Journal of Neurophysiology

Blinks slow memory-guided saccades

Alice S. Powers, Michele A. Basso, Craig Evinger

Abstract

Memory-guided saccades are slower than visually guided saccades. The usual explanation for this slowing is that the absence of a visual drive reduces the discharge of neurons in the superior colliculus. We tested a related hypothesis: that the slowing of memory-guided saccades was due also to the more frequent occurrence of gaze-evoked blinks with memory-guided saccades compared with visually guided saccades. We recorded gaze-evoked blinks in three monkeys while they performed visually guided and memory-guided saccades and compared the kinematics of the different saccade types with and without blinks. Gaze-evoked blinks were more common during memory-guided saccades than during visually guided saccades, and the well-established relationship between peak and average velocity for saccades was disrupted by blinking. The occurrence of gaze-evoked blinks was associated with a greater slowing of memory-guided saccades compared with visually guided saccades. Likewise, when blinks were absent, the peak velocity of visually guided saccades was only slightly higher than that of memory-guided saccades. Our results reveal interactions between circuits generating saccades and blink-evoked eye movements. The interaction leads to increased curvature of saccade trajectories and a corresponding decrease in saccade velocity. Consistent with this interpretation, the amount of saccade curvature and slowing increased with gaze-evoked blink amplitude. Thus, although the absence of vision decreases the velocity of memory-guided saccades relative to visually guided saccades somewhat, the cooccurrence of gaze-evoked blinks produces the majority of slowing for memory-guided saccades.

  • main sequence
  • orbicularis oculi
  • eye movements
  • lid movements
  • superior colliculus

saccadic eye movements are rapid movements of the eyes that reorient the line of sight. They are characterized by their increase in speed with increasing amplitude (Becker 1989; Westheimer 1954). The main sequence, the relationship between saccade amplitude and peak velocity (Bahill et al. 1975), and the relationship between peak and average saccade velocity (Becker 1989; Evinger et al. 1981), are powerful measures that can assess abnormalities in the brain stem circuitry controlling saccade generation (Jürgens et al. 1981; Schmid-Burgk et al. 1982).

Saccadic eye movements are driven by a variety of cues. In the laboratory, a common approach is to use visual stimuli located in the peripheral visual field as targets for saccades, visually guided saccades. If the target appears in the periphery only transiently, and the saccade is made at a later time when the target is no longer visible, then the saccade is referred to as memory guided (Hikosaka and Wurtz 1983). Visually guided and memory-guided saccades exhibit different main sequence functions, in spite of the fact that they share midbrain and brain stem circuits for their generation (Moschovakis et al. 1996; Sparks and Mays 1990). In general, saccades guided by memory are slower than saccades guided by vision (Gnadt and Andersen 1988; Hikosaka and Wurtz 1983; Smit et al. 1987; White et al. 1994). One hypothesis for this difference is that the lack of visual drive to the superior colliculus reduces the activity of neurons within the superior colliculus, which, in turn, leads to slower saccadic velocities. Support for this idea comes from data showing that transient reductions in superior colliculus activity with either lidocaine or muscimol reduce saccadic velocity (Basso et al. 2000; Hikosaka and Wurtz 1985; Lee et al. 1988). Similarly, saccades made to locations without visual stimuli are associated with a reduced discharge of action potentials from superior colliculus neurons compared with saccades made to locations with visual stimuli (Edelman and Goldberg 2001).

