in its generation and maintenance are interwoven in dynamic interaction with each other across different time-scales? We aim to develop an enactive approach to pain. That is, we think about pain and the process of chronification in terms of alterations in the dynamic, interactive, and embodied relation between subjects and their environment (e.g. Varela, Thompson, & Rosch, 1991). An enactive framework seems particularly suited to approach how subjects experience themselves and their environment from the perspective of bodily beings that relate and attune to such environment based on their skills, concerns, and interests. Our approach is in line with a promising trend in the enactive discourse on psychopathologies that has gained popularity in recent years (de Haan et al., 2020; Glackin et al., 2020; Krueger & Colombetti, 2018). So far, the literature is sparse on the question of how enactive approaches can contribute to better understanding pain and inform therapeutic approaches. We aim to show that an enactivist approach to pain is empirically motivated and of practical use. Not only does it do justice to the decisive role action plays in relation to pain (Klein, 2015), but it can also rise to the challenges previously presented.
On modeling mental disorders. Epistemic Richness and the Case of PTSD
In models of mental disorders, there is a tension between the desideratum of providing generalizable inferences for the disorder and the constraint of explaining specific features of individual patients. Moreover, models of mental disorder have to integrate many factors that contribute to the emergence, maintenance or relapse of mental disorders which span many levels from biological vulnerabilities, to personal experiences, up to sociocultural background and, often neglected, subjective sense making. I will call models that are able to achieve a proper balance of both, generalizability and specificity, and which, in addition, are able to integrate a relevant amount of the contributing factors, epistemically rich (strong) models. Here I will try to develop the rough idea of an epistemically rich model further by examining models of post-traumatic stress disorder (PTSD). PTSD is of specific interest for for at least two reasons. First, in contrast to other disorders, it contains a causal event – the trauma – as a necessary diagnostic criterion so that models of PTSD are by necessity linked to (at least one) cause of the disorder. Second, PTSD is well suited as a test case for the concept of epistemically rich models, as the factors contributing to PTSD are highly idiosyncratic. In this presentation, I will discuss some current models of PTSD to develop the notion of epistemic richness further. I will evaluate their limits and potentials and will particularly highlight the problem of integration, a feature which is often claimed in modeling frameworks but rarely achieved.
(Max Planck Institute for Biological Cybernetics)
The nascent field of computational psychiatry includes the application of computational and statistical methods to understand dysfunction in psychiatric disease through the medium of normal information processing and cognition. I will discuss our attempts to use Bayesian decision theory to provide a framework for understanding disorders – coarsely, solving the wrong problem, solving the right problem incorrectly, or solving the right problem correctly, but in the wrong environment. I will illustrate aspects of these flaws using examples drawn from anhedonia, rumination and learned helplessness.
(Humboldt-Universität zu Berlin)
Mental Disorder – An Ability-based Approach
On the standard view, “mental disorder” is defined, roughly, as a harmful biological dysfunction. This view, however, faces a couple of problems. I will argue that the presence of a (harmful) biological dysfunction is neither necessary nor sufficient for having a mental disorder. I will then propose an alternative definition. According to the view I propose, having a mental disorder is a matter of having a certain harmful inability. In a nutshell, the view is the following: an individual has a mental disorder if and only if she is unable to respond adequately to some of her reasons in some of her thinking, feeling, or acting in view of her mental constitution, and is harmed by the condition that results from that inability. In my talk, I will spell out the concept of ability in more detail and show in which respects the conceptual shift from “dysfunction” to “inability” proves fruitful. In sum, the view I propose offers a conceptual framework that is fruitful, because it allows us to combine personal-level with subpersonal-level talk without making commitments neither on the metaphysical level nor on questions of explanation.
