Brain Imaging and Psychiatric Classification

Fielding and Marwede attempt to lay down directions for an applied onto-psychiatry. According to their proposal, such an enterprise requires us to accept certain metaphysical and methodological claims about how brain and experience are related. To put it in one sentence, our critique is that we find their metaphysics questionable and their methodology clinically impracticable.

A first fundamental problem for their project, as it is expressed in their paper, is that their overall aim is unclear. At least three different aims might be read as motivating their efforts, here listed according to their strength:

1. 1. They aim to develop tools for the development of knowledge representational systems that can be used by clinicians in their efforts to diagnose and treat psychiatric illnesses.

2. 2. They aim to develop tools for the development of knowledge representational systems in order to facilitate interdisciplinary biomedical research in psychiatry, for example, research in the etiologies of psychiatric diseases.

3. 3. They aim to provide an ontological description of specific types of images used in the domain of medical research and practice.

We discuss them in turn.

Brain, Cognition, and Diagnostic Practice

Fielding and Marwede Aim to Develop Tools That Can Be Used by Clinicians in Their Efforts to Diagnose and Treat Psychiatric Illnesses

Fielding and Marwede (2011) do not explicitly make this claim but it is suggested by the parallel they draw between x-ray images of fractured bones, computed tomography (CT) images of lungs and tumors, and finally functional magnetic resonance imaging (fMRI) of patterns of brain activity. Beside facilitating research in anatomy and certain medical pathologies, the value of the different imaging techniques is obviously that they enable clinicians to diagnose the patients, articulate prognostic indications, and tailor therapeutic strategies. With an image of a fractured bone, we have a manifest picture of what is wrong. With a picture of a tumor, the issue is more complicated, but still we would have a manifest indication of the cause of the patient's problems. Indeed in both cases the imaging method reproduces salient information structurally tied (on a primarily biomechanical and anatomical-topographical level of description) with the causal disease entity (i.e., a physical fracture of the bone, an abnormal proliferation of neoplastic tissues). Such connection, even adopting more sophisticated, functional imaging techniques (like functional connectivity, diffusion tensor imaging, magnetoencephalography) remains inescapably problematic within psychiatry, first because of the intrinsic questionability of the correspondence between consensus-based contemporary diagnostic classifications and their subsumed domains of pathological "entities" (Boteva and Lieberman 2003; Jablensky 2005; Oulis 2008) and second because of the reductivist difficulties of the current paradigms trying to bridge the personal and subpersonal descriptive domains (Fuchs 2002; Hohwy and Rosenberg 2005; Fusar-Poli and Broome 2006). In sum, unlike x-rays of fractured bones, in psychiatry the relation between the brain image and the disease entity is highly problematic.

To come back to Fielding and Marwede's proposal, information in a knowledge representation system is structured in accordance with a knowledge model called an "applied ontology" (Fielding and Marwede 2011, 301). An applied ontology structures information in classes and lays down rules for the different types of relations between classes. This allows the software to generate inferences concerning specific entities contained in the ontology. As Fielding and Marwede point out, for these inferences to be sound and of any practical value, they need to be constrained by what the world is like. Not just any kind of classification of entities will do; only classifications made in accordance with a realist ontology. What a realist ontology (in philosophy) is supposed to be is in itself problematically vague in the paper. It would seem to be something like our best available scientific understanding and explanation, here understood as a form of materialism or physicalism.

Again, this is all very well with fractured bones and tumors in the brain. There we do have pictures of what is wrong. We know what we have a picture of. This kind of knowledge is simply missing in the case of neuroimaging studies of schizophrenic patients...