Another possible reason for the slowing of memory-guided saccades is that a blink is more likely to occur with a memory- than a visually guided saccade. Just as reflex blinks slow saccadic eye movements by superimposing a “down and in” movement on the ongoing saccadic eye movement (Goossens and Van Opstal 2000a; Goossens and Van Opstal 2000b; Rambold et al. 2004; Rottach et al. 1998), a blink associated with a saccade, a gaze-evoked blink, may slow the saccadic eye movement that it accompanies. Consistent with this hypothesis, gaze-evoked blinks are less likely to occur in the presence of visual stimuli that are behaviorally significant (Evinger et al. 1994; Williamson et al. 2005). For example, gaze-evoked blinks rarely occur when monkeys make saccades to visual targets for a reward, but frequently occur during unrewarded saccades after the trial (Powers et al. 2006). Here, we test the hypothesis that the occurrence of blinks associated with saccades is the dominant reason for the slowing of memory-guided saccades. We measured saccades guided by vision and guided by memory in three monkeys and compared the main sequence and curvature of saccades with and without blinks. We found that the slowing of saccades as well as the curvature of saccades increased as the amplitude of the gaze-evoked blink increased. We conclude that the reduced velocity of memory-guided saccades results primarily from the occurrence of a gaze-evoked blink.

METHODS

Physiological procedures.

Three monkeys (Macaca mulatta) were implanted with eye loops for monitoring eye movements and cranial cylinders for electrophysiological recording of single neurons, as described previously (Kim and Basso 2010; Li and Basso 2011). These animals participated in electrophysiological experiments ongoing in the laboratory. After initial sedation with ketamine (5.0–15.0 mg/kg im), monkeys received atropine to reduce salivation (0.5 mg/kg) and were intubated and maintained under isoflurane anesthesia for the duration of the procedure. A subconjunctival eye loop was implanted on the right side of the monkeys (Judge et al. 1980). The skull was exposed, and a plastic headholder for restraint and three cylinders for microelectrode recording were secured to it with titanium screws and dental acrylic. One day before the surgery and for 4 days afterward, Cefadroxil, an antibiotic (25 mg/kg), was given. Postsurgical analgesia was provided as needed for 72 h by the administration of buprenorphine (0.01–0.03 mg/kg) and Flunixin (1–2 mg/kg). One to two weeks of postoperative recovery occurred before behavioral testing. All experimental protocols were approved by the Institutional Animal Care and Use Committee and complied with or exceeded standards set by the Public Health Service policy on the humane care and use of laboratory animals.

General behavioral procedures.

We used a real-time experimental data acquisition and visual stimulus generation system (Tempo and VideoSync; Reflective Computing, St. Louis, MO) to create the behavioral paradigms and acquire eye position data. A digital light projector (LP335; Infocus, Wilsonville, OR) with a native resolution of 1,024 × 768 and operating at 60 Hz was used to project visual stimuli on a tangent screen at a distance of 51 cm. The luminance of the background was 0.28 cd/m2. The visual stimuli were controlled by VideoSync software (Reflective Computing) running on a dedicated PC with a 1,024 × 768 VGA video controller (Computer Boards). Accurate timing was assured by a photocell placed on the screen that sent a transistor-transistor logic pulse to the PC within 1 ms.

Behavioral tasks.

Trained monkeys performed visually guided and memory-guided delayed-saccade tasks. Each monkey had experience with these tasks for at least 2 yr. As with eye position, we measured lid position using a magnetic induction technique (Evinger et al. 1991; Fuchs and Robinson 1966). To measure blinks, a lid coil (4-mm diameter, 30 turns, 25 mg) made of Teflon-coated stainless steel wire, 0.003-in. diameter (A-M Systems, Everett, WA) was attached to the lower margin of the upper eyelid of the left eye using surgical tape and false eyelash adhesive. The lid coil was calibrated after each experiment by rotating it through known angles. Monkeys were habituated to the application of the lid loop using treat reinforcement and learned to tolerate the procedure well.

