Tool use and the distalization of the end-effector

In the study of imitation (Dautenhahn & Nehaniv, 2002), the correspondence problem for the imitator is to decide how to match end-effectors of the imitatee’s body to its own. We have brains and bodies that allow us to master the use of varied tools, but that rests, for most of us, on early training from caregivers that enables us to learn how to attend to a range of necessary affordances and master a set of basic uses of our hands and body, providing the effectivities which are dual to the affordances provided by perception (Turvey, 1992; Turvey et al., 1981; Zukow-Goldring & Arbib, 2007) This allows us to “get the idea” of novel tools by a mix of verbal instruction, complex imitation, mechanical reasoning (Goldenberg & Hagmann, 1998), and trial-and-error—the latter being crucial as we hone skill in using a tool through extended practice with it.

Unfortunately, the data do not support ready discrimination between F4 and F5 cells.

Note that even in grasping, the hand can be reconfigured to provide different end-effectors, e.g., the pads of thumb and index finger in a precision pinch, the palm and inner finger surfaces in a power grasp. So the hypothesis that F5 neurons encode the motion of the end-effector in extrinsic coordinates is still consistent with these new data and previous research in F5 (Rizzolatti et al., 1988; Raos et al., 2006; Umilta et al., 2007) that relate neuron activity to grasp type and phase and wrist orientation.

An informal description of ACQ was published in Arbib and Bonaiuto (2008), an extended description of the model with further simulation results will appear in Bonaiuto and Arbib (2009).