Abstract
Psychopathology classification is at a conceptual crossroads. It is becoming increasingly accepted that the flaws of the DSM relate to its struggles to pick out “real” entities as opposed to clusters of symptoms. The Research Domain Criteria (RDoC) was formulated in response to this failure, and attempts to address the concerns confronting the DSM by shifting to a causal and continuous model of psychopathology. Noting key criticisms of neurocentricism and problems with conceptual validity leveled at the RDoC, we argue that they stem from its grounding in the metaphysical position of eliminative materialism, or at least material-reductionism. We propose that 3e cognition (viewing the mind as embodied, embedded, and enactive) offers a superior alternative to eliminative materialism. A 3e-informed framework of mental disorder is sketched out and its advantages as a basis for classifying and conceptualizing mental disorders are considered.
The classification of individuals’ clinical presentation into dimensions or categories, such as depression, facilitates communication and access to relevant scientific literature, thereby guiding both data exploration and theory generation. The impact of a diagnostic system increases as we move up from the study of psychopathology in individuals to populations, where, in many ways, diagnostic systems shape the landscape in which psychopathology research is done. Diagnostic systems constrain which putative disorders get studied and how we conceive of disorders in the first place. The challenges arising from these issues apply to the research and practice applications of the Diagnostic and Statistical Manual of Mental Disorders (DSM), and to alternative models of conceptualizing psychopathology such as the Research Domain Criteria (RDoC). There are conceptual links between models of mental disorder, classification systems, and psychological science that are of paramount concern to both researchers and clinicians in the field of psychopathology.
In this article we first briefly overview some of the conceptual assumptions of the DSM and the problems that arise from these. We then explore in more detail how the RDoC attempts to address these challenges. We argue that in doing so, it also makes some problematic assumptions concerning the nature of psychopathology and human functioning in general. On the basis of the work of Murphy (2017), we propose that these underlying assumptions reflect a commitment to a view of the mind known as eliminative materialism, or at least material-reductionism. While theoretical commitments about the way the mind is organized and its relationship to the body and environment are needed to support any psychopathology classification system, in our view eliminative materialism has led RDoC researchers astray. We propose that a framework referred to as 3e cognition 1 (the mind as embodied, embedded, and enactive) provides a more integrative and richer framework within which to study mental disorders. In this paper we present a theoretical sketch of what disorders look like from the perspective of 3e cognition, and argue that it provides greater conceptual clarity as to why some behaviors should be considered disordered.
Assumptions of the DSM and RDoC
Diagnostic and Statistical Manual of Mental Disorders (DSM)
The DSM is often referred to as a “signs and symptoms” approach to classification. It works from the reasonable assumption that the causal processes supporting psychopathology are complex and hard to obtain knowledge of. The DSM sidesteps this difficulty by being “atheoretical” with regards to etiology. This means that rather than basing diagnostic constructs on a set of causes or underlying processes as is the case in other areas of medicine (causalism), diagnostic constructs are inferred from observed patterns of clinical features across the relevant population (descriptivism). Under the DSM’s descriptivist model, signs and symptoms observed to co-occur and be associated with harm and functional impairment are given a label of mental disorder. The central issue of relevance is that this model is solely focused on reliability of diagnosis, and not whether these diagnostic constructs pick out common causal processes (Zachar & Kendler, 2017).
One strength of the DSM model is that it makes an attempt at what Wakefield (2014) calls conceptual validity. Since the publication of the DSM III, a mental disorder diagnosis has required the presence of some degree of harm or functional impairment for the individual concerned (American Psychiatric Association, 1980). This is vital as it helps to justify why a particular set of signs and symptoms should be labeled as a disorder rather than just an atypical variant of what is essentially normal functioning. Some authors have noted, however, that this requirement has been watered down in DSM-5, with a change of wording in the preamble concerning the definition of disorder. This now states “Mental disorders are usually associated with significant distress or disability in social, occupational, or other important activities” (American Psychiatric Association, 2013b, p. 20), alongside a removal of a harm criterion from many diagnoses (Cooper, 2013).
The DSM’s model of disorder is descriptivist rather than causal as described above. Yet, seemingly in conflict with this, DSM constructs are often treated as essential, objective, and categorical entities (these labels are discussed later). As has been explored elsewhere, this conception is problematic and has given rise to many criticisms including the issues of artifactual comorbidity 2 (Andrews, Slade, & Issakidis, 2002), symptomatic and etiological heterogeneity (Lilienfeld, 2014), false positives (Cooper, 2013; Wakefield, 2015), and the problem of reification (Hyman, 2010). All things considered, the DSM is conceptually lacking. The key point for our discussion, however, is that it does not adequately pick out valid clinical entities. This has important ramifications for both research and treatment, where we want to be able to assume that a disorder has similar causes and solutions respectively.
Research Domain Criteria (RDoC)
RDoC is a research framework proposed by the US National Institute of Mental Health (NIMH) in direct response to the acknowledged problems with the DSM. An important research goal is to shift attention from surface features to the underlying causal processes that generate signs and symptoms; it is a causalist model (Insel et al., 2010). RDoC adopts a central organizing structure in the form of a two-dimensional grid, with the horizontal axis containing seven “units of analysis” which are largely structural in nature, and the vertical axis containing five domains/constructs, also referred to as systems, which are functional (Cuthbert & Insel, 2013; Cuthbert & Kozak, 2013; Lilienfeld & Treadway, 2016; Morris & Cuthbert, 2012).
While the RDoC is not a diagnostic system, it is intended to lay the groundwork for one (Insel et al., 2010), although it is explicitly uncommitted to the form that this diagnostic system might take (Cuthbert & Kozak, 2013). However, it is reasonable to presume that diagnostic entities in the system will represent dysfunctions of the identified functional systems, observed at or through the lens of the various units of analysis.
