by Chuck Ruby, Ph.D.

A recent article in the Journal of Abnormal Psychology, also publicized in Science Daily, reports on the results of research that is attempting to develop a better way to classify mental disorders by using a dimensional rather than categorical approach. The system is called the Hierarchical Taxonomy of Psychopathology (HiTOP). While it might correct some of the shortcomings of the present Diagnostic and Statistical Manual of Mental Disorders (DSM) diagnostic system, it nonetheless falls short. Its weakness is three-fold: 1) as with the DSM, it still regards human struggles as pathology; 2) despite its use of dimensions, it still retains the main categorical challenge of determining normal from abnormal human experiences; and 3) the factor analysis statistical approach does not always account for all variation among individuals' experience of distress.

First and foremost, the HiTOP continues to use a medical model and invokes the term “psychopathology”. It must be remembered that this is figurative language at best and misinformation at worst. There is no evidence that anything psychological or mental can literally be pathological. If any “mental disorder” is shown to be caused by true brain pathology, then it would fall within the medical specialty of neurology. The HiTOP continues to perpetuate the scientifically empty hypothesis that existential human problems are illnesses. While it does use a dimensional approach in an attempt to recognize human diversity, it still couches those problems in language that implies illness (e.g., pathology, patient, symptoms, syndromes, clinical, illness, comorbidity, diagnostic, psychopathology, etiology, pathophysiology, etc.)

More over, the HiTOP might actually be increasing the extent to which natural human struggles are considered pathological, even more than the DSM. For example, it is said that “[d]imensions are psychopathologic continua that reflect individual differences in a maladaptive characteristic across the entire population (e.g., social anxiety is a dimension that ranges from comfortable social interactions to distress in nearly all social situations)." But how can “comfortable social interactions” be considered psychopathological (or even problematic) and entire dimensions of human experience be "psychopathologic continua"? This seems to be saying that all human activity can be considered pathological.

So in continuing to use the medical model, but absent any evidence of biomarkers that identify the putative pathology, the HiTOP remains a pseudo-medical endeavor and suffers the same weakness of the DSM. It claims pathology but it must rely on moral judgements, not science, to identify pathology. Given the lack of evidence of literal pathology, who determines whether something is maladaptive?

The researchers consider the HiTOP to be a project alongside the National Institute of Mental Health (NIMH) program called the Research Domain Criteria (RDoC) as promising dimensional alternatives to the DSM system. But the RDoC and HiTOP suffer from the same weakness: they fail to start with evidence of real pathology. Rather, they assume it. In the case of the RDoC, neurological concomitants of human distress are described as pathology. With the HiTOP, natural variation of human distress is considered pathological.

The HiTOP’s second weakness has to do with how the researchers use the dimensional model. They pass up an opportunity to use the factor analytic statistical approach to create a system of hierarchical continua with which to describe the intensity and types of distress outside a medical and pathology model. Such an approach is seen in the NEO Personality Inventory (NEO-PI-R), which is an instrument providing a measurement of five personality factors, each with six facets. It is one of the few psychological instruments that does not conjure up pathology.

To the contrary, the HiTOP retains a focus on distinguishing between abnormal and normal human experiences. Thus, despite using a dimensional approach, it still saddles itself with a categorical decision.  This is the exact opposite of what the HiTOP was intended to do. It is said, “[t]his quantitative approach responds to all aforementioned short- comings of traditional nosologies. First, it resolves the issue of arbitrary thresholds and associated loss of information.”  Really? If the task is to differentiate between normal and abnormal, there will always be the need to set an arbitrary threshold between psycho-normality and psycho-pathology.

This problem is manifest in the following statement: “A common concern with dimensional classifications is whether they are applicable to clinical settings, as clinical care often requires categorical decisions. Indeed, actionable ranges of scores will need to be specified on designated dimensions for such a classification to work effectively in clinical practice. Rather than being posited a priori, these ranges are straightforward to derive empirically, as is commonly done in medicine (e.g., ranges of blood pressure, fasting glucose, viral load, etc.).”

But how do we determine that point (or range) between normal and abnormal human experiences? In real medicine, that decision is based on the extent to which the pathological factor threatens the physical viability of the patient. We do not have that with “mental health” because none of the ranges or levels of "mental illness" affect the physical viability of people any more than other behaviors not typically considered pathological affect physical viability, such as high risk recreational pursuits and sports.

The problem with using dimensional tools to assess “illness” is that the assessors must arbitrarily determine the separation between abnormal (ill) and normal (healthy) without any scientific foundation to do so. It gets back to what I said earlier, the assessor will be making a moral judgement about how much distress is normal.

Lastly, like all research, the variation of human experiences of distress cannot be explained completely by factor analyses. In some research a sizable portion of the natural variation of human distress is unexplained. This can be a problem for the HiTOP, but it is difficult to determine as the authors do not report the specific statistics that would allow one to make that determination. For instance, the eigenvalues, loadings, and communalities of the factor analyses in studies used to construct the HiTOP are not given. There can be a substantial amount of the factor analytic data that isn’t accounted for by the dimensional model.

Also, factor analyses are dependent on the scales/instruments that are used in collecting the original data. Therefore, information that is not collected via these instruments is not considered. The factor analysis identifies the relationship between the variables assessed by the instrument. It doesn’t identify variables that are not assessed by the instrument. If there is an important variable that is overlooked by the original instruments, it won’t be included in the factor analysis. So, if the instrument only looks at variables consistent with a medical model, as is done here, then only medical model factors will result.

The HiTOP is the latest in a line of "newer and better" diagnostic systems. Even though it might offer a more realistic view of human variation, it nonetheless still suffers the same weaknesses as other attempts to classify the scientifically threadbare pathology model of human distress.