Intrinsic interest and extrinsic reward: The effects of a continuing token program on continuing nonconstrained preference