The three foundational postulates of the RDoC are stated clearly by Morris and Cuthbert (2012): (a) psychiatric disorders are dysfunctions of brain circuits, (b) the tools of neuroscience can identify these dysfunctions, and (c) clinical neuroscience alongside genetics research will yield bio-signatures of dysfunction that will augment classical clinical signs and symptoms of disorder.
Within the RDoC the explanatory focus has been shifted from DSM diagnostic entities (clusters of signs and symptoms) to transdiagnostic mechanisms that are thought to underlie them (Cuthbert & Insel, 2013; Hoffman & Zachar, 2017). The hope is that the identification of such mechanisms will allow for faster scientific progress, translation across levels of analysis, more precise medication and treatment, and perhaps even lead to the development of reliable bio-markers of psychopathology (Cuthbert, 2014; Cuthbert & Insel, 2013; Cuthbert & Kozak, 2013; Insel et al., 2010; Morris & Cuthbert, 2012). Hoffman and Zachar (2017) also point out that the narrowing of scientific attention from disorders to transdiagnostic mechanisms (while being mindful of the original purpose for seeking an explanation) may hopefully provide a better understanding of the relationships between levels of analysis; the logic being that diverse phenomena at the macro level will be constituted by simpler and more homogenous sets of mechanisms at the micro level.
To clarify the RDoC’s value as a possible classification system, we evaluate it against a conceptual taxonomy presented by Zachar and Kendler (2007). This taxonomy features six important factors upon which conceptions of psychopathology often differ, and offers a concise way of sketching out conceptual positions in this area. Along the way we will state some key criticisms of the RDoC approach.
Causalism/descriptivism
This factor relates to the question “Should psychiatric disorders be categorized as a function of their causes (causalism) or their clinical characteristics (descriptivism)?” (Zachar & Kendler, 2007, p. 557). The primary motivation for RDoC is to shift to a causal model, one that picks out etiologically valid constructs in a way that the descriptivist DSM does not.
Categories/continua
This factor relates to the question “Are psychiatric disorders best understood as illnesses with discrete boundaries (categorical) or the pathological ends of functional dimensions (continuous)?” (Zachar & Kendler, 2007, p. 559). RDoC is committed to viewing the symptoms of psychopathology in dimensional terms, in opposition to the categorical DSM approach (Cuthbert & Insel, 2013; Cuthbert & Kozak, 2013; Lilienfeld, 2014; Lilienfeld & Treadway, 2016). This means that features of psychopathology are viewed as quantitative extensions of normal behaviors or biological states and therefore exist in degrees. This is a significant strength of RDoC and aligns with current evidence for the vast majority of constructs in the field of psychopathology (Haslam, Holland, & Kuppens, 2012; Markon, Chmielewski, & Miller, 2011).
Essentialism/nominalism
This factor relates to the question “Are categories of psychiatric disorder defined by their underlying nature (essentialism), or are they practical categories identified by humans for particular uses (nominalism)?” (Zachar & Kendler, 2007, p. 558). In order to first locate the hypothesized transdiagnostic mechanisms it is a primary intention of RDoC to reverse the DSM psychopathology research model of noting clusters of signs and symptoms within the population and then investigating them. The resulting research model is one whereby abnormalities across units of analysis are discerned during the study of both typical and atypical populations, and it is later observed how these atypicalities may be serving as causal mechanisms (Cuthbert & Insel, 2013). Given the stated commitment to a dimensional conception of disorder, if mental disorders exist in degrees, then research methods that cover the whole population and capture such continuous variation are needed. For this model to make sense, the assumption has to be made that the atypicalities exist in nature, waiting to be discovered. In this way there is an essentialist element to RDoC. However, in RDoC’s current state of development it is not clear whether this is the full story. If these observed atypicalities themselves constitute a disorder in the classification system that evolves from RDoC, then this would indeed seem quite essentialist. However, if these mechanisms, or perhaps regularly observed mechanistic clusters, are labeled as “disorder” and this is done for normative or practical reasons, then this would situate the resulting diagnostic system as moderately nominalist.
Objectivism/evaluativism
This factor relates to the question “Is deciding whether or not something is a psychiatric disorder a simple factual matter (“something is broken and needs to be fixed”) (objectivism), or does it inevitably involve a value laden judgment (evaluativism)?” (Zachar & Kendler, 2007, p. 558). Given RDoC’s empirical intentions to work from the bottom up noting atypicalities across the population and from this inferring disorder, it seems very likely that the RDoC belongs in the objectivism camp.
This, and the discussion around essentialism versus nominalism above, is relevant to a criticism of RDoC raised by Wakefield (2014). In his paper, Wakefield criticizes the RDoC for its lack of conceptual validity, arguing that it has great difficulty explaining why a set of phenomena should be labeled a disorder. In its current form the RDoC relies merely on the abnormality of a phenomenon and its probabilistic association with some poorly defined harmful outcome. Grounded in his Harmful Dysfunction model, Wakefield demonstrates that this is not sufficient, and is actually a step backwards from the DSM, which at least attempts to delineate the disordered from the simply atypical. Abnormality is not disorder. Diagnoses should be given with the interest of the client in mind, and we should not pathologize behavior unless there is evidence of dysfunction or harm.
Internalism/externalism
This factor relates to the question “Should psychiatric disorders be defined solely by processes that occur inside the body (internalism) or can external events also play an important (or exclusive) defining role (externalism)?” (Zachar & Kendler, 2007, p. 558). The RDoC would certainly fall under the internalism banner. Of the seven units of analysis, five are “beneath the skin,” with the other two being behavior and self-report, which are still focused on the individual rather than on interpersonal or situational factors. Further, explicit conceptual focus is given to the “brain-circuit” level, at which disorders are primarily located. The privileged status of neural circuits in the explanation of mental disorders makes it vulnerable to the criticism of being overly “neurocentric,” and reductionistic (Berenbaum, 2013; Hershenberg & Goldfried, 2015; Hoffman & Zachar, 2017; Kirmayer & Crafa, 2014; Lilienfeld, 2014; Lilienfeld & Treadway, 2016).
