Neuronal encoding of sound direction in the auditory midbrain of the rainbow trout

Acoustical stimulation causes displacement of the sensory hair cells relative to the otoliths of the fish inner ear. The swimbladder, transforming the acoustical pressure component into displacement, also contributes to the displacement of the hair cells. Together, this (generally) yields elliptical displacement orbits. Alternative mechanisms of fish directional hearing are proposed by the phase model, which requires a temporal neuronal code, and by the orbit model, which requires a spike density code. We investigated whether the directional selective response of auditory neurons in the midbrain torus semicircularis (TS; homologous to the inferior colliculus) is based on spike density and/or temporal encoding. Rainbow trout were mounted on top of a vibrating table that was driven in the horizontal plane to simulate sound source direction. Rectilinear and elliptical (or circular) motion was applied at 172 Hz. Generally, responses to rectilinear and elliptical/circular stimuli (irrespective of direction of revolution) were the same. The response of auditory neurons was either directionally selective (DS units, n = 85) or not (non-DS units, n = 106). The average spontaneous discharge rate of DS units was less than that of non-DS units. Most DS units (70%) had spontaneous activities < 1 spike per second. Response latencies (mode at 18 ms) were similar for both types of units. The response of DS units is transient (19%), sustained (34%), or mixed (47%). The response of 75% of the DS units synchronized to stimulus frequency, whereas just 23% of the non-DS responses did. Synchronized responses were measured at stimulus amplitudes as low as 0.5 nm (at 172 Hz), which is much lower than for auditory neurons in the medulla of the trout, suggesting strong convergence of VIIIth nerve input. The instant of firing of 42% of the units was independent of stimulus direction (shift <15 degrees), but for the other units, a direction dependent phase shift was observed. In the medial TS spatial tuning of DS units is in the rostrocaudal direction, whereas in the lateral TS all preferred directions are present. On average, medial DS units have a broader directional selectivity range, are less often synchronized, and show a smaller shift of the instant of firing as a function of stimulus direction than lateral DS units. DS response characteristics are discussed in relation to different hypotheses. We conclude that the results are more in favor of the phase model.