In order to learn novel actions, it is essential to observe and imitate similar actions performed by other individuals. Two experiments were conducted to investigate whether the participants bound together five meaningless postures during observation. The participants were divided into three groups according to the experimental conditions: the observation of postures, the imitation of postures, and the imitation of postures including the arm movements between subsequent postures, respectively. The participants' task was to memorize five sequentially-presented meaningless postures, after which the participants were either to recall the subsequent posture neighboring the one presented (Experiment 1), or to recognize and recall the same postures (Experiment 2). The participants were asked to discern the strategy judgments (visual or verbal encoding) they used for each trial. A number of joints functioning in the changes in the neighboring postures (1- and 4-step conditions) were manipulated. The results of Experiment 1 and 2 indicated that the correct recall rates (the binding effect) under the 1-step condition were higher than the 4-step condition in the imitation and the movement imitation groups, but that the binding effect was not to be found in the observation group. However, in both Experiment 1 and 2, binding effects were found in the correct recall rates in all three groups based on the visual encoding when the verbal encoding had been eliminated. The results suggest that observation elicits the binding between neighboring meaningless postures regardless of the imitation of body movement.