Network Neuroscience in Lesional Brain Disease
Cognitive deterioration in lesional brain disease, such as cerebrovascular accidents, multiple sclerosis and glioma, weighs heavily on patients, their caregivers and society, particularly since curative treatment is unavailable. Variation in cognitive decline is large: some patients suffer from progressive deterioration, while others do not, despite comparable disease parameters. A traditional localizationist view of brain functioning has not yielded accurate understanding or treatment targets for such cognitive decline. Using concepts from graph theory, cognition is increasingly seen as a combination of segregation and integration occurring in the brain network. This network can be measured anatomically and functionally using neuroimaging and neurophysiological modalities. Brain regions form the network nodes, while their number of interconnections and/or extent of functional interdependency define connections. Acute and long-term cognitive decline in lesional brain disease have been linked to brain network alterations, but it is largely unclear whether and how networks can be used for prediction of cognitive decline and ultimately treatment thereof. The most insights into these questions will be discussed from unimodal and multilayer network perspectives.
Correlation, Causation and Variable Selection in Multifactorial Models
Understanding psychopathology requires looking at a variety of different factors such as neurophysiology, genetics, environment, and behavior. If this is correct, promising models of mental disorders must be multifactorial in nature. One way to build such multifactorial models is by extracting patterns of correlations from big amounts of real-world data using computational and machine learning methods and subsequently visualizing them as some kind of graph or network-like structure.
While these network models are intuitive to grasp, they also raise a number of questions. First, observed correlations between variables underdetermine the underlying relations between variables. Second, multifactorial models are often rather unconstrained, viz. there are currently no clear criteria as to what variables precisely should be included. In this talk, we shall examine what useful constraints on multifactorial mental disorder models might be—and how this might also help disambiguate the sources of observed correlations in psychiatric data.
Model-Based Diagnostic Reasoningin Clinical Psychiatry
A central task forclinical psychiatrists is to assess the psychopathological condition of patients. Usually, this means to determine present signs and symptoms and select a diagnostic category from a diagnostic manual. How do psychiatrists fulfill this task? While “it is not sufficient to check off the symptoms in the diagnostic criteria to make a diagnosis” (APA, 2013, p19), most expert communities’ taciturn answer to this question is “clinical judgment” and “clinical training”. Unfortunately, this answer only relocates the question: What does an adequately trained psychiatrist do while diagnosing patients? What distinguishes a good clinical judgment from a bad one? What is needed to answer these questions is to understand the reasoning responsible to ensure diagnostic quality. Philosophers and scientists interested in medical cognition have offered different approaches to do so. In my talk, I will first focus on one of these answers that various proponents of phenomenological oriented psychiatry have adopted (e.g., Zahavi, Fuchs, Parnas): the pattern recognition approach. I will show what makes this approach inadequate to account for proper diagnostic reasoning in psychiatry and, subsequently, present a more adequate alternative: The Theory of Model-Based Psychiatric Diagnostics.
Mental Illness, Brain Disorders, and Metacognitive Skill
What we think mental illness is is important. Our conception of mental illness not only affects treatment; it can directly impact how the afflicted are treated by society, and how they view their conditions and their prospects of recovery. There is a growing push amongst mental health organizations and patient advocacy groups to classify mental disorders as ‘brain disorders’. Such proposals are justified on both theoretical and pragmatic grounds. The theoretical rationale is that this is the most accurate way of describing mental disorders, since science is showing that mental functions are deeply connected to brain functions. The pragmatic rationale is that a brain-based conception of disorder (e.g., ‘depression is a disease just like cancer is a disease’) will help to reduce stigma and encourage increased access to mental health resources. In this talk, we defend an alternative, skill-based conception of mental illness, on the grounds that it is both theoretically and pragmatically superior to brain-based views.
The theoretical problems of the brain disease model are well known. Notably, the model relies on concept of ‘mental dysfunction’ to explain what is disordered about mental disorders, however there is currently no consensus on what constitutes normal mental functioning (and it is unclear if such a consensus is forthcoming). There are also serious pragmatic problems with the brain disease model. While there is some evidence that conceiving of mental illnesses as brain disorders decreases stigma (though, this is controversial), it also appears to increase essentialist conceptions of illness (i.e., beliefs that illness is caused by biological processes over which individuals have little control), and pessimism about recovery. While the ‘brain disorder’ narrative may increase access to treatment, it may also hinder recovery and diminish patients’ agency. We believe that a more plausible view emerges if we replace the focus on brain functions with a focus on metacognitive skills. Just as one can develop embodied skills (e.g., athletic or musical abilities), or intellectual skills (e.g., chess or mathematical competence), one can develop skills at regulating one’s own cognitive capacities and dispositions (e.g., emotion, attention, learning, belief fixation, etc.). Such metacognitive skills enable the kind of self-regulation that arguably constitutes mental health.
