Biomedical Engineering Reference
In-Depth Information
The model predictions for all saccades match displacement data and estimates of velocity
very well, including saccades with a dynamic or a glissadic overshoot, with accuracy similar
to those in Figure 13.52.
The 8
saccade shown in Figure 13.52a of data andmodel predictions has dynamic overshoot.
Note that the saccade with dynamic overshoot is caused by a PIRB firing in the antagonist neu-
ral input at approximately 220 ms. The PIRB induces prominent reverse peak velocity as shown.
Figure 13.52b shows model predictions and data for an 8
saccade with glissadic over-
shoot. The glissade is caused by the PIRB in the antagonist neural input at approximately
223 ms. Notice the peak firing for a saccade with glissadic overshoot is smaller than one
with dynamic overshoot. The PIRB induces reverse peak velocity that is smaller than the
one with dynamic overshoot. In glissadic overshoot, the eye has an overshoot that returns
to steady state more gradually. As a result, the glissade has a smaller peak velocity.
A -12
normal saccade is shown in Figure 13.52c. Normal saccades usually do not have a
PIRB, although this is not absolute, since the timing of the PIRB might offset the impact of
the burst.
The main sequence diagram is shown in Figure 13.53. Peak-velocity estimates from the
model are in close agreement with the data estimates of peak velocity and follow an expo-
nential shape as a function of saccade magnitude. Duration has a linear relationship with
saccade magnitude for saccades above 7
. For saccades between 3 to 7
, duration is approx-
imately constant. It should be noted that saccade duration is difficult to determine, espe-
cially for small saccades, and may be a source of differences with other published data.
The latent period is relatively independent of saccade magnitude.
The estimated agonist pulse magnitudes and durations are shown in Figure 13.54 for all
127 saccades. The agonist pulse magnitude does not significantly increase as a function of
saccade magnitude for saccades larger than 7
, consistent with the time-optimal controller
proposed by Enderle [11], Zhou and coworkers [50], and Enderle and Zhou [18]. For sac-
cades under 7
, agonist pulse magnitude shows a linear increase in pulse magnitude versus
saccade magnitude, again in agreement with our theory for the saccade controller. A great
variability is observed in the pulse magnitude estimates for saccades of the same magni-
tude, which is also observed by Hu and coworkers [34] in their analysis of the firing rates
in the monkey EBN. The agonist pulse duration increases as a function of saccade magni-
tude for saccades larger than 7
. For saccades between 3 to 7
, the duration of the agonist
pulse is relatively constant as a function of saccade magnitude. Note that for all saccades,
the pulse magnitude is tightly coordinated with the pulse duration.
13.8.5 Postinhibitory Rebound Burst and Postsaccade Phenomena
Inhibition of antagonist burst neurons
3
is postulated to cause an unplanned postinhibi-
tory rebound burst (PIRB) toward the end of a saccade that causes dynamic overshoots or
glissades [11]. While some studies do not observe the rebound firing in the abducens
neurons in monkeys [24, 38, 45], PIRB are observed in the abducens motoneurons at the
3
The neurons describe in this section are fully described in Section 14.9, including a neural network. Only a
brief description of the neurons involved in PIRB is given here.
