Anterior Intraparietal Area: A Hub in the Observed Manipulative Action Network

Abstract

Current knowledge regarding the processing of observed manipulative actions (OMAs) (e.g., grasping, dragging, or dropping) is limited to grasping and underlying neural circuitry remains controversial. Here, we addressed these issues by combining chronic neuronal recordings along the anteroposterior extent of monkeys' anterior intraparietal (AIP) area with tracer injections into the recorded sites. We found robust neural selectivity for 7 distinct OMAs, particularly in the posterior part of AIP (pAIP), where it was associated with motor coding of grip type and own-hand visual feedback. This cluster of functional properties appears to be specifically grounded in stronger direct connections of pAIP with the temporal regions of the ventral visual stream and the prefrontal cortex, as connections with skeletomotor related areas and regions of the dorsal visual stream exhibited opposite or no rostrocaudal gradients. Temporal and prefrontal areas may provide visual and contextual information relevant for manipulative action processing. These results revise existing models of the action observation network, suggesting that pAIP constitutes a parietal hub for routing information about OMA identity to the other nodes of the network.

Keywords: action observation; anatomical connectivity; macaque monkey; parietal cortex; visuomotor processing.

Figure 1.

Anatomical reconstruction and behavioral paradigms. (A) Reconstruction of the probes location along the intraparietal sulcus of Mk1 and Mk2. Vertical dashed lines indicate the position of each probe’s track, illustrated in the coronal sections below (a–d for Mk1 and a′–c′ for Mk2). Asterisk indicates the location of the 2 probes that have been considered together for the analysis of neuronal responses. Cgs, cingulate gyrus; cs, central sulcus; ias, inferior arcuate sulcus; ips, intraparietal sulcus; ls, lateral sulcus; lus, lunate sulcus; ps, principal sulcus; sas, superior arcuate sulcus; sts, superior temporal sulcus. (B) Observation task (OT). (C) Behavioral setup for the visuomotor task (VMT). Temporal sequence of task events is shown in Figure S1A. (D) Examples of initial (Epoch 1) and middle (Epoch 2) frames for each OMA exemplar. (E) Average speed of motion (degree/second) for the 4 variants of each OMA exemplar during the 2.6 s video presentation period.

Figure 2.

Single neuron examples, population activity and tuning properties of AIP units in the OT. (A) Percentage of OMA-selective, -nonselective, and –unresponsive units in each monkey. (B) Examples of facilitation (Neuron 1) and suppression (Neuron 2) OMA-selective neurons. Each neuron’s raster and peri-stimulus response is aligned to the video presentation (green triangles and dashed lines). Red triangles, reward delivery. (C) Time course of the net normalized population activity (including single- and multiunits) of OMA-selective facilitated (Left) and suppressed (right) units. The shading area around each line indicates 1 standard error, gray and light-blue shaded areas superimposed on each plot represent epochs 1 and 2 used for statistical analysis.

Figure 3.

Temporal dynamic of OMA processing in AIP. (A) Regression plot of preference index (PI) values calculated on OMA-selective unit activity during Epoch 1 and 2. See also Figure S2D. (B) Cross-validation of the ranking of all OMA exemplars performed with the average activity (±1 standard error) during Epoch 1 as a function of the same ranking performed with the average activity during Epoch 2 (Kruskal–Wallis, χ² = 139.34, P < 0.001). (C) Percentage of units selective for each OMA exemplar in each monkey during Epoch 2. (D) Classification accuracy of OMA exemplars as a function of test and training time. The superimposed white line represents the classification accuracy of the population along the diagonal (scale on the right). The red line in the lower part of the plot indicates the period of time during which the decoding accuracy is significantly above chance level (see Materials and Methods). (E) Cross-validation of the best OMA exemplar (rank = 1) calculated with the activity during Epoch 2 (E2) as a function of time during the entire action unfolding period (bin width 300 ms, step 20 ms). For each bin the color code (see inset) represents the local rank of the OMA ranking 1 in Epoch 2. Bins in which neural activity was not significantly different from baseline (sliding window ANOVA, bin width 300 ms, step 20 ms, P > 0.5 uncorrected) have been blanked out (E1 = Epoch 1).

