If you work in a discipline such as toxicology, psychology, or neurology, you may well find yourself in the position of having to decide whether you will harm animals in research. Those who decide to perform harmful animal research are either doing it for the benefit of the actual animals used, as in veterinary work with sick or injured animals, or have been able to find for themselves an answer to the question: what gives us the right to inflict pain and suffering on innocent creatures? The justifications offered for harming animals are almost always consequentialist in form; by harming a small number of animals in as limited a way as possible, goes the argument, we can produce great benefits for humans and other animals. In this chapter we will explore this utilitarian way of thinking.

According to the US Department of Agriculture, 1,134,693 vertebrate animals were used in research in the USA in 2010 (Animal and Plant Health Inspection Service 2011). Of that number, 71,317 were nonhuman primates and roughly 85,000 were cats or dogs. More than 300,000 animals were used in experiments involving pain that was treated with drugs, and almost 100,000 were caused pain with no drugs administered.

By all accounts, these figures vastly underestimate the actual numbers of animals used because the Animal Welfare Act does not require researchers to report figures on the most commonly used species: rats, mice, birds, fish, and frogs. According to Larry Carbone, the number of rodents alone used in 2001 was 80 million (Carbone 2004). (The numbers are also a fraction of the total numbers of animals killed for food and fiber in the USA, which the US Department of Agriculture estimates at 35 million cows, 110 million hogs, and 8.8 billion chickens (United States Department of Agriculture 2011).) The commonly accepted rationale for using animals in biomedical research is to advance basic knowledge of human disease and function and so to improve human life without having to subject humans to the experiments being run on the animals.