There was no difference between dogs that had had previous exposure to the mirror and those that had not, as per owner reports (Fisher exact test, 2-sided, P > 0.60). Latency to obtain the treat among dogs in the experimental group also differed significantly from the control group dogs, with dogs in the experimental group fi nding the treat faster than controls, exp(B) ¼ 2.46; df ¼ 1; SE ¼ 0.42; P ¼ 0.03. Among only dogs that found the treat, there was no significant difference in latency between experimental and control groups (Cox proportional hazards regression, P ¼ 0.59).