Entities/agents
This factor relates to the question “Should psychiatric disorders be considered to be things people get, or are they inseparable from an individual’s personal subjective make up?” (Zachar & Kendler, 2007, p. 559). The RDoC would seem to fall somewhere in the middle of these two possibilities, perhaps leaning slightly towards the entity view. Given its stated focus on lower units of analysis, it seems to lack the holism required to encapsulate an agential purposive perspective. At the same time RDoC does not see sufferers entirely as mere vehicles of mental pathogens. Rather, it alludes to the biological norms of an organism and includes behavior as a unit of analysis, which suggests an awareness of an interplay between disease processes and individual agents.
Conclusions regarding the RDoC
In this discussion we have focused on two key criticisms which we see to be fundamental, and therefore intractable without radically changing the RDoC’s central assumptions. First, the claim that the RDoC is too neurocentric (Berenbaum, 2013; Hershenberg & Goldfried, 2015; Hoffman & Zachar, 2017; Kirmayer & Crafa, 2014; Lilienfeld, 2014; Lilienfeld & Treadway, 2016). RDoC authors have attempted to rebut the claim of neurocentricism, but it remains a popular criticism (Cuthbert & Kozak, 2013). Second, the argument that it lacks conceptual validity; it is not clear why an atypicality noted in the RDoC framework should be seen as a dysfunction/disorder (Wakefield, 2014).
We propose that these two problems stem from the same root; RDoC’s underlying assumptions concerning the nature of the mind. Murphy (2017) has recently argued that the RDoC is grounded in a philosophy of mind position known as eliminative materialism, albeit in a moderate form. This is the view that phenomena at higher levels—such as human cognition and behavior—are ultimately reducible to lower levels such as the biological or molecular, and that “folk psychology” explanations rooted in higher levels will be eliminated or heavily revised as science progresses. Whether one agrees with Murphy’s particular labeling or not, it is at least fair to say that he demonstrates that the RDoC has reductive and materialist aspirations. This observation is all that is required for our argument to stand. Such reductive aspirations make sense of RDoC’s neurocentricism, and the conceptual validity issue. This is because the labeling of a phenomenon as a disorder seems unjustified without reference to norms and values from which to demarcate harm or dysfunction. As we will show in following sections, values are better conceptualized as emergent properties of the entire organism and are therefore difficult to account for under a reductionist framework. This of course begs the question of whether there is an alternative to eliminative materialism. The following section argues that 3e cognition is such an alternative, and offers a sketch of what psychiatric disorder might look like from this perspective.
Before moving to this, however, we should briefly mention that we are aware there will be disagreement over how the RDoC is represented here. Regarding the view that RDoC sees mental disorders as brain disorders, Cuthbert and Kozak (2013) state that “this controversial assumption is neither essential nor inherent to the RDoC initiative” (p. 931). They make the alternative claim that “statements [which appear neurocentric] are interpretable as an expression of the need to move beyond symptom-based nosologies for mental disorders” (p. 931). Such a position aligns with the more integrative aspirations alluded to on the RDoC website regarding the need for developmental and environmental considerations. However, it is directly contradictory to the core assumptions of the RDoC stated by Insel et al. (2010) and Morris and Cuthbert (2012), which are clearly brain focused (listed earlier in this article). This inconsistency suggests variation in the conceptual positions of the RDoC authors, with some positions being less neurocentric than others. For clarity, this article assumes RDoC to take the more neurocentric position as per its stated core assumptions. However, the general argument is still applicable if subscribing to RDoC a more moderate position, despite the fact that such views may not constitute eliminative materialism proper. While the claim of neurocentricism is partially weakened against these more moderate views, the criticism of lacking conceptual validity still holds. The RDoC of Cuthbert and Kozak (2013) fares much better than that of the more neurocentric parties, but still underrepresents both socio-cultural factors and the normative nature of diagnosis. In contrast, our 3e framework places values and norms as central to the diagnostic process, an implication that we will explore elsewhere (Nielsen & Ward, 2018).
3e cognition
By 3e cognition, we refer to the view that the mind is embodied, embedded, and enactive. (Key resources outlining the 3e viewpoint include Butz & Kutter, 2017; Durt, Fuchs, & Tewes, 2017; Fuchs, 2017; Gallagher, 2017; Gibbs, 2005; Maiese, 2016; Thompson, 2007.) Under this view, the mind is seen as wholly natural and constituted by the entire brain-body system, not only the brain; i.e., we are embodied beings. Mental processes are completely and necessarily embodied in the brain, nervous system, and all of the biological systems of the body. The embodiment conception of human functioning is based on a relatively simple idea: human psychological functioning and sense of meaning is shaped in fundamental ways by bodily experience. Additionally, every bodily system (i.e., sensory, motor, nervous, immune, endocrine, etc.) is either constitutive of, or is causally implicated in, psychological functioning and subjective experience (Gibbs, 2005; Johnson & Rohrer, 2007). It is not possible for minds to function independently of the body; they are essentially interrelated sets of processes. More than this, interactions with the physical and social environments within which the organism is situated are seen as necessary conditions for the development of the mind over time; i.e., we are embedded. Finally, we are also enactive creatures (Gallagher, 2017; Thompson, 2007). According to enactivism, the mind is not a thing but rather an interrelated set of psychological capacities and powers that are essentially dispositions to act in accordance with an organism’s needs, interests, and respective goals; the mind emerges (is enacted) by virtue of the organism’s needful relationship with the world.
