Torsional control of eye-head saccades

Under natural conditions the head and the eye are both free to
rotate about three mutuall orthogonal axes each (horizontal,
vertical and torsional). Theoretically, these six degrees of
freedom would allow any two-dimensional direction of the line of
sight to be obtained by infinitely many torsional head and eye
orientations. Yet, for any one gaze direction our brain chooses
specific angles of torsion for the head and the eye. For steady
fixation of distant targets with the head fixed and upright this
observation is known as Donders' law (1847). It has been shown to
hold independently of the direction of the rapid gaze shift
(saccade) preceding a fixation. Surprisingly, despite considerable
research on head and eye coordination the full implications of
Donders' law still have not been analyzed for head-unrestrained
gaze shifts. It has merely been studied whether torsional
constraints hold, when gaze is returned repeatedly to the targets
from single initial positions. The aim of this study was to see
whether Donders' law holds after head-unrestrained saccades,
independently of the saccade direction. Secondary objectives were
to analyze whether the neural controls of the eye and the head are
dependent or independent during this task and to collect and
present control data for comparison with patient recordings in
clinical context. Therefore, seven healthy human subjects made
large head-unrestrained gaze shifts to a single set of visual
targets during two separate conditions: 1) Repeated saccades to
individual target positions from the same direction respectively
(Star paradigm). 2) Repeated saccades to every target position from
several different directions (Diamond paradigm). Three-dimensional
orientations of head and eye were measured simultaneously with the
magnetic search coil technique and consecutively plotted in
three-dimensional space so that those orientations obeying Donders'
law formed a surface. For each of the three body units the static
orientations formed subspaces that resembled surfaces in the shape
of twisted double saddles. Surfaces of head orientations had the
most pronounced twist, eye in head surfaces were the most planar
and surfaces of gaze orientations showed intermediate twist. The
standard deviation of torsional residuals of the approximated
surfaces (torsional thickness) was bigger for gaze than for the eye
and smallest for the head. Head and eye torsion, as averaged over
individual fixations, were correlated differently within every
subject, but between subjects there was no correlation. In summary,
neither surface shapes nor torsional thickness of gaze, head or eye
differed between the two conditions (Star/Diamond). With this it is
shown for the first time that Donders' law of torsional control
holds true for gaze, head and eye orientations independently of the
direction of the preceding saccade. The absence of correlation
between head and eye torsion can be explained by independent
controllers of head and eye movements. This yields a new, further
argument supporting recent models of neuronal gaze control that are
based on the assumption of independent head and eye controllers.
Studies with patients carrying lesions in possible target
structures of such neuronal controllers are needed to further
investigate these models. Finally, clinically-diagnostic relevance
of this study arises from the comparison to results of studies on
gaze coordination after midbrain lesions where patients exhibit an
altered form of Donders' law.