Monkeys sat with head fixed in a custom-designed primate chair and made memory-guided and visually guided delayed saccades. In the delayed-saccade task, a centrally located, visual spot appeared initially, and monkeys fixated this spot. After a random time of 500–1,500 ms, a peripheral spot appeared 10° or 20° from the fixation point at one of 8 angles. After another delay (800–1,200 ms), the fixation spot disappeared, and its disappearance served as a cue for the monkeys to make a saccade to the visual spot located in the periphery. The memory version of this task (Hikosaka and Wurtz 1983) was identical except that the spot located in the visual field appeared only transiently (200 ms). Monkeys remembered the location and made a saccade to that memorized location when the central spot disappeared. Memory-guided and visually guided saccade trials were interleaved randomly. Target location was also randomly interleaved. For a visual saccade, by 500 ms after the cue to make a saccade appeared, the eye position had to be within a 3° or 6° square around the target, depending upon target eccentricity. For a memory-guided saccade, the eye position had to be within a 5° or 10° square around the target, depending on target eccentricity. Eye position remained at this location for 300–500 ms. When performed correctly, monkeys received a drop of water (0.1 ml) as reward. Each session consisted of a total of 300–400 correct trials with an intertrial interval of 1,500–2,000 ms.

Data acquisition.

Eye and lid position were recorded using a magnetic induction technique (Evinger et al. 1991; Fuchs and Robinson 1966) (Riverbend Instruments, Birmingham, AL). Voltage signals proportional to horizontal and vertical components of eye position and lid position were filtered (eight-pole Bessel, −3 dB, 180 Hz), digitized at 16-bit resolution, and sampled at 1 kHz (CIO-DAS1602/16; Measurement Computing, Middleboro, MA). The data were saved and analyzed offline using an interactive computer program designed to display and measure eye and lid position and velocity. The analysis program detected saccadic and blink onsets and ends using a velocity criterion of 5.0% of peak velocity. In trials when the gaze-evoked blink began after the upward saccade began (Fig. 1C), blink start was defined as when the lid began to lower, and the end as when the lid stopped moving. To measure lid amplitude without including lid saccades that accompany saccadic eye movements, we measured the amplitude of a blink from the beginning of the rapid downward sweep of the lid, which occurs only when a blink occurs and is independent from the lid saccade, to the lowest position reached by the lid. Our measurements thus included only the amplitude of a blink, not the amplitude of a lid saccade, if one accompanied the blink. The experimenter verified the accuracy of computer-chosen values for every trial and adjusted them if necessary.

Fig. 1.

Blinks are often associated with saccades. Examples of visually guided (A and B) and memory-guided (C and D) eye and lid movements, with (B and D) and without (A and C) gaze-evoked blinks made by monkey 2. For each record, the top two traces show polar eye position (Pos) and polar eye velocity (Vel). The bottom two traces show upper eyelid position (LidPos) and eyelid velocity (LidVel).

Data analysis.

Statistical analyses were performed using SPSS. The saccadic velocity and duration data were analyzed by t-tests comparing visually guided to memory-guided saccades and trials with blinks to trials without blinks. Curvature of the vertical and horizontal components of saccades was determined independently. For each sample between the beginning and end of the saccade, we calculated the square of the difference between the actual movement and a straight line connecting the start and end of the saccade. We summed these values, took the square root, and normalized the value to the size of the movement described by a straight line to estimate saccade curvature. The curvature data were analyzed by t-tests comparing curvature for saccades with and without blinks.

RESULTS

Without being accompanied by a gaze-evoked blink, both visually guided and memory-guided saccadic eye movements showed typical, bell-shaped velocity profiles (Figs. 1, A and C, and 2) (Becker 1989). As expected from these velocity profiles, there was a strong correlation between peak saccadic eye velocity and the average velocity with the typical slope of 1.9 (Becker 1989; Evinger et al. 1981) (see also Fig. 4). As the sample saccades in Fig. 1 had an upward component, the eyelid exhibited a normal upward lid saccade associated with the vertical gaze shift (Evinger et al. 1991) (Fig. 1, A and B). When a gaze-evoked blink occurred with the saccadic eye movements, however, the eye velocity profile changed, disrupting the relationship between peak and average velocity (see also Fig. 4). A gaze-evoked blink occurring in the middle of a visually guided saccadic eye movement reduced peak saccade velocity and increased saccade duration relative to a saccade without an accompanying blink (Fig. 1, A and B). For memory-guided saccades, the occurrence of a gaze-evoked blink substantially decreased peak saccadic velocity and increased saccade duration relative to the same saccade without a blink (Fig. 1, C and D). The alterations seen in the velocity profiles of the saccadic eye movements with blinks were qualitatively similar for all three monkeys.

