Encoding of movement in near extrapersonal space in primate area VIP

Abstract

Many neurons in the macaque ventral intraparietal area (VIP) are multimodal, i.e., they respond not only to visual but also to tactile, auditory and vestibular stimulation. Anatomical studies have shown distinct projections between area VIP and a region of premotor cortex controlling head movements. A specific function of area VIP could be to guide movements in order to head for and/or to avoid objects in near extrapersonal space. This behavioral role would require a consistent representation of visual motion within 3-D space and enhanced activity for nearby motion signals. Accordingly, in our present study we investigated whether neurons in area VIP are sensitive to moving visual stimuli containing depth signals from horizontal disparity. We recorded single unit activity from area VIP of two awake behaving monkeys (Macaca mulatta) fixating a central target on a projection screen. Sensitivity of neurons to horizontal disparity was assessed by presenting large field moving images (random dot fields) stereoscopically to the two eyes by means of LCD shutter goggles synchronized with the stimulus computer. During an individual trial, stimuli had one of seven different disparity values ranging from 3° uncrossed- (far) to 3° crossed- (near) disparity in 1° steps. Stimuli moved at constant speed in all simulated depth planes. Different disparity values were presented across trials in pseudo-randomized order. Sixty-one percent of the motion sensitive cells had a statistically significant selectivity for the horizontal disparity of the stimulus (p < 0.05, distribution free ANOVA). Seventy-five percent of them preferred crossed-disparity values, i.e., moving stimuli in near space, with the highest mean activity for the nearest stimulus. At the population level, preferred direction of visual stimulus motion was not affected by horizontal disparity. Thus, our findings are in agreement with the behavioral role of area VIP in the representation of movement in near extrapersonal space.

Keywords: VIP; disparity; macaque monkey; multisensory; parietal cortex; self-motion.

Figure 1

Graphical scheme of the experimental setup. Monkeys always fixated a target on the screen (blue axes). Moving stimuli were presented at seven different disparities, ranging from 3° crossed disparity (near space; negative values) to 3° uncrossed disparity (far space; positive values). These disparity values corresponded to simulated depths ranging from 27 cm to 229 cm in front of the monkey.

Figure 2

Experimental validation of the stereoscopic setup. The left and the right column depict data obtained with (left) and without (right) the LCD goggles being synchronized with the stimulus computer. Panels in the upper row (A and B) show mean discharges (±standard error) of neural activity for stimuli presented at a certain disparity (depth, only perceivable with the goggles being synchronized). Panels in the bottom row (C and D) show the monkey's oculomotor behavior during a sample trial each. In these panels, the ordinate depicts the mean eye position of both eyes while the abscissa depicts vergence. Neural activity strictly depended on horizontal disparity, while oculomotor behavior, which was perfectly centered on zero vergence, did not.

Figure 3

Disparity tuning: single cell level. The diagram shows the disparity tuning of a representative neuron (mean discharge ±standard error). The horizontal dotted lines indicate the mean discharge for stimuli presented at screen distance, i.e., zero disparity, and those disparities evoking the maximum (−3° disparity) and the minimum (+1° disparity) response. Activity induced by stimuli at zero disparity was normalized to 100%.

Figure 4

Disparity tuning: population level. The histogram in panel (A) shows the number of cases for which the peak discharge of a single cell was found at a certain disparity. The largest number (50) was found for the stimulus closest to the monkey: −3° disparity corresponding to 27 cm in front of the monkey. The smallest number (4) occurred for the stimulus furthest away, i.e., at +3° disparity, corresponding to 229 cm in front of the monkey. Panel (B) depicts the mean population activity (±standard error) taken from all 140 neurons with a significant response to visual stimulus motion and a significant tuning for horizontal disparity.

Figure 5

Normalized population discharge. In this population analysis, we first normalized maximum discharges of each neuron to a value of 1.0. The resulting disparity tuning curves of each neuron are shown as light grey lines. We then averaged these normalized response curves (solid black line). Error bars indicate the standard error. The resulting response curve was strongly monotonic with the strongest discharge for the closest stimulus.

Figure 6

Directional tuning as a function of disparity: single cell level. The panels in the top row shows the response (spike density function) to a complete cycle (corresponding to 1600 ms) of the circular pathway stimulus of the neuron whose average discharge for the preferred direction is shown in Figure 3. The sequence of stimulus directions is shown on the very right: right (R), down (D), left (L), up (U), and again right. The vertical lines indicate the interval which was considered to compute this average discharge. The bottom row shows the same data in polar plots. The preferred directions (PD) are computed based on vector averaging, i.e., weighting each motion direction with the neuron's discharge for this direction.

Figure 7

Average angular differences in directional tuning. The scatter plot shows for each neuron the average angular difference between preferred directions (PDs) of visual motion at different disparities. The red symbol indicates the value from the neuron whose disparity tuning was shown in Figures 3 and 6. The histogram on the right gives the distribution of these angular differences sorted in 45° bins. Seventy-five percent of the cells showed a mean angular difference of less than ±45°.

Figure 8

Directional tuning as a function of disparity: population level. The three-dimensional plot shows color coded the distribution of the angular differences between preferred directions (PDs) of visual motion at different disparities. We computed for each neuron on a pairwise basis the angular difference between PDs obtained at different disparities in case the tunings were significant (Rayleigh test, p < 0.05). Given the seven different disparity values, this resulted in a maximum of 21 pairwise comparisons per neuron, represented by the y-axis of the x-y-base of the cube. Values were sorted for each condition (1–21) into 45° bins (x-axis of the x-y-base). For all conditions, the angular differences clustered in the small value bins, i.e., around zero.

Figure 9

Angular differences between preferred directions obtained at different disparities. We computed for each neuron on a pairwise basis the angular difference between PDs obtained at different disparities in case the tunings were significant (Rayleigh test, p < 0.05). For this analysis, we averaged for each pairwise comparison the angular difference. The grand average as computed for the full population is shown color coded. None of the values was different from zero (Bonferroni correction, t-test, p > 0.3).