Chapter 2. Prehension

8 WHAT IS PREHENSION?
movement. Dominating the early work in this field, the physiologist
Sir Charles Sherrington (1906) coined the term proprioception for the
collective position sense stemming from sensors in the muscles,
joints, and tendons. Ultimately, this work has led to questions such
as: in what frame of reference do movements occur; how does
automatic regulation of movement occur, what happens when different
parts of the brain are lesioned; how are multiarticulate movements
coordinated; and what are the dynamical characteristics of the limb.
Today, these fields blend as tools and techniques for measurement and
analyses become more sophisticated, and the complexity of the human
body becomes realized.
In a similar vein, engineers, mathematicians, and computer and
cognitive scientists design various forms of computational architec-
tures that may allow a simulation of the mapping between inputs and
outputs within the black box of Figure 1.2. These simulations have
been designed using computational neural networks, artificial intelli-
gence programming techniques, expert systems and even implemented
mechanically in robotic systems. Struggling to develop robot con-
trollers for systems working in unknown environments, roboticists
address many of the same problems faced by researchers in motor be-
havior: how can a computer system coordinate multiple degree of free-
dom limbs as well as the brain does; how can sensors be designed and
multiple modalities integrated; how is sensory information integrated
with movement; how does stable grasping occur, how does movement
occur in spite of the complex forces acting on the system; how are ob-
jects perceived; and how does planning occur.
Marr (198 1) suggested three levels for analyzing systems that per-
form complex information processing. These are the task level, the
representation or algorithmic level, and the implementation level. An
example to demonstrate the distinctions is derived from flying. While
both airplanes and birds fly, the algorithms they use are different: air-
planes use jet propulsion while birds flap their wings. At the imple-
mentation level, airplanes use jets made out of materials such as steel,
while birds use muscles and feathers. We will make similar distinc-
tions in terms of prehension.
The close relationship between computational and experimental re-
searchers is seen in Figure 1.3. Advances in science unfold through
the interaction between theory and data. To start with, conceptual
models are important for understanding complex systems and act to
suggest experimental and computational models. Conceptual models
are the theory part of science. They begin on a piece of paper (or a
napkin!), and might leave out the details of how they might be imple-