Fig. 2.

Gaze-evoked blinks alter saccade kinematics. A: averaged eye velocity for 9° to 11° saccades made in the visually (blue) and memory-guided (red) tasks from monkey 2 made on a single day (n = 20 each). B: averaged eye velocity for 9° to 11° visually guided saccades made with (blue) and without (gray) a gaze-evoked blink (n = 10 each). C: averaged eye velocity for 9° to 11° memory-guided saccades made with (red) and without (gray) a gaze-evoked blink (n = 10 each). Error bars are 1 SEM.

The averaged polar velocity of 40 saccadic eye movements between 9° and 11° from one monkey showed that memory-guided saccades achieved lower peak velocities and had longer durations than visually guided saccades (Fig. 2A). Separating the visually guided saccades into those with (Fig. 2B, blue) and without gaze-evoked blinks (Fig. 2B, gray) demonstrated that the occurrence of a gaze-evoked blink slowed peak saccadic velocity slightly and considerably increased the duration of visually guided saccades. Separating memory-guided saccades into those with (Fig. 2C, red) and without gaze-evoked blinks (Fig. 2C, gray) revealed that a blink substantially decreased peak saccadic velocity and increased the duration of memory-guided saccades. Thus gaze-evoked blinks affected the kinematics of memory-guided saccades more than those of visually guided saccades. Combining the stronger effect of gaze-evoked blinks on memory guided saccades with a higher probability of gaze-evoked blinks with memory-guided saccades points to the importance of gaze-evoked blinks in the slowing of memory relative to visually guided saccades.

As expected from previous studies on gaze-evoked blinks (Evinger et al. 1994; Williamson et al. 2005), blinks were more likely to accompany memory-guided than visually guided saccades. We analyzed 844 visually guided saccades in all, 98 with blinks (13%) and 746 without blinks, and 1,199 memory-guided saccades, 380 with blinks (32%) and 819 without. For monkey 1, blinks occurred with only 3% of visually guided saccades, whereas the probability of a gaze-evoked blink occurring with memory-guided saccades was 46%. For monkey 2, the probability of blinks with visually guided saccades was 27% compared with 87% for memory-guided saccades. For monkey 3, the probability of blinks with visually guided saccades was 7% and was 12% for memory-guided saccades.

Both the peak velocity and duration of memory- and visually guided saccades (Fig. 2, B and C) were different. To characterize these differences and to assess the influence of blinks on these eye movement parameters, we plotted the peak velocity and duration as a function of saccade amplitude for all visually and memory-guided saccades (Fig. 3, A and B) and for visually and memory-guided saccades with and without gaze-evoked blinks for all monkeys (Fig. 3, C–F). As reported previously (Gnadt and Andersen 1988), memory-guided saccades achieved lower peak velocities and had longer durations than visually guided saccades (cf., blue and red points Fig. 3, A and B). The slopes of the peak velocity-polar amplitude relationships were 32.8°/s/° for visually guided (r2 = 0.92) and 28.9°/s/° for memory-guided (r2 = 0.77) saccades. The mean peak velocity for visually guided saccades, 504.8 ± 4.6°/s, was significantly higher than the mean peak velocity for memory-guided saccades, 415 ± 4.8°/s [t(2,039) = 13.0, P < 0.001], even though their mean saccade amplitudes (visual: 12.3 ± 0.13°, memory: 12.5 ± 0.12°) were not significantly different [t(2,039) = −1.2, P > 0.05; Fig. 3A]. Likewise, the mean duration of visually guided saccades, 65.3 ± 1.7 ms, was significantly shorter than that of memory-guided saccades, 106.3 ± 2.2 ms [t(2,039) = −13.7, P < 0.001], despite similar amplitudes (Fig. 3B).

Fig. 3.

