Only domain-specific imitation practice makes imitation perfect

Does imitation involve specialized mechanisms or general-unspecialized-learning processes? To address this question, preschoolers (3- and 4-year-olds) were assigned to one of four “practice” groups. Before and after the practice phases, each group was tested on a novel Spatial Imitation sequence. During the practice phase, children in the Spatial Imitation group practiced jointly attending, vicariously encoding, and copying the novel spatial sequences. In the Item Imitation group, children practiced jointly attending, vicariously encoding, and copying novel item sequences. In the Trial-and-Error group, children practiced encoding and recalling a series of novel spatial sequences entirely through individual (operant) learning. In the Free Play (no practice) control group, children played a touchscreen drawing game that controlled for practice time on the touchscreen and mirrored some of the same actions and responses used in the experimental conditions. Results of the difference between pre- and post-practice effects on novel spatial imitation sequences showed that only the Spatial Imitation practice group significantly improved relative to the Free Play group. Individual Spatial Trial-and-Error practice did not significantly improve spatial imitation. The effect of Item Imitation practice was intermediate. These results are inconsistent with the hypothesis that general processes alone--or primarily--support imitation learning and is more consistent with a mosaic model that posits an additive-interaction-effect on imitation performance where a more general social cognitive mechanism (i.e., natural pedagogy) gathers the relevant information from the demonstration and another more specialized mechanism (i.e., imitation specific) transforms that information into a matching response.