Multimodal coding of three-dimensional rotation and translation in area MSTd: comparison of visual and vestibular selectivity

Abstract

Recent studies have shown that most neurons in the dorsal medial superior temporal area (MSTd) signal the direction of self-translation (i.e., heading) in response to both optic flow and inertial motion. Much less is currently known about the response properties of MSTd neurons during self-rotation. We have characterized the three-dimensional tuning of MSTd neurons while monkeys passively fixated a central, head-fixed target. Rotational stimuli were either presented using a motion platform or simulated visually using optic flow. Nearly all MSTd cells were significantly tuned for the direction of rotation in the absence of optic flow, with more neurons preferring roll than pitch or yaw rotations. The preferred rotation axis in response to optic flow was generally the opposite of that during physical rotation. This result differs sharply from our findings for translational motion, where approximately half of MSTd neurons have congruent visual and vestibular preferences. By testing a subset of neurons with combined visual and vestibular stimulation, we also show that the contributions of visual and vestibular cues to MSTd responses depend on the relative reliabilities of the two stimulus modalities. Previous studies of MSTd responses to motion in darkness have assumed a vestibular origin for the activity observed. We have directly verified this assumption by recording from MSTd neurons after bilateral labyrinthectomy. Selectivity for physical rotation and translation stimuli was eliminated after labyrinthectomy, whereas selectivity to optic flow was unaffected. Overall, the lack of MSTd neurons with congruent rotation tuning for visual and vestibular stimuli suggests that MSTd does not integrate these signals to produce a robust perception of self-rotation. Vestibular rotation signals in MSTd may instead be used to compensate for the confounding effects of rotatory head movements on optic flow.

Figure 1.

Schematic illustration and photographs of recording sites as shown in coronal sections. A, B, Right hemisphere of monkey Z. C, D, Left hemisphere of monkey Q. Thick rectangles in the schematic illustrations show the area of magnification in the corresponding photographs. Thin vertical lines in A and C indicate examples of electrode tracks, which can also be recognized in the photos (B, D, asterisks). Dashed lines in photos indicate gray matter/white matter boundaries and the border between MST and MT. The location of these borders is based on MRI data as analyzed by CARET software (Van Essen et al., 2001).

Figure 2.

Schematic of the 26 rotational and translational directions used to test MSTd neurons. A, Illustration of the 26 movement vectors, spaced 45° apart, in both azimuth and elevation. B, Top view: definition of azimuth. C, Side view: definition of elevation. Straight arrows illustrate the direction of movement in the translation protocol. Curved arrows illustrate the direction of rotation (according to the right-hand rule) about each of the movement vectors.

Figure 3.

Normalized population responses to visual and vestibular stimuli (thick gray and black curves) during rotation (A) and translation (B) are superimposed on the stimulus velocity and acceleration profiles (solid and dashed black lines). The dotted vertical lines illustrate the 1 s analysis interval used to calculate mean firing rates. Data are along the preferred direction of the cell and were normalized relative to the peak visual response.

Figure 4.

Examples of 3D direction tuning for an MSTd neuron tested during vestibular rotation (A), visual rotation (B), vestibular translation (C), and visual translation (D). Color contour maps show the mean firing rate as a function of azimuth and elevation angles. Each contour map shows the Lambert cylindrical equal-area projection of the original spherical data (see Materials and Methods). In this projection, the ordinate is a sinusoidally transformed version of elevation angle. The tuning curves along the margins of each color map illustrate mean ± SEM firing rates plotted as a function of either elevation or azimuth (averaged across azimuth or elevation, respectively).

Figure 5.

Summary of the differences in direction preference of MSTd neurons between the visual and vestibular conditions, plotted separately for rotation (left column; n = 113) and translation (right column; n = 167). A, B, Histograms of the absolute differences in 3D preferred directions between the visual and vestibular conditions (|Δ preferred direction|) for the rotation and translation protocols, respectively. Data are included only for neurons with significant 3D tuning in both stimulus conditions. C, D, Distributions of the differences in direction preference as projected onto each of the three cardinal planes: X–Z (front view), Y–Z (side view), and X–Y (top view). Note that the data from the 2D projections now cover a range of 360°.

Figure 6.

Distribution of 3D direction preferences of MSTd neurons, plotted separately for vestibular rotation (A; n = 127), visual rotation (B; n = 127), vestibular translation (C; n = 183), and visual translation (D; n = 307). Each data point in the scatter plot corresponds to the preferred azimuth (abscissa) and elevation (ordinate) of a single neuron with significant tuning (ANOVA, p < 0.05). Histograms along the top and right sides of each scatter plot show the marginal distributions. Also shown are 2D projections (front view, side view, and top view) of unit-length 3D preferred direction vectors (each radial line represents one neuron). The neuron in Figure 4 is represented as open circles in each panel.

