There are three reasons for this:
First, while everyone understands very broadly what is meant by the term ‘experimental psychopathology’, it doesn’t seem to have a ‘manifesto’ as such (it is indeed the study of psychopathology using experimental and empirical methodologies – but, students, don’t quote that, it’s a tautology!). What I mean by this is that there is no manual that you can consult that tells you exactly how to do this kind of research and what the guiding principles are. As a result we get a lot of mixed quality research masquerading under the label experimental psychopathology, and this is something I’ve alluded to in other blog posts (e.g. “does a menu explain a restaurant?”).
Second, one important feature of experimental psychopathology is the creation of psychological models of mental health problems, and the development of such psychological models seems to have been significantly overshadowed in recent years by the bugeoning influence of neuroscience and genetics. One reason for this is that mental health research is still viewed primarily as a domain of medicine. The advent of DSM-5 has continued to fuel the misguided need to seek neurobiological markers for psychological disorders, and – most unfortunately – medics rarely, if ever, get training in how to develop or understand psychological models, and so inevitably undervalue them. The challenge here is to project Experimental Psychopathology as an empirical approach to knowledge that is equivalent to neuroscience and genetics, but is explaining psychopathology at a different level to the latter two. This, of course, means that neuroscience, genetics and psychological models are not mutually exclusive forms of explanation, but each complement each other to provide a rich understanding of mental health problems.
Third, much Experimental Psychopathology research is conducted on healthy individuals – often university undergraduates - and there are many ways in which this fact is used to project experimental psychopathology as a ‘second class’ contributor to mental health research. For example, many journals will now only publish psychopathology research conducted on individuals with a clinical diagnosis (e.g. “The Evils of Journal Scope Shrinkage”), and – perhaps quite reasonably – experimental psychopathologists are regularly urged to justify how their findings from nonclinical populations have relevance to clinical populations. It is this latter issue that needs both clarification, and a well-articulated set of validation rules. These validation rules are needed to describe quite clearly the pathway from basic psychological models developed with healthy individuals to models that have obvious relevance to the distinctive characteristics of clinical populations.
So what are the advantages of developing psychological models of mental health symptoms using an experimental psychopathology approach? Here are just a few:
- It permits the use of experimental models to mimic psychopathology processes in healthy individuals.
- It allows the study of clinically-relevant processes under highly controlled conditions.
- Controlled experiments provide evidence of causal relations between events that are a critical component of theory building.
- Experimental Psychopathology regularly borrows models and procedures from other areas of core psychological knowledge in order to understand mechanisms of psychopathology.
- Through controlled experimentation Experimental Psychopathology can help to identify the active ingredients in interventions developed directly from clinical practice.
- It allows the testing of psychopathology models in circumstances where doing so with clinical populations may be problematic.
These benefits can provide clinical psychology with a rigorous set of scientific principles for developing psychological models of psychopathology, and – perhaps more importantly – can help to prevent clinical psychology from re-inventing the wheel. What do I mean by this latter statement? Well, unfortunately many clinical psychology researchers tend to research psychopathology directly from their clinical experience, and without exploring the existing core knowledge available in the literatures on cognitive, social and biological psychology that may provide many of the answers to the theoretical questions they are asking.
Finally, the $64,000 question – how do we convincingly argue that our psychological models developed on healthy individuals have relevance to clinical populations? Well, start by reading this excellent paper by Bram Vervliet & Filip Raes published in Psychological Medicine in 2013. They define a number of different levels of external validity that can be used as criteria to bridge the translational gap between healthy individuals and clinical populations – and these range from weak criteria to relatively strong criteria. What also occurred to me is that you can use these various levels of external validity to construct a research pathway that begins with the development of simple psychological models or proof of concept studies and then moves with each step closer to the strict criteria that will determine whether those models have genuine relevance for clinical populations.
The pathway goes like this (from weak to strong criteria):
Step 1 – Face Validity: This is the degree of formalistic or phenomenological similarity between the behaviour in the laboratory model and the symptoms of the disorder. This is a relatively weak criterion for validity, and might simply represent the fact that the behaviour in the model simply “looks like” the behaviour in the disorder. Examples might include the induction of anxious or depressed moods in healthy participants and observing the effect of this on cognitive processes relevant to the psychopathology (e.g. Davey, Bickerstaffe & MacDonald, 2006; Hawksley & Davey, 2010)
Step 2 – Predictive Validity: This is the degree to which performance in the laboratory model predicts performance in the disorder. Basically, can you use the laboratory model to predict the performance or outcomes in your clinical population? This may require the collection of case histories to demonstrate that the processes specified by your model can be identified in the etiology of clinical cases (e.g. Davey, de Jong & Tallis, 1993), or the application of the model to clinical patients in experimental procedures to show that the same processes operate in the disorder as in the laboratory model (e.g. Chan, Davey & Brewin, 2013)
Step 3 – Construct Validity: This criterion seeks to show that the model developed in the laboratory can be compared favourably with existing clinical models of the disorder. It needs to be shown that the processes described in the laboratory model parallel the clinical process of interest, and that theories of the psychopathology and processes specified by the laboratory model allude to the same theoretical processes. For this criterion to be useful, there needs to be a well-elaborated laboratory model of the disorder as well as a valid clinical model of the disorder. In this case, elaborating and developing the laboratory model should be a valid means of extending and developing the clinical model. For example, in the 1980s and 1990s, there were many clinical models of anxiety disorders based on conditioning theory, and because we had an extensive knowledge of conditioning theory from animal learning we could use this knowledge to extend and test the clinical models (e.g. Davey, 1989; White & Davey, 1989).
Step 4 – Diagnostic Validity: Finally, it may be necessary to demonstrate that your laboratory model taps into processes that are unique to patients or are genuinely representative of clinical populations. I was slightly skeptical of this criterion to begin with because there is good reason to believe that many common psychopathologies are on a continuum of severity rather than being qualitatively different to sub-clinical forms of the disorders, so we wouldn’t necessarily expect processes to be ‘unique’ to individuals with a diagnosis. However, if your model has been developed on healthy individuals, this does raise the question of why everyone doesn’t exhibit symptoms of psychopathology? Perhaps everyone does – but in all probability clinical populations may possess characteristics or have had experiences that make them significantly more vulnerable to the important causal variables in your model. I now believe this is probably the final link required to validate your laboratory model, and is a criterion that would help to explain the jump in severity in symptoms between your healthy laboratory volunteers and individuals with a clinical diagnosis. We’ve begun to explore this possibility in our own mood-as-input model of perseverative psychopathologies, and examples of this can be found in Meeten & Davey, 2011, p1266-1269).
My preferred pathway for taking experimental psychopathology research from a simple laboratory model to a genuinely clinical-relevant theory is to plan your research to sequentially explore Steps 1, 2 and 4 in that order. In Step 1 you can use existing core psychological knowledge (together with with clinical experience) to develop your model on healthy individuals in the laboratory, in Step 2 you can test out whether the relevant processes in your model predict symptoms in clinical populations, and Step 4 allows you to clinically validate your model by identifying features in your model that might be representative of clinical populations and make them specifically vulnerable to more severe symptoms.
Perhaps principles such as these might form the beginnings of a ‘manual’ for Experimental Psychopathology. It would be a manual that would allow researchers to plan research programmes for psychological models of mental health that would be the intellectual and scientific equals of neuroscience and genetic approaches.