Thus, in essence, the mind is not a linear symbol-processing machine with a defined input and output, but rather an emergent property of the whole organism arising from interactions in the brain-body-environment system (Thompson, 2007). This also has the effect of making human cognition much less dependent on cognitive representation and processing: “a natural cognitive agent—an organism, animal or person—does not process information in a context-independent sense. Rather it brings forth or enacts meaning in structural coupling with its environment” (Thompson, 2007, p. 58). More than this, under 3e cognition, the valenced nature of our experience—the meaning that is immediately apparent in the world around us or what Maiese (2016) calls affective framing—can be seen as real.
Readers familiar with these ideas will note that we are using the term 3e, when often the term 4e is used. We do so because we do not subscribe to the fourth “e” (the extension of the mind, where the mind is seen as partially constituted by the external environment; Clark & Chalmers, 1998). Briefly, this is because full extension is not compatible with enactivism and embodiment, and that much of what “extension” achieves can be achieved via the concept of embedment (see Maiese, 2018; Thompson & Stapleton, 2009).
Previous work in this area
Limited work has been done to bring conceptual analysis of the nature of mental disorders together with the 3e cognition perspective. Drayson (2009) lays down the challenge “for embodied cognitive science … to … come up with an explanatory model of the origin and development of psychiatric disorders that can adequately compete with the current orthodox model” (p. 339). She argues that such a model would potentially show great promise for those disorders with large bodily components such as impairments of mood or eating, but questions how it could be applied to psychiatric problems in which representational content, such as delusional disorders, has a prominent role. We view these as open questions. It is our intention to begin answering Drayson’s (2009) challenge in this paper, and attempt to sketch out a preliminary conceptual framework; an image of what kind of thing mental disorders may be, using 3e cognition as a guiding set of ideas. This conceptual framework can later be used to guide the construction of explanatory models in line with empirical evidence.
Fuchs (2009), offers some insight into what such a conceptual framework could look like. His main conclusion is that it will necessitate multi-level analysis and a focus on circular causality operating across all levels. As a consequence, in contrast to the cognitively dominated orthodox models of psychopathology, Fuchs prescribes greater focus on perception and action as these are the primary modes by which we are coupled with our environment. Taking such a multi-level approach allows for a more comprehensive view across brain, body, and environment, as required by the 3e framework (see also Fuchs & Schlimme, 2009).
Some authors have attempted to generate 3e explanatory models of particular disorders. These models are grounded in a phenomenological approach which features a rich but often confusing terminology. We have attempted to translate these ideas into standard psychological concepts but acknowledge that in the process of doing so some coherency and richness may be lost.
Zautra (2015) presents an enactive account of addiction. He describes how current models fail to offer an account of the first-person experience of addiction, how they do not give sufficient weight to interactions with the social and physical environment, and how they tend to be based on an entity conception of addiction and fail to recognize the agency of the individual suffering. Zautra addresses these problems by emphasizing the “lived experience” of addiction. His model describes how exposure to the drug has developed a need for it within individuals, and how they meet this and other needs in accordance with the affordances and constraints of their environment. Further, the model emphasizes how having this dominant need changes the agent’s embodied experience in a multitude of ways; from attentional processes, through impulsivity, to their relationships with emotion.
While we think Zautra’s work provides a valuable description of addiction, one that is compassionate and that emphasizes agency, there are many issues with this model. Greater detail is needed concerning how the relevant need is constituted within the individual, how this is triggered by the drug, and why this need becomes dominant for some and not others. Essentially, whilst offering an insightful perspective on the subjective experience of addiction, the model does not offer the multi-level view that Fuchs (2009) suggests a 3e perspective should generate. Our analysis of Zautra’s model suggests two conclusions. First, an optimal model of disorder from the 3e perspective should align with the subjective experience of the sufferer, and thereby both generate compassion and define the explanandum. Second, it should offer mechanistic insights at multiple levels above and below the level of the first-person perspective in order to best guide treatment.
Fuchs and Röhricht (2017) and Maiese (2016) present embodied and enactive accounts of schizophrenia. Within these models the primary dysfunction is seen as a breakdown in the experience of the basic or bodily self, also referred to as the experience of ipseity; the first-person “givenness” of all experience. This breakdown results in a lack of unity of perception, action, and thought, whereby the relation to objects in the world, thoughts in the mind, as well as body parts and their actions, lack qualities of wholeness and “for-me-ness.” Response to these experiences produces hyper-reflexive self-observation and feelings of isolation and detachment. Trust in others thereby becomes very fragile, and the shared understanding of the world is damaged by this, fostering the development of delusions. Maiese’s model differs slightly, with greater focus on the role of affective relations. Both offer rich subjective accounts of schizophrenia, with some explanatory value in that they account for many signs and symptoms of schizophrenia as understandable psychological responses to a central feature. However, as with Zautra’s (2015) account of addiction, we do not believe these are fully explanatory models. It is unclear what the origins of the disruption to the experience of the bodily self are, and how this disruption to experience is constituted at lower levels of analysis. Finally, more specific detail is also required concerning the role of interpersonal, social, and cultural factors.
All three models offer first-person accounts of their respective disorders which have epistemic and pragmatic value. However, they miss factors situated at levels of analysis above or below the first person experience. This seems to be due to their grounding in phenomenology, which is largely descriptive and concentrates on subjective experience. Hence, these accounts focus on the experience of embodiment but de-emphasize the lower levels of analysis that constitute individuals, and factors from higher levels that constrain human functioning (e.g., social institutions); seemingly because both elements require a third-person perspective. In order to offer a sufficiently comprehensive causal account, cross-level analysis is required (Kendler, 2012a, 2012b; Kinderman, 2005).
