Choice-Related Activity during Visual Slant Discrimination in Macaque CIP But Not V3A

Abstract

Creating three-dimensional (3D) representations of the world from two-dimensional retinal images is fundamental to visually guided behaviors including reaching and grasping. A critical component of this process is determining the 3D orientation of objects. Previous studies have shown that neurons in the caudal intraparietal area (CIP) of the macaque monkey represent 3D planar surface orientation (i.e., slant and tilt). Here we compare the responses of neurons in areas V3A (which is implicated in 3D visual processing and precedes CIP in the visual hierarchy) and CIP to 3D-oriented planar surfaces. We then examine whether activity in these areas correlates with perception during a fine slant discrimination task in which the monkeys report if the top of a surface is slanted toward or away from them. Although we find that V3A and CIP neurons show similar sensitivity to planar surface orientation, significant choice-related activity during the slant discrimination task is rare in V3A but prominent in CIP. These results implicate both V3A and CIP in the representation of 3D surface orientation, and suggest a functional dissociation between the areas based on slant-related choice signals.

Keywords: 3D orientation; CIP; V3A; macaque; perception; slant.

Figure 1.

Surface orientation tuning. A, The 3D orientation of a planar surface can be described by two variables, slant and tilt. Tilt specifies the axis within the frontoparallel plane about which the plane is rotated, and slant specifies how much it is rotated. These variables define a polar coordinate system. Only a subset of the stimuli used in the study are shown. B, C, Distributions of the SODI for 396 CIP (B) and 60 V3A (C) neurons. Open bars denote tuned neurons (215 CIP and 44 V3A), and filled bars denote untuned neurons (181 CIP and 16 V3A). D, Equal area projection (Rosenberg et al., 2013) showing joint distribution of preferred slants and tilts for the 215 tuned CIP neurons (black circles) and 44 tuned V3A neurons (gray circles).

Figure 2.

Slant discrimination task and behavioral performance. A, Temporal sequence of events in the slant discrimination task. Each trial began with the presentation of a fixation point at the center of the screen. The monkey fixated this point for 300 ms after which an RDS plane was presented for 1000 ms while fixation was maintained. The monkey then reported which direction the plane was slanted away from frontoparallel by making a saccade to the upper target if the slant was positive (top-far) or the lower target if the slant was negative (top-near). B, Side view of the task illustrating positive versus negative slants. Solid lines depict planes centered at the fixation depth (screen distance, ∼32.5 cm). Dashed and dotted lines depict planes centered at either near or far depths (2.25 cm in front of or behind the display), respectively. C, Discrimination behavior plotted as the proportion of top-far choices as a function of slant. Data are fit with a cumulative Gaussian for each depth (N = 450 trials/data point). D, The P.S.E. as a function of depth for each monkey. For comparison, the gray line shows the expected dependency of the P.S.E. on stimulus depth if the task was performed based on local absolute disparities rather than slant. The line reaches ±14° at ±2.25 cm but is clipped at ±2° to not obscure the data.

Figure 3.

Surface orientation tuning of example CIP and V3A neurons. A, B, Slant–tilt tuning profiles of representative CIP (A) and V3A (B) neurons. Firing rate is color coded with red hues indicating larger firing rates. The peak of the CIP tuning profile is in the lower right corner, indicating that the cell responded best to a planar surface with the lower right corner closest to the monkey. The peak of the V3A tuning profile is in the upper portion of the plot, indicating that the cell responded best to a planar surface with the top closest to the monkey. White dashed lines correspond to the 90°/270° tilt axis along which the slant discrimination task was performed. C, D, Slant tuning curves of the same CIP (C) and V3A (D) neurons measured during the slant discrimination task. Error bars denote SEM. E, F, Neuronal response distributions for three pairs of slant angles (±20°, ±5°, ±1.25°) for the CIP (E) and V3A (F) neurons. Negative slants are shown as black bars and positive slants are shown as white bars.

Figure 4.

