Excerpt: In 1975, Hans Krebs coined the “August Krogh Principle,” based on the Danish physiologist’s insight that “For a large number of problems there will some animal of choice, or a few such animals, on which it can be most conveniently studied” (Krogh, 1929, p. 247; Krebs, 1975). Krogh, in fact, goes one step further, offering, somewhat facetiously, that some animals almost seemed to be “created” for the express purpose of certain physiological investigations (p. 247). While the use of what we would now call experimental organisms has a long history (e.g., Galvani’s use of frogs in the study of animal electricity in the 18th century), it was largely during the 20th century that a relatively small subset of these animals began to be systematically employed throughout biology, psychology, and neuroscience in the manner that Krogh suggests (Leonelli & Ankeny, 2013). The impact of these organisms, in both basic science and applied biomedical contexts, would be difficult to overstate. One hardly needs to scour the archives at Nature to grasp the importance of the fruit fly (Drosophila melanogaster) to genetics, rat (Rattus norvegicus) to behavioral psychology, or mouse (Mus musculus) to the study of a myriad of human diseases (Morgan et al., 1923; Donaldson, 1909; Gould et al., 2015; Sullivan et al., 2020). In biomedical research in particular, animal models are taken to be the most important tool in the researcher’s toolbox (LaFollette & Shanks, 1996, p. 108).

While the practice of experimenting on other species to learn something about our own is widespread, how animal models are taken to accomplish this task is still a matter of some debate (Levy & Currie, 2015; Leonelli & Ankeny, 2011; Schaffner, 1998). One common suggestion found in both the philosophical and scientific literature is that animal models function as surrogates for human beings (Germain, 2014; Committee on New and Emerging Models in Biomedical and Behavioral Research, 1998). Surrogates are, in essence, stand-ins for other systems (e.g., human beings) that for epistemic, practical, or moral reasons, we cannot directly intervene on. Although details differ between accounts, most surrogacy views, as Germain (2014) calls them, take analogy to play a key role in guiding the extrapolation from model system (e.g., an animal model of disease) to target (e.g., a human patient). Unfortunately, the theory of analogy animating the majority of these positions is outdated and leaves much to be desired as a reliable method for generating sound inferences from model to target. Additionally, the majority of these views do little to engage with the contemporary psychological and cognitive science literature on analogical reasoning, including research on analogical reasoning in science specifically. The resulting conception of analogy often rests on a minimally constrained and shallow notion of similarity, which makes any model that relies on such reasoning an easy target for criticism.

My goal in this paper is to develop an account of animal models as surrogates that is both philosophically rich and psychologically plausible. To do this, I begin with an exploration of how animal models might function as surrogates for human populations. I consider three possibilities for the kind of relationship that would need to hold between a surrogate and target — simple similarity, homology, and analogy — before settling on analogy as the most suitable option. Next, I take on the theory of analogy often encountered in the literature and argue that such an account leaves the surrogacy view on unsteady footing, vulnerable to what is often called the disanalogy argument. I offer an alternative theory of analogy, based on research from psychology and cognitive science, which I argue better fits the canonical description of surrogative reasoning first offered by Swoyer (1991) and is able to meet the challenge offered by the disanalogy argument. I use this framework to suggest a narrower understanding of what qualifies as a disanalogy and provide two strategies for mitigating legitimate disanalogies when researchers encounter them. Finally, I close by considering some interesting implications of this view, such as the proper role of homology as part of the an analogical conception of surrogacy.