Kyselo (2016) makes a similar observation concerning enactive accounts of schizophrenia. She outlines a model developed by Parnas and Sass (2010) which is centered around dysfunction in the experience of ipseity, and notes that it is descriptive rather than causal. Alongside this analysis she explores a proposal by Ebisch and Gallese (2015), based on an empirical review of recent neuroscientific evidence, that disturbances in the experience of ipseity and the distinction between self and others may be partly caused by a disruption in multi-sensory integration within the ventral pre-motor cortex. Kyselo (2016) argues that these two perspectives can be seen as complementary, one as a description, the other as a potential account of underlying causes. Kyselo goes on to offer two criticisms of importance. First, she states that both models’ stress on the first-person perspective results in a failure to emphasize social elements. She argues that such an individualistic focus does not make room for incidence and prognostic factors like social support and socio-economic status (Agerbo et al., 2015; Bhavsar, Boydell, Murray, & Power, 2014; Buchanan, 1995; Lim, Barrio, Hernandez, Barragán, & Brekke, 2017; Tsai et al., 2014), nor for consideration of cultural factors that may shape both contexts and the individuals’ understanding of their experiences. Second, she criticizes both papers for defining disorder simply in opposition to the normal, rather than in terms of the norms and values of the individual. This is related to the idea of conceptual validity mentioned earlier. A conceptual model of psychiatric disorder ought to make sense of why a cluster of clinical phenomena 3 should be labeled a mental disorder, and not do so simply on the basis of deviation from the normal. Kyselo goes further, presenting a view of psychopathology as “an altered form of striving for quasi-equilibrium in the organization of a person’s self” (2016, p. 607). This refers to seeking a balance between connecting to and differentiating ourselves from others, and is based on Kyselo’s socially defined conception of the self. Problematically, it is not clear how this model accounts for the phenomena which currently define the schizophrenia construct such as hallucinations, delusions, and negative symptoms. We do not therefore view Kyselo’s model as explanatory either. However, a particular strength of her model is that it attempts to define disorder on the basis of the norms of the individual rather than by the abnormality of the observed behavior. 4
Reflecting on these previous attempts, we arrive at three conclusions. First, previous 3e models of mental disorders have typically focused on the subjective level of explanation, likely due to their roots in phenomenology. While this approach is of value, comprehensive explanatory models need to employ a broader multi-level analysis (Kendler, 2012a, 2012b; Kinderman, 2005). Such a perspective is required given our embodiment as biological and social organisms (Fuchs, 2009; Fuchs & Schlimme, 2009). Second, as alluded to by Kyselo (2016), and in alignment with the criticism of RDoC raised by Wakefield (2014) concerning the need for conceptual validity, disorders need to be defined by more than just abnormality. Third, previous conceptual and specific explanatory attempts which draw on embodied and enactive perspectives have not attempted to integrate current literature surrounding conceptions of psychopathology.
What does 3e cognition give us?
A rich picture of human functioning
As a field, 3e cognition is consistent with naturalist and non-dualist assumptions. One key strength of the 3e view however, is how it meets these assumptions. It does this in a way that places equal value on biological processes and on first personal and interpersonal levels of explanation. In accordance with the comprehensive view prescribed by Fuchs (2009), genes and neuronal networks are vitally important for understanding the etiology and symptoms of psychiatric disorder, but so are emotional regulation skills, interpersonal relationships, and culture. Further than this widening of the lens across levels, 3e cognition also broadens our view laterally, to biological factors that have been historically overlooked because they lie outside the central nervous system. A good example of the relevance of extended biological processes are recent findings concerning the importance of the gut biome and nutrition for mental health (Kaplan, Rucklidge, Romijn, & McLeod, 2015; Rucklidge & Kaplan, 2013).
There are also important theoretical ideas contained within 3e cognition, such as those of emergence and constitution. Emergence is the view that a whole may gain properties from the interaction of its parts rather than being simply the sum of them. A classic example of this is the phenomenon of starling murmurations, where birds have been shown to respond to the seven or so birds in their local environment, resulting in what appears to be coordinated behavior of the whole flock, confusing predators (King & Sumpter, 2012). A simpler example is water. The property of water being a liquid is not held by a single H2O molecule, rather it is emergent from the interaction of multiple H2O molecules repelling each other due to their dipole structure.
Constitution is the idea that wholes can be made up of parts without the whole being eliminated or becoming meaningless as an explanatory entity. For example, if you build a tower of Lego, both the form of the tower and the Lego blocks exist and can be useful in an explanation of why the tower fell over under certain conditions. During the time that the Lego blocks constitute the tower, the blocks are the tower. Similarly, organisms are made up of many parts, and derive properties, such as mindedness, from the interactions between these parts. Both the parts and the organism are no less real because of the knowledge we gain about their parts and how they manage to constitute a minded creature. These conceptual tools are not available to reductionist perspectives such as eliminative materialism, yet they are arguably necessary for a comprehensive conception of psychopathology where an understanding of both wholes and parts, as well as the interactions between them, is required.
Related to its roots in dynamic systems theory and evolutionary theory, a final strength of a 3e cognition framework is that it provides a way to account for something akin to what Aristotle referred to as final causes (Falcon, 2015), and Kant called purposiveness (Ginsborg, 2014). Intuitively, the idea that a desired end state can cause an action seems to be a teleological error where the future acts backwards on the past, and is at odds with a mechanistic view of the universe. 3e cognition offers an elegant solution: the organism system is shaped by its ontogenetic and phylogenetic past to act in accordance with its needs, and to do so in accordance with the constraints of the environments that shaped it. This allows an organism to have purposes, goals, and even values inherent as tendencies within the organism system (Maiese, 2016; Thompson, 2007), and is achieved without the future acting backward on the past, thus granting a sort of naturalized teleology. To clarify, we do not wish to endorse final causes as a function of essence as per Aristotle, rather we wish to suggest that life forms can be seen as having purposes and goals in a non-trivial sense, in so far as they have been naturally selected across time to self-maintain and adapt (Thompson, 2007).