Figure 4.

Relationship between motor selectivity for the grip type and visual selectivity for observed grasping actions. (A) Percentage of grip-selective (blue) and nonselective (red) units showing grip selectivity over time during the VMT performed in darkness. The dashed line below the plot indicates the time bins in which the relative number of tuned units was significantly different between the 2 subpopulations (sliding χ² test performed on 20 ms bins, P < 0.05, only sets of at least 5 contiguous bins are shown). (B) Percentage of grip-selective (blue) and nonselective (red) units (same as in panel A) showing OMA selectivity over time. All conventions as in panel A. The additional curves indicate the percentage of units with specific selectivity for observed grasping (sliding χ² tests, P < 0.05) among grip-selective (light blue) and nonselective (orange) units. (C, D) Heat maps of facilitation OMA-selective units with visual preference for grasping (C) or for OMA other than grasping (D) during the OT. Units in the heat maps have been ordered (from bottom to top) according to the timing of their peak activity after video presentation onset (vertical dashed line in the panels on the right). Superimposed on each heat map, the black lines in the left panels represent the percentage of units of the entire subpopulation showing significant tuning for grip type (sliding window one-way ANOVA, bin size 200 ms, step 20 ms, P < 0.05 uncorrected). In the right panels, the colored curves on the heat maps indicate the percentage of units in the subpopulation displaying preference for a specific OMA exemplar (i.e., the exemplar with highest activity value see color code in the legend; sliding window one-way ANOVA with 7 levels of the factor “OMA,” bin size 200 ms, step 20 ms, P < 0.05 uncorrected). Further analyses on the same set of data are provided in Figure S4. (E) Time course of the net normalized mean activity for each subpopulation illustrated in panels C and D during VMT (left) and OT (right). Shaded regions around each line represent 1 standard error.

Figure 5.

Rostrocaudal differences in neural selectivity for OMA, visually presented object and grip type along AIP. (A) Preference index for OMA during OT. (B) Preference index for the visually presented object during go trials of the VMT. (C) Preference index for the grip type during reaching/grasping execution (left column) and during object pulling (right column) in the dark (top) and in the light (bottom). (D) Preference index for hand visual feedback (HVF, light vs. dark) calculated from reaching/grasping (left column) and object pulling (right column) responses. Each set of data has been analyzed by means of a 2 × 3 factorial ANOVA (factors: monkey, position), **P < 0.001 for the factor position.

Figure 6.

Anatomical connectivity of the rostral, intermediate and caudal sectors of area AIP in the 2 monkeys. (A) Three dimensional anatomical reconstructions illustrating the distribution of labeled cells after injections at the different AIP levels of Mk1 (top) and Mk2 (bottom). For each monkey, connectivity maps are presented for the rostral (left), intermediate (center) and caudal (right) injections. To facilitate the comparison, the maps of Mk1 (right hemisphere) were flipped and shown as a left hemisphere and the mesial walls were shown as right hemispheres. The color scale indicates the relative density of labeled cells, counted within regions of 600 × 600μm², and expressed as a percentage of the maximum value obtained within the cortical surface for any given injection. Cgs, cingulate sulcus. Other abbreviations as in Figure 1. (B) Graphical representation of the strength of the main (>1%) connections of the rostral, intermediate and caudal sectors of AIP (data from corresponding injection positions in the 2 monkeys have been merged). The width of the wedges indicates the percentage of labeled neurons following injections at the various anteroposterior positions (see scale on the left). Acronyms: rIPL, rostral inferior parietal lobule; cIPL, caudal inferior parietal lobule; cIPS, caudal intraparietal sulcus; Par.Op, parietal operculum; SPL, superior parietal lobule; mIPS, medial intraparietal sulcus; vIPS, ventral intraparietal sulcus.

Figure 7.

Rostrocaudal gradients in AIP anatomical connectivity. Each bar represents the percentage of labeled cells observed in the various functional territories. Each territory is defined based on functional similarities of the labeled areas. The areas included in each cluster are listed under the histograms. The bars in each cluster are arranged in a rostral to caudal fashion relative to the location of the injection sites. Acronyms as in Figures 1 and 6.