Neuropsychiatric disorders such as schizophrenia, depression and bipolar disorder are serious health problems. They have substantial negative affects on a significant subset of the population and are still largely not understood. While the molecular targets of many psychotherapeutic drugs have been successfully reverse engineered, this was done in the 1960's. Despite ongoing efforts to further understand these disorders, little progress has been made since then. This raises the question: why? Two scientists, Eric J Nestler and Steven E Hyman have published an article in the journal Nature Neuroscience seeking to find a solution to this. In their paper, Animal models of neuropsychiatric disorders, they claim that the primary thing holding back research in the field is the difficulty of creating animal models of human psychiatric diagnoses. The authors then seek to contribute to the field by doing an analysis of the currently used models and discussing which ones are most likely to be valid and productive.
It can be very difficult to translate between animal and human thoughts and feelings. Whenever one does so they risk unfairly anthropomorphizing. Since animals are incapable of reporting their feelings researchers need to find round-about ways to determine what is actually going on within an animal's head. The typical methodology then used to study animal behavior and use it as a proxy for mental activity. However for most of the neuropsychiatric disorders that are professionally addressed what constitutes a legitimate disorder is not clearly separated from what constitutes normal variation. Furthermore, the same neuropsychiatric diagnose can be proscribed to two completely different sets of symptoms. This leaves researchers in a position where they must decide for themselves what constitutes a legitimate disorder, how to define it, and the subsequently how it can be represented in behavioral models.
In order to usefully discuss the efficacy of studies relating to these neuropsychiatric disorders, the authors of the article referred to a framework for validating studies with the components construct, face and predictive validity. Construct validity is a measure of how well a model's construction is relevant to a disease. Face validity is a measure of how well a model reiterates the physical and behavioral features of a human disease. Predictive validity is a measure of how well a model's response to treatments compares to patients actual responses to these same treatments.
The article then discusses different things that can be modeled in schizophrenia, depression and bipolar disorder and the validities of potential models. For schizophrenia, it is stated that blunted affect, asocial behavior, diminished motivation and deficits in working memory and/or conscious control of behavior are all symptoms that a behavioral model should seek to measure. The article claims that a good way to creat constructively valid models is to use genetic animal models with highly penetrant human mutations, although it doesn't consider these models to be perfect. It also states that a good (but not sufficient on its own) measure of face validity is a deficit in prepulse inhibition (PPI) of a phenomenon where weak starting stimuli reduce a startle response produced by a following more intense stimulus.
For depression it's stated that neurovegetative symptoms such as abnormalities in sleep, appetite, weight and energy along with psychomotor agitation or retardation are all potential indicators. With the caveat that no abnormality has proven sufficiently robust or consistent enough to validate an animal model the paper claims that chornic social defeat stresses along with chronic mild and chronic unpredictable stress are all capable of inducing states of depression which have some face value. These methodologies are criticized however as potentially setting off an anxiety disorder with similar symptoms instead of modeling depression. The authors suggest that measures of other homeostatic symptoms such as alterations in sleep, circadian rhythms and feeding with attendant metabolic parameters would strengthen claims of depression in animal models.
For bipolar disorder it's stated that the diagnosis comes from periods of mania with or without depression. The article states that transgenic mice have exhibited manic-like behavior when they were programmed to over express glycogen synthase kinase-3beta. These mutants are assesd to meet partial criteria for face validity along with predictive validity. However they failed to meet requirements for contruct validity. The article suggests that mania investigative studies use a broad range of behavioral tests and interpret their data cautiously.
Finally, the article listed some generalized recommendations towards researchers. These recommendations included listing the specific aspects of the illness meant to be model and stating the types of validators applied to the model. The researcher also noted that construct validity is most compelling of the different validities and that it's best to use a broad range of behavioral assays.
It's clear that research into these neuropsychiatric disorders still faces a great deal of hurdles, especially when it comes to assessing bipolar disorder. However, as this paper shows, there is constructive focus being brought to the forefront of this area. With genetic and technological advances combined applied to models with clearly stated rationales and sober discussion of validity significant progress can potentially be made in the field.