To briefly sum up what human functioning looks like from the 3e perspective (referring to both external and internal behaviors), multi-level explanations are required to account for behavioral phenomena, with no preference given to any particular level simply because it is higher or lower. Particular levels of explanation may be of specific importance in any instance, while the behavior itself from both a first- and third-person perspective is obviously of import as the explanandum (i.e., explanatory target). Up until this point, the picture is one similar to that painted by RDoC, however, this is where the similarities end. Relations between levels can explicitly be constitutional, a point left unclear in current formulations of the RDoC (Hoffman & Zachar, 2017); and phenomena can emerge at higher levels that would be impossible to predict from an understanding of lower level structures and processes. Furthermore, higher level processes can act downward to constrain lower levels. Finally, the explanatory tools outlined above allow us to see how, shaped by evolutionary and developmental histories, behaviors can have purpose in supporting the continuation and flourishing of the organism. The key point is that behavior cannot be accounted for by simply looking at neural processes. The body, the physical and socio-cultural environment, as well as considerations of evolution and development, are vital for understanding both why behavior is performed, and why it takes the form that it does.
Conceptual validity
3e cognition offers an account of how values can be emergent properties of organisms, developed over both evolutionary and lifetime time scales as tendencies that have served the maintenance of self-maintenance and adaption to the environment in the past (Di Paolo, 2005; Thompson, 2007). This is most clearly true for biological values such as food, warmth, comfort, and reproduction, which directly serve the function of maintaining biological norms. Maiese (2016), and Di Paolo (2005) explore how, in social and conceptually sophisticated animals, values of a less immediately biological nature can also be accounted for. Evolutionary and developmental pressure is generated by the fact that values such as mastery, autonomy, fairness, and so on allow us to fare well in the social environment in which we reside. Thus, over time the constraints of the social level act downward to shape the ontogenetic and phylogenetic development of the organisms that constitute the social group, and from this circular causality, values embodied in the dynamic tendencies of both the organisms and the groups emerge. Maiese (2016) offers the example of being a good driver: we wish to be good drivers to demonstrate our mastery which has positive social implications for us, not only so that we can avoid crashing.
We have seen that 3e cognition supplies a kind of naturalized teleology, and that this in turn allows for an account of values as objective properties of individuals. This is important because an understanding of values is vital for comprehending why a cluster of clinical phenomena should be labeled as a disorder. An understanding of values is required in order to conceptualize harm and dysfunction and demarcate it from functionality. Under 3e cognition we have argued that values are real and meaningful explanatory entities. They are also broad, serving not only biological norms, but agential and societal ones as well.
Pulling the framework together
Definitions
From a 3e cognition viewpoint, a behavior is harmful if it does not support the maintenance of biological norms or our ability to fare well, both of which can be seen through the lens of values as discussed above (as failing to support continued self-maintenance and adaption). However, the presence of such harm should not be seen as dysfunctional per se because we all do things at times that run counter to our values. For example, eating too much chocolate in one sitting, and feeling so terrible that you decide to attempt to induce vomiting does not constitute atypical harm at a population level, and therefore, while not ideal, is not inherently dysfunctional. However, if a recurring pattern of binging and purging develops to a degree that faring well in the social environment is challenged, autonomy is under threat, and healthy bodily functioning becomes a concern, this pattern of behavior can be viewed as dysfunctional. Help with moving away from a pattern of dysfunctional eating is likely to result in significant benefits to the person concerned. If similar patterns of behavior, with similar underlying causal factors, are seen across individuals, it seems reasonable to label this phenomenon as a mental disorder/diagnostic entity.
From this example and the above discussion, we can formulate initial working definitions of dysfunctional behavior, and mental disorder from a 3e cognition perspective. Note that for these working definitions, and indeed throughout this paper, “behavior” is used broadly, referring to both what have traditionally been termed external (movements) and internal behaviors (thoughts, affective responses, perceptions). It may also refer to dynamic patterns of, or tendencies in, behavior, and explicitly includes non-deliberative behaviors. Also note that atypicality plays a role in these working definitions, but that it is atypicality of the degree of harm, rather than atypicality of the behavior itself that is of interest here. Finally, “values” here refer to both biological values that directly serve the function of maintaining biological norms, and interpersonal and prudential values (e.g., autonomy, mastery, etc.) which indirectly serve such a function, and also directly support faring well in our social environments.
We offer the following tentative definitions: A dysfunctional behavior is a behavior that violates the values of the organism to a significant or atypical degree. 5 A mental disorder (diagnostic entity) is a pattern of dysfunctional behavior that occurs regularly across individuals.
Sketching the framework
Combining these definitions with the above discussion of what behavior looks like from the perspective of 3e cognition, a view of mental disorders emerges. What we have initially defined as dysfunctional behaviors are relatively stable dynamic patterns within the brain-body-environment system of individuals that run counter to their values to a significant or atypical degree. Mental disorders are recognizable patterns of such behavior occurring across individuals. 6 This sketch aligns well with the concept of a homeostatic property cluster, originally posited by the philosopher Richard Boyd (1991), and explored in reference to psychiatric disorder by Kendler and colleagues (Kendler, 2016; Kendler, Zachar, & Craver, 2011). It also aligns with the network approach to psychopathology (Cramer, Waldorp, van der Maas, & Borsboom, 2010; McNally, 2016), as well as the imperfect community model (Zachar, 2014). This ontological picture of a multi-level stable dynamic pattern then, is not particularly novel. What is novel is that our framework adds a normative and agential focus, thereby bringing greater conceptual validity while still maintaining biological and naturalist assumptions.
