According to Einstein's equivalence principle, inertial accelerations durin
g translational motion are physically indistinguishable from gravitational
accelerations experienced during tilting movements. Nevertheless, despite a
mbiguous sensory representation of motion in primary otolith afferents, pri
mate oculomotor responses are appropriately compensatory for the correct tr
anslational component of the head movement. The neural computational strate
gies used by the brain to discriminate the two and to reliably detect trans
lational motion were investigated in the primate vestibule-ocular system. T
he experimental protocols consisted of either lateral translations, roll ti
lts, or combined translation-tilt paradigms. Results using both steady-stat
e sinusoidal and transient motion profiles in darkness or near target viewi
ng demonstrated that semicircular canal signals are necessary sensory cues
for the discrimination between different sources of linear acceleration. Wh
en the semicircular canals were inactivated, horizontal eye movements (appr
opriate for translational motion) could no longer be correlated with head t
ranslation, instead, translational eye movements totally reflected the erro
neous primary otolith afferent signals and were correlated with the resulta
nt acceleration, regardless of whether it resulted from translation or tilt
. Therefore, at least for frequencies in which the vestibule-ocular reflex
is important for gaze stabilization (>0.1 Hz), the oculomotor system discri
minates between head translation and tilt primarily by sensory integration
mechanisms rather than frequency segregation of otolith afferent informatio
n. Nonlinear neural computational schemes are proposed in which not only li
near acceleration information from the otolith receptors but also angular v
elocity signals from the semicircular canals are simultaneously used by the
brain to correctly estimate the source of linear acceleration and to elici
t appropriate oculomotor responses.