We show that a focus on skill, rather than function, provides us with a more explanatorily robust and pragmatically useful theory of mental health. On a theoretical level, this conception of mental illness allows for more intuitive assignment of extensions to our mental illness categories, because it introduces a kind of ‘slack’ between biological function and mental health: not all deficits in skill need involve biological dysfunctions. On the pragmatic level, the focus on metacognitive skill rather than function enables a conception of mental disorder that promotes human agency while also avoiding essentialist conceptions of mental disorder. Finally, conceiving of mental health as skilled metacognition is consistent with the scientific study of psychopathology. In the same way that empirical study of the brain is relevant to understanding embodied and intellectual skills, so it is to understanding the metacognitive skills that constitute mental health. This is not an anti-psychiatry view. It merely reconceptualizes mental health: rather than falling into the category of biologically proper functioning, it falls into the category of skillful metacognitive regulation.
(Dimence Mental Health Institute, Zwolle)
Mosaic Models & Clinical Reality
In the past years, the importance of shared-decision-making in psychiatric treatment and the management of psychiatric symptoms and their consequences has received increased and much-deserved attention. A central challenge for shared decision making is how clinicians, patients, and significant others are to co-design treatment plans and implement them given all these different perspectives on psychiatric disorders.
The idea of “perspectival mosaic unity” is introduced as a pragmatic foundation for shared decision making in psychiatric practice. Different aspects of the idea are highlighted in relation to clinical reality. For instance, perspectival mosaic unity can serve as a means for clinicians to explore the phenomenological alienness of psychiatric illness. Furthermore, it can be used to help those suffering from psychiatric symptoms to develop their mentalization skills and heighten their autonomy (with treatment resistant symptoms, the improvement of personal autonomy must procede in spite of persisting psychiatric symptoms).
Interventions in PTSD
Posttraumatic stress disorder (PTSD) may develop upon exposure to an extremely threatening event or a series of such events. Symptoms include re-experiencing the traumatic event, avoidance of thoughts and memories of the trauma and of activities, people, or situations reminiscent of the event(s), and persistent perceptions of heightened current threat. According to clinical guidelines, trauma-focused psychotherapies (TF-PT) such as trauma-focused cognitive behavioural therapy (TF-CBT) and eye movement desensitization and reprocessing (EMDR) are recommended as first-line treatments for PTSD. As some patients fail to respond or have comorbid symptoms or disorders that only partially decline with TF-PT, there is growing interest in augmenting TF-PT through adjuvant interventions. Preliminary evidence suggests that endurance sports, breathing biofeedback, or cortisol administration may have an adjuvant effect on PTSD symptom reduction.
(Western University, Ontario)
New Frontiers in Translational Research: Touchscreens, Open Science and The Epistemic Community
There is widespread consensus that progress in understanding the neural mechanisms that underlie impairments in high-level cognitive functions that accompany many neurodegenerative, neuropsychiatric and other brain disorders is not achievable within the confines of a single laboratory; rather, it requires a paradigm shift in the way that neuroscience is done, with teams of researchers across the globe actively collaborating to propel discovery forward. During the past two decades, rodent behavioural researchers combining the touchscreen approach to assessing cognition in animal models with cutting-edge visualization and intervention techniques have led an unprecedented charge for standardized, collaborative, open, methodologically transparent, and reproducible neuroscience. Indeed, over the past decade, a global research community has emerged around the use of rodent touchscreen technology with “over 300 different research groups at over 200 research institutes in at least 26 countries” using the technology (Dumont, Salewski & Beraldo 2020). Knowledge mobilization and global community building efforts spearheaded by Western researchers have culminated in the development of a novel state-of-the-art technologically innovative and methodologically integrative service platform—the Mouse Translational Research Accelerator Platform (MouseTRAP)—which is centered on rodent touchscreen technology and directed at accelerating discovery in translational research by means of open, accessible, and reproducible science (Sullivan, Dumont, Memar et al. 2020).
