There is provided an operator member displaceable, in response to operation by a human operator, relative to at least one displacement axis (more preferably, multi-dimensionally). Operating state of the operator member is detected and model operation information, indicative of an operating state to be taken by the operator member, is generated, so that reaction information is generated in accordance with a difference between the detected actual operating state and an operating state indicated by the model operation information. Reactive force is imparted to the operator member in accordance with the generated reaction information. In this way, a model operation amount can be informed to the human operator in the form of the reactive force. Further, by imparting separate reactive forces to a multi-axially operable operator member, appropriate information can be provided to the human operator during operation of the operator member, thereby achieving sensory feedback to the human operator.