Of mice and men: on the validity of animal models of psychiatric disease


As biomedical researchers, we use animal models as a compromise. We hope to understand human disorders and improve human health, but the experiments we do are often too risky for human subjects. One largely unspoken concern about this compromise is the degree to which these animals’ behaviors accurately model the disorder in question. What do we even mean when we say that a particular rodent behavior “models” a human syndrome? And why is it that, very often, treatments that work in animal models fail once they reach the clinical setting (1)?

There is an extensive literature in psychology on the various ways to assess the validity of tests and models (2), and the biomedical research community would do well to consider this long philosophical struggle. But, as a behavioral ecologist and ethologist, there seems to be one potential gold-standard question for animal models that is rarely, if ever, discussed. Are apparent similarities between the human and the animal behavior driven by homology,  or are they analogies, driven  by convergent evolution?

Analogy vs. Homology

As I see it, one major flaw in the design of animal models is in mistaking analogy for homology. That is, neuroscientists often study an animal’s behavior because it resembles an interesting human behavior. Take, for example, mouse models of obsessive-compulsive disorder. The goal is not to understand why some mice groom too much, but instead to understand why some humans wash their hands too much. Mouse grooming is an analogy for hand washing. These studies are only useful, then, if mouse grooming and human hand-washing rely on the same neural circuitry. For these studies to be meaningful, the two behaviors must be homologous.

What does it mean to be homologous?


Homology means evolved from the same ancestral structure or behavior. If, for example, you wanted to understand the structure of bat wings, but could not get the permits to study bats, you could reasonably study bird wings as a model. You could also study human arms, or even whale flippers. The only reason such studies would be useful is that bat wings, bird wings, human arms, and whale flippers have very similar, evolutionarily homologous, structures (see figure). Even though whale flippers are not used for flight (“And the rest, after a sudden wet thud, was silence…”), their structure can tell you a lot about how bat wings are likely put together.

An analogous behavior or structure, on the other hand, is one that looks similar across species but likely occurs for different reasons or through entirely different mechanisms. A bat wing and a butterfly wing are analogous—while they look similar, and evolved to promote the same behavior, they are evolutionarily and structurally distinct. Attempting to learn about the skeletal structure of bats’ wings by studying butterflies would be a largely fruitless endeavor.

The difficulty, of course, in studying psychiatric disease is that most psychiatric diseases are defined by a cluster of symptoms—not by an underlying physiological process. For the researcher, this means that it is challenging to know whether you are studying the right physiological process at all. If a particular assay, based originally on analogy, repeatedly fails to translate in clinical trials—for example, if social behavior assays in mouse autism, or over-grooming in mouse OCD, or refusing to swim in mouse depression repeatedly let clinicians down—perhaps we, as a community, should consider this potential reason why.


  1. http://www.plosmedicine.org/article/info%3Adoi%2F10.1371%2Fjournal.pmed.1000245
  2. Messick, S. (1989). Meaning and Values in Test Validation: The Science and Ethics of Assessment. Educational Researcher, 18, 5-11.