An animal model is an animal, or an experimentally defined population of animals, used to investigate a biological process relevant to another setting, often human physiology or disease. Its usefulness depends on the specific mechanism and measured outcome. Sharing a visible phenotype does not establish that every cause or response is shared.
For example, experiments in rats tested whether a compound enhanced fear extinction by measuring fear-potentiated startle after repeated presentations of a previously threatening cue. The defined behavioural protocol made a learning mechanism experimentally accessible.[ref]
Facilitation of conditioned fear extinction by systemic administration or intra-amygdala infusions of D-cycloserine as assessed with fear-potentiated startle in rats by David L. Walker and others. 2002, abstract and conditioning/extinction experiments
This establishes an effect on a rat learning assay; whether the same intervention improves a human clinical outcome is a further question. A good animal model makes the proposed correspondence explicit enough to test. New to topics? Read the docs here!