Social Familiarity As A Predictor Of Observational Learning In The Domestic Dog (Canis familiaris)

Experiments utilizing observational learning in dogs are becoming more frequent, but few investigate the familiarity relationship between model and observer. Two experiments were designed to explore the familiarity relationship between conspecifics and heterospecifics. The first experiment utilized a heterospecific-conspecific training dynamic and the conditions of the experiment varied how familiar (familiar or unfamiliar) observing dogs were to the heterospecific and conspecific models. Forty-eight dogs were assigned to four conditions varying in familiarity of the human demonstrator and model dog. A separate control condition (n=12) were trained with operant techniques and did not receive demonstration or familiarity variables. Analysis of data concludes that effects of learning were not due to observational learning, rather, dogs were learning more efficiently due to familiarity relationships. More in-depth analysis reveals that dogs only learned due to the familiarity relationship with the human demonstrator. The conspecific demonstrator did not influence performance according to the analysis of variance. The second experiment replicated apparatus from another experiment (Pongracz, Miklosi, Kubinyi, Gurobi, Topal, Csyani, 2001) to determine if demonstration by a socially familiar or unfamiliar demonstrator (human or conspecific) influenced learning rates in a V-shaped fence task (n=48). Dogs did not show learning rates that differed from the control condition (n=12). Two tasks were given where dogs had a more difficult task (n=35) or an easier task (n=30). While significant task differences were exhibited, regression analysis reveals that these are not to due familiarity effects. Dogs did appear to utilize observational learning in the more difficult task, but individual learning took precedence in the easier task.