Figure 7.

Summary of tuning strength and the differences in direction preference between rotation and translation, plotted separately for the vestibular (left column; n = 48) and visual (right column; n = 61) conditions. A, B, Scatter plots of the DDI during rotation and translation. Filled symbols indicate cells with significant tuning under both rotation and translation protocols (ANOVA, p < 0.05); open symbols denote cells without significant tuning under either one or both of the rotation and translation protocols (ANOVA, p > 0.05). C, D, Histograms of the absolute differences in 3D preferred direction (|Δ preferred direction|) between rotation and translation for the vestibular and visual conditions, respectively (calculated only for neurons with significant tuning in both conditions). E, F, Distributions of preferred direction differences as projected onto each of the three cardinal planes: X–Z (front view), Y–Z (side view), and X–Y plane (top view). G, H, The ratio of the lengths of the 2D and 3D preferred direction vectors is plotted as a function of the corresponding 2D projection of the difference in preferred direction (red, green, and blue circles for each of the X–Z, Y–Z, and X–Y planes, respectively).

Figure 8.

A–C, Examples of 3D rotation tuning for an MSTd neuron tested during both fixation and in darkness, illustrating vestibular rotation (A; fixation of a head-stationary target), visual rotation (B; fixation of a head-stationary target), and vestibular rotation in complete darkness (C). The format is as in Figure 4. D, Comparison of peristimulus time histograms from a single motion direction (azimuth, 0°; elevation, 0°) between the standard vestibular rotation condition (left) and rotation in darkness (right). Red curves indicate the time course of the motion stimulus.

Figure 9.

Population summary of vestibular rotation selectivity during fixation and in darkness. A, Distribution of the absolute difference in 3D preferred direction (|Δ preferred direction|) for the 23 neurons that had significant tuning under both fixation conditions and in complete darkness. B, C, Scatter plot of the maximum–minimum response amplitude (R_max–min) and DDI for the 34 cells tested during both fixation and in darkness. Symbols indicate neurons with (filled circles) and without (open circles) significant tuning during rotation in darkness (ANOVA, p < 0.05; all tested cells had significant tuning under the fixation condition).

Figure 10.

Comparison of tuning strength and response amplitude between visual and vestibular conditions before and after bilateral labyrinthectomy. A, B, Tuning strength is quantified by the DDI and is plotted separately for rotation and translation protocols. C, D, Scatter plots of response amplitude (R_max − R_min). Filled symbols indicate cells for which both visual and vestibular tuning was significant (ANOVA, p < 0.05); open symbols denote cells without significant vestibular tuning (ANOVA, p > 0.05). Histograms along the top and right sides of each scatter plot show the marginal distributions (including both open and filled symbols). Red symbols and bars denote data from labyrinthine-intact (normal) animals; blue symbols and bars denote data from labyrinthectomized animals.

Figure 11.

Examples of 3D rotation tuning for four MSTd neurons (A–D) tested during vestibular, visual, combined (100% visual coherence), and combined (35% visual coherence) conditions. The format is as in Figure 4.

Figure 12.

Summary of the differences in direction preference and comparison of tuning strength between the combined and each of the vestibular and visual conditions. A, B, Histograms of the absolute difference in 3D preferred directions (|Δ preferred direction|) between combined (100% coherence) and either the vestibular or the visual condition, respectively (n = 23). C, D, Scatter plots of the DDI for the combined (100% coherence) and either the vestibular or the visual condition, respectively (n = 25). E, F, Histograms of the |Δ preferred direction| between combined (35% coherence) and either the vestibular or the visual condition, respectively (n = 14). G, H, Scatter plots of the DDI for the combined (35% coherence) and either the vestibular or the visual condition, respectively (n = 16). Filled symbols indicate cells for which both the combined and vestibular (C, G) or visual (D, H) tuning was significant (ANOVA, p < 0.05). Open symbols denote cells for which either the combined or the vestibular (C, G)/visual (D, H) tuning was not significant (ANOVA, p > 0.05).

Figure 13.

Quantification of vestibular and visual contributions to the combined response for both the rotation and translation protocols. A, B, Relationship between vestibular gain (a₁) and visual gain (a₂). C, D, Gain ratio (a₁/a₂) plotted as a function of relative tuning strength between visual and vestibular responses (VVR). Number of neurons: n = 23 (A, C); n = 133 (B, D).