These developments offer a novel occasion to reflect on how to transform the global community of rodent touchscreen researchers into a properly “epistemic culture” – a distributed community of researchers working together to “create and warrant” translational “knowledge” (Knorr Cetina 1988) of the kind needed to facilitate cross-species translational research and propel forward mechanistic and therapeutic discovery. Engaging in such reflection from a philosophical perspective is the aim of this talk, with an eye towards deriving broader lessons for translational research.
(University of Texas at San Antonio (UTSA))
Conceptual Engineering in Psychiatry: Expertise Naturalized
Recent methodological debates on the value of traditional conceptual analysis in philosophy (Strevens 2019; Machery 2017) has resourceful implications for both philosophy of psychiatry and scientific psychiatry. For example, the idea that concepts are universal in nature has been criticized, with experimental philosophers showing how different social groups have different intuitions about the application of certain concepts. This has led to the development of new approaches in philosophical inquiry, i.e., naturalized conceptual analysis, and conceptual engineering (Machery 2017). Naturalized conceptual analysis requires empirical methods to be pursued successfully which will then allow philosophers to arrive at knowledge that is within their epistemic reach. Such empirical tools include, but are not limited to, experimental philosophy and actual empirical knowledge about the domain the concepts are about (e.g., psychiatry).
The goal of conceptual engineering is to remedy the epistemic flaws in our concepts such as obscurity, imprecision, etc.
Taking cues from this debate, the proposed talk calls for conceptual engineering (Machery 2007) of the concept of “expertise” in scientific psychiatry. In other words, I propose using both methods of naturalized conceptual analysis and empirical research to clarify and modify the definition of the “expertise” operative in scientific psychiatry so that it includes not only the perspectives of those who are formally trained in psychopathology as recognized by academic degrees, e.g., MD, but also the perspectives of patients and their care-givers who have direct experience with mental disorders. Expanding the concept of “expertise” to include experience- based experts (Collins and Evans 2002) not only reduces the epistemic flaws pervasive in the notion of “expertise” used in contemporary scientific psychiatry but also paves the way to a “developing, untidy, methodological pluralism” (Solomon 2015) in psychiatric epistemology.
The dominant perspectives in scientific psychiatry ascribe the status of expertise only to those who have clinical training in psychopathology. Take the Diagnostic and Statistical Manual of Mental Disorders (DSM), and the Research Domain Criteria (RDoC), i.e., the primary classificatory schemas currently used to expand scientific knowledge on mental disorders. In the various versions of the DSM since 1952, and the RDoC the notion of “expertise” has been limited to those with clinical training in psychopathology. An unanalyzed assumption in both of these documents has been that patients and their care-givers lack the expertise required for the scientific investigation of mental disorders.
Empirical evidence that challenges this assumption comes from three sources; (i) the first-person accounts of those who experience or witness mental disorders –usually available in the form of memoirs (Flanagan 2013); (ii) clinical diagnostic scales or medical case studies that are prepared in light of direct encounter with patients (Parnas 2012, Ankeny 2017) and (iii) amateur/citizen/user-led research conducted outside of traditional academic settings by the mental health users themselves (Cooper 2017). I invite philosophers of psychiatry and scientific psychiatrists to modify the concept of expertise so that it opens up multiple venues for expanding psychiatric epistemology.
Naturalized conceptual analysis is epistemically beneficial because it helps
philosophers identify concepts that are invalid, obscure, or ambiguous. What follows conceptual
analysis is conceptual engineering, which is defined as the modification of an existing concept in
light of naturalized conceptual analysis and empirical data.