Blinks alter the kinematics of saccades. A: peak velocity of all visually guided (blue circles) and memory-guided (red squares) saccades plotted as a function of polar amplitude. B: duration of all visually guided (blue circles) and memory-guided saccades (red squares) plotted as a function of polar amplitude. C: peak velocity of visually guided (blue circles) and memory-guided (red squares) saccades without gaze-evoked blinks plotted as a function of polar amplitude. D: duration of visually guided (blue circles) and memory-guided saccades (red squares) without gaze-evoked blinks plotted as a function of polar amplitude. E: duration of visually guided (blue circles) and memory-guided (red squares) without gaze-evoked blinks plotted as a function of polar amplitude. F: duration of visually guided (blue circles) and memory-guided saccades (red squares) accompanied by gaze-evoked blinks plotted as a function of polar amplitude of saccades.

Separating visually and memory-guided saccades with blinks from those without blinks revealed that the occurrence of a gaze-evoked blink accounted for most of the difference in velocity and duration between visual and memory-guided saccades. For saccades without blinks, the slope of the amplitude-peak velocity relationship for visually guided saccades, 32.6°/s/° (r2 = 0.94; Fig. 3C, blue symbols), was only slightly steeper than the slope of memory-guided saccades, 31.7°/s/° (r2 = 0.77; Fig. 3C, red symbols). This small difference in slope created a significant difference in the peak velocity between visually and memory-guided saccades. The mean peak velocity for visually guided saccades, 515.1 ± 5.1°/s, was significantly higher than the mean peak velocity of memory-guided saccades, 457.3 ± 5.9°/s [t(1,553) = 7.4, P < 0.001], even though the average amplitude of the two groups of saccades was not significantly different [visual: 12.4 ± 0.15°; memory: 12.7 ± 0.16° t(1,553) = −1.2, P > 0.05; Fig. 3C]. For visually and memory-guided saccades without gaze-evoked blinks (Fig. 3E), the difference between saccade duration for the two types of saccades was similar to the difference between peak velocities. Despite the overlap in saccade duration for visually and memory-guided saccades, the mean duration of visually guided saccades, 50.0 ± 0.44 ms, was significantly shorter than the mean duration of memory-guided saccades, 60.2 ± 0.65 ms [t(1,553) = −12.7, P < 0.001]. The decreased velocity and longer duration of memory-guided saccades relative to visually guided saccades without blinks were consistent with the argument that a lack of visual drive is responsible for the slowing of memory-guided saccades. This reduction in saccade velocity, however, was only a fraction of the difference between visually and memory-guided saccades when a blink occurred.

When a gaze-evoked blink occurred with either a visually or memory-guided saccade, peak velocity (Fig. 3D) decreased and duration (Fig. 3F) increased substantially compared with saccades without blinks. The slope of the amplitude-peak velocity relationship for visually guided saccades was 24.4°/s/° (r2 = 0.45) and 16.7°/s/° for memory-guided saccades (r2 = 0.30) accompanied by a blink (Fig. 3D). Consistent with the large change in slope, the mean peak velocity for visually guided saccades with a gaze-evoked blink (437.3 ± 8.9°/s) was significantly higher than for memory-guided saccades with a blink [324.1 ± 5.9°/s; t(488) = 9.6, P < 0.001], even though saccade amplitude was not significantly different between the two groups [t(488) = −1.2, P > 0.05; Fig. 3D]. Likewise, the duration of visually guided saccades with a blink (168.9 ± 6.9 ms) was significantly shorter than memory-guided saccades with a blink [205.0 ± 3.1 ms; t(488) = −5.3, P < 0.001; Fig. 3F]. Thus for visually guided saccades, a gaze-evoked blink decreased average peak velocity by 16% and increased average saccade duration by 337%. For memory-guided saccades, a gaze-evoked blink reduced average peak velocity by 29% and increased average saccade duration by 340%.

