Temporal interactions of air-puff-evoked blinks and saccadic eye movements: insights into motor preparation

Abstract

Following the initial, sensory response to stimulus presentation, activity in many saccade-related burst neurons along the oculomotor neuraxis is observed as a gradually increasing low-frequency discharge hypothesized to encode both timing and metrics of the impending eye movement. When the activity reaches an activation threshold level, these cells discharge a high-frequency burst, inhibit the pontine omnipause neurons (OPNs) and trigger a high-velocity eye movement known as saccade. We tested whether early cessation of OPN activity, prior to when it ordinarily pauses, acts to effectively lower the threshold and prematurely trigger a movement of modified metrics and/or dynamics. Relying on the observation that OPN discharge ceases during not only saccades but also blinks, air-puffs were delivered to one eye to evoke blinks as monkeys performed standard oculomotor tasks. We observed a linear relationship between blink and saccade onsets when the blink occurred shortly after the cue to initiate the movement but before the average reaction time. Blinks that preceded and overlapped with the cue increased saccade latency. Blinks evoked during the overlap period of the delayed saccade task, when target location is known but a saccade cannot be initiated for correct performance, failed to trigger saccades prematurely. Furthermore, when saccade and blink execution coincided temporally, the peak velocity of the eye movement was attenuated, and its initial velocity was correlated with its latency. Despite the perturbations, saccade accuracy was maintained across all blink times and task types. Collectively, these results support the notion that temporal features of the low-frequency activity encode aspects of a premotor command and imply that inhibition of OPNs alone is not sufficient to trigger saccades.

FIG. 1

Temporal traces of eye amplitude, eye velocity, and eyelid channel during control and air-puff trials for (A) step, (B) gap, and (C) delayed saccade tasks. Waveforms of eye and eyelid movements of each trial are color-coordinated. Eyelid signals are drawn in arbitrary units. Schematics of temporal evolution of behavioral tasks are shown at bottom. All traces are aligned on cue to initiate saccade, which is target onset for step and gap tasks and fixation point offset for delayed saccade task. Duration of saccade target and overlap period are arbitrarily schematized. See methods for details.

FIG. 2

Lawful variation in saccade latency as a function of blink time. Both measures are computed relative to time of saccade cue. Columns: data from individual datasets collected during (A) step, (B) gap, and (C) delayed saccade tasks. Rows: data from the 2 monkeys. Circles, rightward target; squares, leftward target. Dashed horizontal line drawn at saccade latency equal to 0.

FIG. 3

Change in saccade latency relative to control trials of matched target conditions plotted as a function of blink time. A: same dataset as shown in bottom left panel of Fig. 2. Dashed horizontal line drawn at 0 to distinguish trials for which saccade reaction time relative to control trials was reduced (negative) or prolonged (positive). The 4 visually identified clusters indicate the differential effects of blink time on saccade latency. B–D: linear regressions analysis applied to all datasets in both monkeys. Gray lines show best fits for individual datasets of (B) step, (C) gap, and (D) delayed saccade tasks. Thick black traces are means of individual fits. Details and statistics are provided in text and Table 1.

FIG. 4

Histogram comparing the transitional blink onset times for the step (black bars), gap (white bars), and delayed (gray bars) saccade tasks. Each bar represents an average value across datasets collected for each behavioral condition; error bars denote SD. A: latest blink time to increase saccade latency (rightmost points of cluster 2, Fig. 3A). Saccade onset was delayed for blink times as late as 150 ± 28, 7 ± 31, and 127 ± 20 ms after the initiation cue for step, gap, and delayed saccade tasks, respectively. B: earliest blink time to reduce saccade latency (leftmost points of cluster 3, Fig. 3A). Saccade onset was facilitated for blink times as early as 35 ± 53, −84 ± 50, and −84 ± 88 ms relative to the initiation cue for the step, gap, and delayed saccade tasks, respectively. C: Transition blink time after which blink onset reduces saccade latency in >50% of trials. Mean ± SD transition blink times were 105 ± 34, −25 ± 15, and 97 ± 27 ms across datasets of the step, gap, and delayed saccade tasks, respectively. *P < 0.01, ^#P < 0.05; 2-tailed t-test.

FIG. 5

Schematic representation of spatial profiles of control and blink-perturbed saccades. Saccade metrics were analyzed by 1st determining radial amplitude of each blink (R_p) and a median radial amplitude of selected control trials (R_c) to matched target conditions (see methods). Next, a radial error measure (e_r) was defined as R_p − R_c. A direction deviation (θ) parameter, defined as the difference in directions of puff and control trials, was also computed.

FIG. 6

Saccade amplitude plotted as a function of blink time for (A) step, (B) gap, and (C) delayed saccades. Same datasets and format as shown in Fig. 2.

FIG. 7

Quantitative comparison of blink-perturbed trials with control movements to the same target location. Radial error (A–C), direction deviation (D–F), and normalized peak velocity (G–I) are plotted as a function of blink time for step (left), gap (middle), and delayed saccades (right). Different colors represent the 4 clusters, as determined from latency analysis (see Fig. 3A). Saccades to rightward and leftward targets are not differentiated. Only 3 datasets, 1 of each behavioral task, are shown. Data from other datasets were similar (see Table 1).

FIG. 8

Interactions of velocity and latency of saccades. A: normalized peak velocity is plotted as a function of the difference between blink and saccade onset times. Panel shows that normalized peak velocity is significantly attenuated only when blinks and saccades coincide temporally (shaded region). B: velocity profiles of a subset of attenuated trials, aligned on onset (20°/s velocity criterion). Vertical dashed line denotes alignment time. Note that several traces exhibit a negative velocity profile prior to onset; it represents the eye movement component due to the blink that precedes saccade. C: reaction time of saccades in shaded region of A plotted against instantaneous velocity 5 ms after onset alignment in B.

FIG. 9

Minimal perturbation of saccade metrics as a function of blink time when saccade target can appear at 1 of 24 potential locations. Distribution of latency (A), normalized peak velocity (B), radial error (C), and direction deviation (D) are plotted as a function of blink time. Different symbols represent the 4 clusters, as determined from latency analysis (diamonds, cluster 1; squares, cluster 2; circles, cluster 3; triangles, cluster 4). Data across all target locations are pooled together. E: accuracy of blink-perturbed movements. Mean horizontal component of control (squares) and blink-perturbed (circles) trials are plotted against mean vertical component for each of 24 target locations. Bars in both directions denote SD. Data are pooled across all blink times.