Comparison of behavioral and neuronal sensitivity. A, B, The proportion of top-far choices made during the recordings of the CIP (A) and V3A (B) neurons from Figure 3 are plotted as a function of slant (+ symbols). Simultaneously measured neuronal responses were converted into neurometric functions using ROC analysis and the proportion of top-far choices of an ideal observer are plotted as a function of slant (* symbols). Dashed and solid curves show cumulative Gaussian fits to the psychometric and neurometric functions, respectively. C, D, Gray curves show cumulative Gaussian fits to the neurometric functions of each neuron recorded during the slant discrimination task. Black symbols and curves show average neurometric functions across animals and neurons. Error bars denote SEM. E, F, Behavioral and neuronal thresholds are compared for all individual experiments for monkeys N (triangles), P (circles), and Z (squares) for CIP (E) and V3A (F). Neuronal thresholds are multiplied by $\sqrt{2}$. Diagonal histograms show distributions of neuronal to behavioral threshold ratios. Triangles above the histograms mark median threshold ratios.

Figure 5.

Summary of choice-related activity in CIP and V3A. A, B, Distribution of firing rates for example CIP (A) and V3A (B) neurons (same as in Figs. 3, 4A,B) in response to the ambiguous 0° slant stimulus. Responses are sorted according to whether the monkey made a top-near (black) or top-far (white) choice. For the CIP neuron, the choice-related difference in responses yielded a choice probability significantly different from chance (grand CP = 0.65, p = 0.001). For the V3A neuron, there was no significant choice-related difference in responses (grand CP = 0.50, p = 0.36). C, D, Histograms of grand choice probabilities for all 65 CIP (C) and 23 V3A (D) neurons. Gray bars denote CPs that are significantly different from the chance value of 0.50 (p < 0.05, permutation test). Mean CPs are marked by triangles. E, F, Choice probability as a function of neuronal threshold (multiplied by $\sqrt{2}$). There is a significant negative correlation between CP and neuronal threshold in CIP (E) and no significant correlation between CP and neuronal threshold in V3A (F). Solid lines show linear fits and dashed lines show 95% confidence intervals for the slope. Filled symbols denote CPs significantly different from chance (0.50, p < 0.05, permutation test). Different symbols correspond to different animals.

Figure 6.

Example CIP and V3A neurons illustrating the effect of choice on slant tuning. For each neuron, the black curve shows the slant tuning curve created by averaging responses regardless of choice. Orange and purple curves show choice-conditioned slant tuning curves created by separating responses into top-far vs top-near choices, respectively. The SPC, CPC, and CP are listed for each neuron. A, CIP neuron with a positive SPC, a positive CPC, and a CP > 0.50 (p = 0.001). B, CIP neuron with a positive SPC, a negative CPC, and a CP < 0.50 (p = 0.001). C, V3A neuron with a negative SPC, a negative CPC, and a CP > 0.50 (p = 0.29).

Figure 7.

Partial correlation analysis showing relationships between slant partial correlation, choice partial correlation, and CP. Choice partial correlation is plotted as a function of slant partial correlation with individual neurons color coded to indicate CP. Significant CPs are filled, nonsignificant CPs are open. A, B, Data are shown for 65 CIP (A) and 23 V3A (B) neurons. Curves show 95% confidence ellipses fit to data points with CP > 0.50 (green dashed) or CP < 0.50 (blue solid). A, In CIP, as indicated by the oblique orientations of the 95% confidence ellipses, CPs > 0.50 (greener) tended to occur when the slant and choice partial correlations had the same sign (quadrants I and III), whereas CPs < 0.50 (bluer) tended to occur when the slant and choice partial correlations had opposite signs (quadrants II and IV). B, For V3A, choice-related activity was weak, as indicated by the elongated but horizontally oriented 95% confidence ellipses.

Figure 8.

Time courses of choice probability, neuronal threshold, and partial correlations. A, B, Mean values of CP for CIP (A) and V3A (B) neurons as a function of time relative to stimulus onset. C, D, Mean neuronal thresholds (multiplied by $\sqrt{2}$) for CIP (C) and V3A (D) as a function of time. E, F, Mean SDFs (blue) as well as squared slant (SPC, black) and choice (CPC, red) partial correlations for CIP (E) and V3A (F) as a function of time. In all plots, analysis bins are 200 ms in duration, shifted every 50 ms starting at 100 ms. Each point is plotted in the center of the 200 ms time bin. Error bars denote SEM. Vertical dashed lines in A, B, E, and F mark the end of the stimulus presentation. The last time bin is centered at 1150 ms, and thus extends approximately until the median choice time (1271 ms after stimulus onset).