Kp. Krommenhoek et Jam. Vangisbergen, EVIDENCE FOR NONRETINAL FEEDBACK IN COMBINED VERSION-VERGENCE EYE-MOVEMENTS, Experimental Brain Research, 102(1), 1994, pp. 95-109
Recently, a quantitative model for the generation of rapid eye movemen
ts in direction and depth was proposed. In this scheme, the saccadic a
nd the vergence system share a common initiation system and are contro
lled by local feedback loops based on efference copy signals. We have
used a remembered-target double-step paradigm to test the idea that bo
th subsystems are guided by extraretinal signals. The subject was inst
ructed to move the binocular point of fixation to the remembered posit
ions indicated by a double-step movement of the target, in direction a
nd depth. Since both binocular refixations were made in complete darkn
ess, correct execution of this task requires information about both th
e stored visual coordinates of the final target and the coordinates of
the first movement. Binocular eye movements from five subjects were c
ompared with predictions from two feed-forward models and a feedback m
odel. Analysis of the pooled direction data showed that the feedback m
odel performed best and fitted well. Qualitatively the same result was
obtained in the vergence component, but in this case the goodness of
fit was considerably less. These results, confirmed in each individual
subject, show that the saccadic and vergence subsystem can use nonret
inal information about a prior movement in direction and depth. Furthe
r analysis showed that the gain of the direction response of the secon
d movement was, on average, roughly correct. By contrast, the vergence
component of these responses was only about 60% of the required ampli
tude. Since the fit procedure gave the same weighting factors to the s
econd target and to the first movement, we propose that the low vergen
ce gain reflects mechanisms operating after the calculation of the mot
or error signal, possibly at the execution stage. Finally, we discuss
the possibility of a central control stage keeping track of the ongoin
g movement sequence, based on a comparison of desired and current eye
position signals.