The crossed-oscillators hypothesis for 3D steering during undulatory
swimming
explains how a set of motoneuronal
output patterns are generated from the neuronal circuitry.
Corresponding experimental data, i.e. multiple EMG-recordings
during various steering maneuvers, are however not readily available.
What is available is recordings of the actual body movements during
such maneuvers. One way to test the crossed-oscillators hypothesis
would therefore be to simulate also the resulting mechanical
movements.
Using the 3D mechanical simulator SAROS
, we have recently been able to conduct simulations
of this sort. The body and muscle model was taken and adapted from
our earlier 2D simulations
. The mechanical model
was extended to allow for free movements in 3D. Also the simplified
model for calculating the drag forces had to be extended to be used in
the 3D setting.
The simulations confirm that the motor patterns generated by the
crossed-oscillators model are indeed sufficient to produce the desired
steering movements
(Ekeberg, Lansner, Grillner 1995).
Lateral (yaw) turns are made by exaggerating the contractions on one
side of the body. Pitch turns are the result of a slight up- or
downward bend of the body. Rolls result from a rotating activity
pattern giving the body a narrow spiral shape.

The results from these simulations have been presented in video sequences by animating the body movements using a sequence of ray-traced imaged (see figure). The continuation of this work may include incorporating the vestibular feedback signals known to be present in brainstem neurons. These neurons presumably serve as "command neurons" that initiate corrective maneuvers to restore the upright body position.
Next