Internal models underlying grasp can be additively combined.

Our ability to additively combine two learned internal models was investigated by studying the forces people generate when lifting objects with a precision grip. Subjects were required to alternately lift two objects of identical physical appearance but differing weight. Grip force scaling prior to lift-off was used to estimate the output of the internal model associated with each object. Appropriate internal models were formed when alternately lifting two objects of different weight. The objects were then combined by stacking them one upon the other, and the combined object was lifted. Results show that subjects can additively combine internal models of object dynamics but the sum is biased by a default estimate of the object's weight.