We argue that consistent formulation of optimal sensing and control must include information terms, yielding an extension of the standard POMDP setting. To make the standard reward/costs terms consist