Encoding of head acceleration in vestibular neurons I. Spatiotemporal response properties to linear acceleration

1. Extracellular recordings were made in and around the medial vestibular nuclei in decerebrated rats. Neurons were functionally identified according to their semicircular canal input on the basis of their responses to angular head rotations around the yaw, pitch, and roll head axes. Those cells responding to angular acceleration were classified as either horizontal semicircular canal-related (HC) or vertical semicircular canal-related (VC) neurons. The HC neurons were further characterized as either type I or type II, depending on the direction of rotation producing excitation. Cells that lacked a response to angular head acceleration, but exhibited sensitivity to a change in head position, were classified as purely otolith organ-related (OTO) neurons. All vestibular neurons were then tested for their response to sinusoidal linear translation in the horizontal head plane. 2. Convergence of macular and canal inputs onto central vestibular nuclei neurons occurred in 73% of the type I HC, 79% of the type II HC, and 86% of the VC neurons. Out of the 223 neurons identified as receiving macular input, 94 neurons were further studied, and their spatiotemporal response properties to sinusoidal stimulation with pure linear acceleration were quantified. Data were obtained from 33 type I HC, 22 type II HC, 22 VC, and 17 OTO neurons. 3. For each neuron the angle of the translational stimulus vector was varied by 15, 30, or 45° increments in the horizontal head plane. In all tested neurons, a direction of maximum sensitivity was identified. An interesting difference among neurons was their response to translation along the direction perpendicular to that that produced the maximum response ('null' direction). For the majority of neurons tested, it was possible to evoke a nonzero response during stimulation along the null direction. The cells that displayed nonzero sensitivity along the null direction always had response phases that varied as a function of stimulus direction. 4. These spatiotemporal response properties were quantified in two independent ways. First, the data were evaluated on the basis of the traditional one- dimensional principle governed by the 'cosine gain rule' and constant response phase at different stimulus orientations. Second, the response gain and phase values that were empirically determined for each orientation of the applied linear stimulus vector were fitted on the basis of a newly developed formalism that treats neuronal responses as exhibiting two-dimensional spatial sensitivity. Thus two response vectors were determined for each neuron on the basis of its response gain and phase at different stimulus directions in the horizontal head plane. The largest response vector (S(max)) is referred to as the maximum sensitivity vector of the neuron, whereas the smallest response vector (S(min)) represents the minimum sensitivity vector in the horizontal plane. The latter (S(min)) is spatially perpendicular and temporally leads the maximum sensitivity vector by 90°. 5. These two methods, i.e., the traditional one-dimensional approach and the newly developed two-dimensional treatment, were compared to determine which provided a more satisfactory quantitative description of the gain and phase dependence on stimulus orientation. With the use of mean square error (MSE) for a relative comparison of the 'goodness' of fit, the two-dimensional treatment provided fits with MSE values that were statistically significantly lower than those obtained from the traditional one-dimensional fits. Thus two response vectors (S(max) and S(min)) were used to describe quantitatively the spatiotemporal response properties of the sampled vestibular nuclei neurons. 6. The tuning ratio, defined as the quotient of the magnitudes of S(min) over that of S(max), was used as a measure of the two-dimensional sensitivity of vestibular nuclei neurons to linear acceleration stimulation. Neurons with tuning ratios >0.1 were classified as broadly tuned or two-dimensional; Cells with tuning ratios <0.1 were referred to as narrowly tuned or one- dimensional. A total of 56/72 (78%) of the examined cells were thus identified as broadly tuned at 0.6 Hz.