To add further detail to this sketch, we will attempt to describe what a classification system based on 3e cognition would look like, using the six-factor conceptual taxonomy presented in Zachar and Kendler (2007) that we applied to RDoC above.
Causalism/descriptivism
The 3e perspective conceptualizes disorders as relatively stable dysfunctional dynamic patterns within the brain-body-environment system (as per embodiment and embedment). Given the sheer complexity of that system such patterns will have multiple causal components and will differ across individuals. Therefore, categorizing disorders based on their causes is a challenging endeavor. However, this concept of mental disorder is still a causal one. Our position aligns with causalism but stresses the complex nature of the causes at play. The stable dynamic pattern view of disorder ultimately begs two questions: what is it that makes some individuals more likely to fall into this pattern in the first place (causal mechanisms), and what is it that makes the pattern relatively stable (maintenance mechanisms)? Maintenance mechanisms seem more likely to be common across different manifestations of a disorder, and also more relevant to treatment. For this reason, maintenance mechanisms may be better suited for a role in demarcating diagnostic entities. It is likely that during the developmental process of a classification system, causal knowledge in the form of empirical and theoretical science will continue to develop. Our understanding then, of causal and maintenance mechanisms, will in time shift from quite general to more specific until an optimal level for the pragmatic purpose of classification is reached. As a note, the view described here is open to the possibility of transdiagnostic causal and maintenance mechanisms.
Essentialism/nominalism
Interestingly, the conception of psychological disorder expressed here leans slightly more towards essentialism than one might think. If a pattern of dysfunctional behavior is seen to be occurring with some regularity across individuals, then this suggests that there is some tendency within the dynamics of the human brain-body-environment system to fall into such a pattern. This fact makes the disorder real rather than being a purely practical concept. On this view psychiatric disorders are bound together by similarities in their causal network rather than by sharing some essence. They are much more like a biological species than an atomic element in this regard, however they have no causal lineage as a species does. A “wing,” in being multi-realizable across species is perhaps a more apt analogy then. This type of kind is sometimes referred to as a “type-causal” kind (Magnus, 2014); similar to an attractor state in dynamic systems theory. However, a classification system is a practical endeavor and will therefore always be influenced by pragmatic concerns. Furthermore, given that the view presented here acknowledges the complexity and multi-leveled nature of causal structures supporting dysfunctional behavior, it seems very unlikely that a classification system is going to accurately “carve nature at its joints” any time soon. Because of these reasons we certainly do not think diagnostic entities should be viewed as completely essential entities. Thus, our view fits most with what Zachar and Kendler (2007) call moderate nominalism.
Objectivism/evaluativism
The framework presented here comes down on the side of evaluativism. Under this view values are seen as ubiquitous, and therefore necessary for a comprehensive understanding of human behavior (as per enactivism). While biological values will be the same across individuals, higher level values of both social and prudential kinds differ across individuals given different genetic and epigenetic predispositions, and different learning histories. This highlights the need to consider cultural values in both practice and research. Recognition of the ubiquity of values in practices such as clinical psychology and psychiatry, as well as science in general, is viewed here as essential for supporting ethical decision making (Douglas, 2009).
Internalism/externalism
It should hopefully be fairly obvious that the view presented here holds that both internal and external factors, as well as the interactions between them, are vital for a complete understanding of behavior and disorder (as per embedment). We would therefore fall under what Zachar and Kendler (2007) refer to as moderate externalism.
Entities/agents
Zachar and Kendler (2007) describe how the entity position generally views “individuals as vehicles for pathological syndromes,” while the agentic position holds that “each psychiatric disorder as manifest in an individual patient is relatively unique” (p. 559). We would certainly view each manifestation of dysfunctional behavior as unique in important ways, informed by both the view described here and clinical observation. Moreover, the very reason a cluster of phenomena should be seen as a disorder is because it will ultimately run counter to the values of the agent and therefore this concept of disorder is inextricable from a purposive and agential view. Our conceptual framework would therefore sit under the agential position. However, this is not to say that meaningful regularities across agents (e.g., disorders) cannot be extrapolated.
Categories/continua
The position sketched out in this paper views dysfunctional behavior as a relatively stable dynamic pattern, composed of many causal factors. Many of these factors will be continuous in nature, and therefore, the constituted behavior is largely continuous. That is not to say that the distinction between non-ideal behavior and dysfunctional behavior, nor the difference between different kinds of mental disorder will be meaningless. Rather, these distinctions are based on the degree of value-crossing, and empirical regularities across kinds respectively. These boundaries are certainly fuzzy, but far from arbitrary.
Discussion
In our view, 3e cognition offers a framework for understanding human behavior and mental disorders superior to the frameworks underlying both the DSM and RDoC. This is because of the more comprehensive view (Fuchs, 2009) and the explanatory tools that 3e cognition brings to the table. The more comprehensive view prescribed by 3e cognition addresses a key weakness of the RDoC—its neurocentricism (Berenbaum, 2013; Lilienfeld & Treadway, 2016). One important element that these explanatory tools provide is a naturalistic account of values (Maiese, 2016). This in turn allows greater conceptual validity in defining disorder in reference to the individual’s values rather than in contrast to normality—this being the second key weakness of the RDoC (Wakefield, 2014), and a weakness seen in previous attempts to conceive of mental disorder from an enactive perspective (Kyselo, 2016).
