Premotor correlates of integrated feedback control for eye-head gaze shifts

Abstract

Simple activities like picking up the morning newspaper or catching a ball require finely coordinated movements of multiple body segments. How our brain readily achieves such kinematically complex yet remarkably precise multijoint movements remains a fundamental and unresolved question in neuroscience. Many prevailing theoretical frameworks ensure multijoint coordination by means of integrative feedback control. However, to date, it has proven both technically and conceptually difficult to determine whether the activity of motor circuits is consistent with integrated feedback coding. Here, we tested this proposal using coordinated eye-head gaze shifts as an example behavior. Individual neurons in the premotor network that command saccadic eye movements were recorded in monkeys trained to make voluntary eye-head gaze shifts. Head-movement feedback was experimentally controlled by unexpectedly and transiently altering the head trajectory midflight during a subset of movements. We found that the duration and dynamics of neuronal responses were appropriately updated following head perturbations to preserve global movement accuracy. Perturbation-induced increases in gaze shift durations were accompanied by equivalent changes in response durations so that neuronal activity remained tightly synchronized to gaze shift offset. In addition, the saccadic command signal was updated on-line in response to head perturbations applied during gaze shifts. Nearly instantaneous updating of responses, coupled with longer latency changes in overall discharge durations, indicated the convergence of at least two levels of feedback. We propose that this strategy is likely to have analogs in other motor systems and provides the flexibility required for fine-tuning goal-directed movements.

Figure 1.

Example gaze and head velocity (left) and position (right) trajectories made during control gaze (black traces) and gaze shifts during which a torque motor was used to generate head perturbations in the direction opposite to that of the ongoing gaze (gray traces). The interval over which the torque pulse was applied is denoted by the horizontal bars. Shaded vertical bars indicate the short duration of the resultant head perturbation. Following the head perturbation, head velocity resumed in the direction of the gaze shift. Inset, The head velocity evoked in response to the same perturbations when the head was stationary. The average response (black trace) is superimposed on traces from 40 individual trials.

Figure 2.

Head perturbation effects on gaze shift and neuronal responses. A, Matched control (black) and perturbed (gray) gaze shifts (see Materials and Methods). Vertical dashed line, gaze onset; shaded box, average increase in perturbed trials duration; open and filled arrows, average perturbation onset and time at which the resulting head perturbation reached maximum velocity, respectively. B, D, Example OPN activity recorded during control (black) and perturbed gaze shifts (gray). Traces were aligned on gaze shift onset and offset, respectively. C, E, Example EBN activity recorded during the control and perturbed trials shown in A. Traces were aligned on gaze shift onset and offset, respectively.

Figure 3.

Following perturbations, neural responses were updated on-line to remain tightly correlated with gaze shift duration. A, Correlation between movement and pause duration for the example OPN. Squares, circles, and stars indicate data from eye-only saccades, control, and perturbed gaze shifts, respectively. B, Correlation between movement and burst duration for the example EBN. Note: the pause and burst duration of OPNs and SBNs were well related to saccade duration in the head-restrained condition (mean slope, 1.06 ± 0.40, 0.97 ± 0.40, 1.04 ± 0.20; mean r = 0.78 ± 0.14, 0.79 ± 0.20, and 0.78 ± 0.21; for OPNs, EBNs, and IBNs, respectively) and to control gaze shift duration in the head-unrestrained condition (mean slope, 1.00 ± 0.30, 1.00 ± 0.14, and 1.00 ± 0.20; mean r = 0.78 ± 0.19, 0.74 ± 0.21, and 0.82 ± 0.20; for OPNs, EBNs, and IBNs, respectively).

Figure 4.

Distribution of UIs. We define UI as (mean increase in discharge duration)/(mean increase in movement duration), calculated for matched perturbed versus control trials; zero indicates no updating, and one indicates perfect updating. Average UIs are indicated for each panel. A, Distribution for OPNs. The vertical dotted line indicates the average UI. B, Distribution for IBNs. C, Distribution for EBNs. D, Combined distribution for all neurons. The different neuron types are shown using the color scheme from A–C.

Figure 5.

Changes in the instantaneous firing rates of SBNs mirrored gaze shift dynamics after head perturbations. Left, Example gaze velocity (A) and firing rate traces (B) for the example EBN during the control (black traces) and perturbed (gray traces) trials shown in Figure 1. Right, Average gaze velocity and firing rate traces computed from the control and perturbed traces shown to the left. Insets, Differences between the average gaze velocity and firing rate traces (black areas). The top insets illustrate the transient decrease in gaze velocity and firing rate immediately after perturbations, and bottom insets emphasize the overall increase in duration.

Figure 6.

Responses of the example EBN (A, B) and an example IBN (C, D) during control versus perturbed gaze shifts. Responses from individual trials (gray traces) and average responses (black traces) are superimposed, and dashed vertical lines denote the onset and offset of the gaze shift. B, D, Small arrows indicate the onset of the head velocity perturbation. Dashed and dotted lines indicate the best fit to the response of the unit before and after the perturbation, respectively. The gray vertical lines denote the interval between the onset of the perturbation and the intersection of the lines of best fit. The example neurons were typical in that their modulation was attenuated at very short latencies (3 and 2 ms) in response to applied head perturbations. A, C, Control data for which the timing of the head perturbation in perturbation trials (B, D) was used as a reference to produce the lines of best fit to the unit’s response. Inset, Time scales of the mean response and best line fits have been further expanded near the onset of the perturbation. E, Distribution of response latency across the population of SBNs. The vertical dotted line indicates the average response latency. Pre, Before perturbation; Post, after perturbation.

Figure 7.

SBNs dynamically encoded the same eye and head-movement-related modulations during control and perturbed gaze shifts. Predictions of the discharges of a neuron during control gaze shifts (A) and perturbed gaze shifts (B), obtained using an eye-based model (see Materials and Methods) estimated during eye-only saccades. Model predictions (thick curve) are shown superimposed on the firing rate (gray shaded area) for the example EBN. Downward arrows indicate time intervals at which the fits to the data were particularly poor. C, Model fits to the firing rate of the same neuron, obtained using a different model in which eye and head velocity coefficients were estimated (see Materials and Methods) from control gaze shifts. Note the marked improvement in goodness-of-fit. D, Very good model predictions were obtained when the model estimated during control trials (C) was applied to the firing rate of the neuron recorded during perturbed gaze shifts.