Chapter 2. Prehension

Chapter 5 - Movement Before Contact 189
the ballpark concept, the Movefast Schema does not have to put the
hand at this location; instead, it puts the hand within the vicinity of this
location. The actual mechanism for driving the hand can be done using
one of the models described in this chapter, such as the VITE model
(Bullock & Grossberg, 1989) or Jordan’s inverse kinematic model
(Jordan, 1988).
One question is what is the right ballpark, or, more formally, what
is the size of the buffer ill? A pilot study, where one subject was
asked to pick up a light, horizontally resting wooden dowel between
his thumb and index finger pad, offers some evidence (Iberall, 1987a;
Iberall & MacKenzie, 1988). Measurements were taken of virtual
finger lengths and angles at the beginning of the movement, at the time
of the peak aperture between thumb and index finger, and at the time
when the dowel was grasped. For differing dowel orientations, the
results showed that VF1 and VF2 parameters at peak aperture were
within the ballpark of their values at contact, but there seemed to be
tighter constraints on VF1 than on VF2 at peak aperture. This
supports the view that the thumb moves less than the index finger
(Wing & Fraser, 1983, Wing et al., 1986).
In terms of the use of sensory information, it was noted that when
subjects are deprived of visual feedback or do not have enough time to
process it, more errors are made and larger peak apertures are formed.
This suggests a larger ‘ballpark’. With a larger initial buffer factor,
the subjects avoid accidental contact, which is more likely to occur
when visual information is not available or there is not enough time to
update one’s internal model, as was observed by (Wing et al, 1986).
From the reaching perspective (the transport component), a two phase
movement helps to ensure success despite perceptual errors in locating
the object. In terms of grasping an object (the grasping component), a
two phase movement with ballpark positioning of the fingers allows
errors in perceiving object characteristics such as shape.
Other models of the preshaping and enclosing phases have been
suggested. In reevaluating the CCP, Hoff and Arbib (in press)
developed a new model that used expected duration for the
coordination of the preshaping and enclosing schemas. Noting that
the time it takes to enclose the hand seems to be invariant in
perturbation studies, about 200 ms (Paulignan, Jeannerod,
MacKenzie, & Marteniuk, 1991; Paulignan, MacKenzie, Marteniuk,
& Jeannerod, 1991; Gentilucci et al., 1992), Hoff and Arbib call the
model the Constant Enclose Time model (see Figure 5.30). The
Transport Arm Schema takes as input the target distance and it com-
putes how much it needs. At the same time, the Preshape Schema re-