We have attempted to begin a sketch of what mental disorder may look like from the 3e cognition perspective, but recognize that there is much work to be done in developing this picture. Our construal of normativity itself is currently underdeveloped, as well as how exactly norms and values can play the role we have here prescribed for them in defining what counts as disorder. We see this as central and will explore this in another paper on these issues, building on these first steps (Nielsen & Ward, 2018). Another potential next step is to use these ideas in conjunction with current psychological and neuroscientific evidence to weave together a comprehensive causal model of a current disorder. This is beyond the scope of this article and may well constitute a research project in its own right. However, to make our analysis less abstract, we will briefly apply the 3e framework to the example of anxiety disorders and try to outline what a 3e conception might look like.
Anxiety disorders are traditionally conceived of as levels of vigilance and/or fear, disproportional to perceived threats, to a degree that is atypical and produces significant harm or impaired functioning (American Psychiatric Association, 2013a). This pattern of symptoms is usually explained by models which emphasize biological, cognitive, behavioral, social factors, or some combination of these. Explanatory models based on the RDoC matrix will do slightly better, attempting to locate constituent abnormally functioning brain circuitry and causal or risk factors across the seven units of analysis. An explanatory model built on the 3e framework, however, would attempt to construct a more integrated and comprehensive account whereby factors at explanatory levels beneath, at, and above the first-person level constitute and constrain clinical phenomena at the subjective level. To expand, a successful explanatory model would start with a rich description of the clinical phenomena. 7 These behaviors would include features of perception, action, and cognition, which represent the altered dynamics of the organism. Should this pattern of behavior be seen to threaten the values of the organism this would justify further investigation. In anxiety, examples of such phenomena would be: perceptual biases towards potential threat, the affective experience of worry and fear, repeated checking behavior, fatigue, sleep disturbance, irritability, etc. The description may also include reference to causes offering some explanation of the stability of the observed pattern of behavior. An example of this would be the experiential avoidance model of Obsessive-Compulsive Disorder, which is highly reafferent. A successful 3e model would then take this pattern of behavior and show how its components are embodied, e.g., how biological factors constitute the observed phenomena. In anxiety these factors may include but are not limited to: genetic polymorphisms, epigenetic factors, neuro-transmitter levels, hormone levels, gut micro-biota (Foster & Neufeld, 2013), neuronal structures, and the activity and structure of neural circuits and anatomical systems such as the amygdala and HPA axis. At the same time, the model would show how the pattern of behavior is constrained by the organism being embedded within the physical and socio-cultural environment. This will include direct causal links from environmental factors to the behavior, but also indirect causal links via the constituent biological factors. These factors are also often occurring across a larger time scale than the biological processes and may therefore necessitate developmental and longitudinal (or even historical and economic) perspectives. Examples in anxiety may include, but are not limited to: childhood history, the actual threat level of previous environments, modeling of anxious behaviors, the amount of food available and which nutrients and vitamins this food contains, exposure to drugs including licit ones such as caffeine and alcohol, relationship history (including parental relationships) and whether these relationships supported the development of self-efficacy, gender norms concerning management of distress, the culturally mediated understanding of what it means to be anxious, etc. It should be apparent that this is no easy task, indeed, this may reflect an ideal more than a practical possibility. A successful explanatory model from the 3e framework should integrate factors from across all of these levels, and explain how they lead to, shape, or underpin the clinical phenomena. Most importantly, it should also show how the phenomena will lead to an outcome that is counter to the biological or personal values of the agent, and thereby justify the labeling and investigation of said phenomena. It is this normative focus, as well as the more ready incorporation of bodily and socio-cultural factors, that differentiates this framework.
Conclusions
The 3e framework presented in this article allows for the convergence of psychological, neuroscientific, and phenomenological perspectives around a central conception of mental disorder. In contrast to the RDoC’s eliminative materialism, our framework is more agential and privileges phenomena at the subjective and behavioral levels as explanatory targets, necessitating consideration of personal values and culture. At the same time, this framework in no way suggests ignoring the brain and nervous system. To the contrary, it requires a comprehensive multi-level and constitutionally minded view, consisting of brain, body, and environment. Explanations in both research and practice will therefore be richer for being based on a 3e framework. This, however, also presents a challenge, as such comprehensive explanations are likely to be extremely complex, potentially challenging their pragmatic value. We suggest, in line with Kendler (2012a, 2012b), that integrative pluralism 8 (Mitchell, 2002, 2009) may be helpful as a metatheoretical tool when weaving together currently isolated explanatory perspectives, at least until such a time as integration is possible, if ever (Hochstein, 2016a, 2016b). Depending on the grain size of the explanatory components it may also be necessary, for similarly pragmatic reasons, to oscillate between systemic and mechanistic explanatory styles. 9 Given their grounding in dynamic systems theory, 3e perspectives can sometimes miss the specifics of the underpinning causal mechanisms (Bechtel, 1998). The 3e framework also shows conceptual overlap with the network model of psychopathology and therefore encourages the use of the related methodology of network analysis (Cramer et al., 2010; McNally, 2016; McNally et al., 2015). Further, in highlighting the defining role of the individual’s values, we hope that this framework encourages normative considerations in both research and practice, and places priority on the subjective level. Our next step will be to explore the normative nature of our framework and how this is grounded in 3e thinking, a side of our framework that is currently underdeveloped (Nielsen & Ward, 2018). It is at the level of the individual that suffering resides. Ultimately, explaining individual suffering and highlighting ways to alleviate it is arguably the most important goal of psychopathological research and therapy.
Footnotes
Acknowledgements
The authors would like to thank Hannah Hawkins-Elder, Roxy Heffernan, Samuel Clack, and the rest of the EPC Lab at Victoria for their feedback during the writing of this article, as well as helpful comments by Drs. Ben Sedley, Mary Barnao, and Dougal Sutherland. Many thanks also to the anonymous reviewers whose critical and well-considered comments were genuinely helpful and appreciated.
Author’s Note
Tony Ward is now affiliated with Ghent University.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.