OCD Research in the Lab: A Research Pathway for Psychological Models Graham C L Davey University of Sussex, UK

Purpose To make explicit the need to address the external validity of developed models of OCD To provide a clear programme of research to extend research on healthy individuals to diagnostic populations To recommend a closer reciprocal relationship between models of OCD and existing core psychological knowledge

What is Experimental Psychopathology The study of psychopathology processes under highly controlled conditions To develop detailed models of psychopathology EP is often carried out on healthy humans and nonhuman animals Has been significant in furthering animal models of psychopathology and pharmacological treatments

Neglected Issues in Experimental Psychopathology A Pathway of EP Research in developing psychological models of psychopathology is unclear Biological and psychological models have often developed in parallel rather than collaboratively Psychological models should offer a broader perspective by integrating cognitive, behavioural & social factors into single etiological models

Problems in the Elaboration of Psychological Models Psychological models have often developed chaotically and independently of each other Psychological models are often driven by clinical experience rather than core psychological knowledge There is a preponderance of esoteric constructs – especially in OCD research The role of research on healthy individuals has been consistently undervalued There are fewer and fewer outlets for EP research on healthy individuals EP researchers have often been negligent in ensuring the validity of their research to clinical populations

Bridging the Theoretical Gap Between Laboratory Studies & Clinical Populations No need to question the research skills of well-trained empirical researchers It is external validity that is often overlooked Using external validity criteria to validate translational research

External Validity Criteria Validity Criterion Definition Face Validity The phenomenological similarity between the behaviour in the laboratory model and the symptoms of the disorder Predictive Validity Performance in the laboratory model predicts performance in the disorder Construct Validity The model developed in the laboratory can be compared favourably with existing clinical models of the disorder. The processes described in the laboratory model parallel the clinical processes of interest Aetiological Validity It can be shown that the aetiologies in the laboratory model and the disorder are identical (this is very similar to construct validity) Convergent Validity The degree to which outcomes from the laboratory model correlate with measures/outcomes from other models of the same disorder/construct Discriminant Validity A model differs from other models of the same disorder to the extent that it’s outcomes are different to those predicted by other models Diagnostic Validity Demonstrating that the laboratory model taps into processes that are unique to the clinical population exhibiting the disorder

Face Validity Weakest Criterion but Often a Starting Point Formalistic similarity between the behaviour in the lab and the symptoms in the disorder Measuring ‘symptoms’ in an analogue task (e.g. checking)

Predictive Validity Your laboratory model predicts behaviour in the disorder Manipulating critical variables in your model and showing they affect symptoms (e.g. manipulating ‘inflated responsibility’) BUT – is the behaviour in your model and the behaviour in the disorder generated by similar mechanisms?

Construct Validity Requires that the laboratory model compares favourably with a clinical model of the disorder Causal processes in your model can also be identified when those processes are examined in the clinical population Processes in the laboratory model can also be identified in clinical populations through etiological case histories (e.g. Davey, de Jong & Tallis, 1993) Benefit of the laboratory model is that it can be used to explore further implications of the model for the disorder

Diagnostic Validity Requires that the researcher should be able to demonstrate that the laboratory model taps into processes or characteristics that are unique to the clinical population (Vervliet & Raes, 2013) Clinical populations may possess characteristics that make them highly vulnerable to the model Extends the model to identifying psychological markers of vulnerability for psychopathology

A Research Pathway for Experimental Psychopathology Does your model superficially look like the clinical phenomenon? Does your model predict what happens in the clinical phenomenon? Can you show that the processes in your model closely resemble the processes in the clinical phenomenon? Does your model explain why the relevant clinical population is differentially vulnerable to the disorder?

A 3-Stage Research Pathway for Experimental Psychopathology Face Validity Predictive/Construct Validity Diagnostic Validity Validity Questions “Does your model produce measurable behavioural/physiological/cognitive outcomes that resemble the clinical phenomenon?” “Does your model predict what happens in the clinical phenomenon?” “Does your model explain why the relevant clinical population is differentially vulnerable to the clinical disorder?” Participant Mainly healthy participants Healthy participants or clinical populations Clinical populations or healthy participants

Stage 1 The adaptation of core knowledge from other areas of psychology to explain the clinical phenomenon The use of basic experiments under highly controlled conditions to identify causal relationships To use experimental procedures to ‘infer’ the existence of processes that cannot be directly observed

Mood-as-Input & Perseverative Checking Current mood is used as information in decisions to terminate or continue a checking task Mood-as-input is a psychological mechanism first developed in the social psychology literature Can be applied to any perseverative task

MacDonald & Davey (2005)

Stage 1 Proof of concept stage

Stage 1 as the ‘Proof of Concept’ Stage - Disgust & Religiosity

Stage 2 Testing predictions of the model against knowledge of what happens in the clinical population Testing clinical participants to determine whether the variables in the model are relevant in causing or maintaining symptoms Examining case histories for evidence of the critical processes in the model Investigating aetiologies of clinical populations using self-report questions and surveys

Lopatka & Rachman (1995)

Stage 3 What is it about the relevant clinical population that makes them differentially vulnerable to the disorder? Literature Reviews Meta-analyses Evidence that shows the clinical population is significantly more reactive or sensitive to the model’s processes than healthy control participants Comparing healthy individuals sub-clinical populations who score high on symptom measures Manipulating the critical variables in healthy participants to a level where they resemble those in the clinical population (especially if the disorder is a dimensional one such as OCD)

Susceptibility to Mood as Information Effects in the Anxiety Disorders Vulnerable individuals have poor problem-solving confidence and decision-making processes Mood is used as information when the individual is aware that it’s a relevant feature of the current task (as in attempts to repair anxious mood) Anxiety inflicts a high cognitive load and depletes working memory capacity – in such situations mood is more likely to be used as information

Implications of the Research Pathway You can start at any of the 3 stages! It would in principle be possible to answer validity questions at all three stages using entirely healthy individuals Research at each stage is valid in it’s own right Journal Editors & Funding Bodies should NOT: ask researchers at Stage 1 to provide evidence that their model is relevant to clinical populations Insist that research is only clinically valid if it is done on clinical populations