Because the cart of experimentation has been put before the horse of knowledge, scientists routinely end up not only inadvertently harming animals, but unknowingly executing flawed experiments bound to yield inconsistent, and thus ultimately useless, results.
Animals, unlike objects, have emotions. Mice, rats, birds, or apes kept under one set of conditions will react differently to the same experimental stimuli than will mice, rats, birds, or apes kept under another set of conditions. Only now, however, are we coming to realize how incredibly sensitive experimental animals are to differential experimental environments, handlers, and procedures. The implications of this sensitivity have radical implications for every experiment done on animals.
A couple of real life examples, both taken from Bernard Rollin's insightful book Animal Rights and Human Morality, highlight the problem. The first involves mice and the experience of shock. In order to gain insight into the human experience of shock, scientists have long traumatized mice and studied their "microcirculatory shock profile."