These data (Figs. 2 and 3) suggest that average velocity (polar amplitude/duration) would be more informative than peak velocity in assessing the effects of gaze-evoked blinks on saccadic eye movement kinematics because average velocity incorporates both amplitude and duration in a single metric. Additionally, blinks can introduce multiple peaks into the saccadic eye velocity profile (Fig. 1D), making peak velocity an inappropriate measure (Gandhi 2012). In fact, one can assess the magnitude of the effect of gaze-evoked blinks on average velocity by examining the slope of the line relating peak velocity * duration to saccade amplitude. In the absence of a gaze-evoked blink, the slope of the line relating peak velocity * duration to saccade amplitude was 1.89 for visually guided (r2 = 0.78; Fig. 4A) and 1.88 (r2 = 0.7; Fig. 4B) for memory-guided saccades. When a gaze-evoked blink accompanied a saccade, however, the slope of this relationship increased to 3.0 (r2 = 0.17) for memory-guided saccades, and the relationship was lost for visually guided saccades (r2 = 0.001). Clearly, gaze-evoked blinks disrupt the relationship between peak velocity * duration and saccade amplitude for visually and memory-guided saccades. Thus it was critical to determine the effect of gaze-evoked blinks on average saccade velocity of visually and memory-guided saccades independent of peak velocity.

Fig. 4.

Gaze-evoked blinks disrupt the relationship between saccade amplitude and peak velocity * duration. A: peak velocity * duration as a function of amplitude for visually guided saccades with (blue) and without gaze-evoked blinks (gray). B: peak velocity * duration as a function of amplitude for memory guided saccades with (red) and without gaze-evoked blinks (gray).

Plotting average velocity as a function of saccade amplitude revealed a substantial effect of gaze-evoked blinks on saccadic eye movements (Fig. 5). The slope of the average velocity-saccade amplitude relationship was higher for visually guided saccades without gaze-evoked blinks (17.2°/s/°, r2 = 0.76; Fig. 5A, gray squares) than with accompanying blinks (12.6°/s/°, r2 = 0.25; Fig. 5A, blue circles). For memory-guided saccades, the slope of this relationship was 16.6°/s/° (r2 = 0.74; Fig. 5B, gray squares) without blinks and 5.7°/s/° (r2 = 0.45; Fig. 5A, red circles) with gaze-evoked blinks. For visually guided saccades, gaze-evoked blinks significantly decreased the mean average velocity from 252 ± 2.9°/s to 92.3 ± 6.1°/s [t(856) = −20.4, P < 0.001], even though there was no significant difference in saccade amplitude for these two groups [t(856) = −1.9, P > 0.05]. Gaze-evoked blinks also significantly reduced the mean average velocity of memory-guided saccades from 220.9 ± 3.2°/s to 65.9 ± 1.6°/s [t(1,197) = −32.5, P < 0.001], even though saccade amplitudes were not significantly different [t(1,197) = −1.62, P > 0.05]. Thus gaze-evoked blinks decreased the mean average velocity of memory-guided saccades by 70% and visually guided by 63%. These data show that the occurrence of gaze-evoked blinks dramatically slowed both visually and memory-guided saccades (Figs. 2, 3, and 5) with a slightly stronger effect on memory- than on visually guided saccades.

Fig. 5.

Gaze-evoked blinks reduce average velocity of saccades. A: average velocity of visually guided saccades with (blue circles) and without (gray squares) gaze-evoked blinks for all monkeys. B: average velocity of memory-guided saccades with (red circles) and without (gray squares) gaze-evoked blinks for all three monkeys.

Gaze-evoked blinks cause the same pattern of blink-evoked eye movements as occur with reflex, voluntary, and spontaneous blinks. The eyes rotate downward and medially during lid closure with reflex blinks (Bour et al. 2000; Collewijn et al. 1985; Goossens and Van Opstal 2000a, 2010). As expected, gaze-evoked blinks increased the curvature of saccadic eye movements. This curvature was most prominent for leftward and upwardly directed movements (Fig. 6A). Saccades directed downward and nasally exhibited little increase in curvature. We quantified blink-induced curvature of saccadic trajectory by determining the difference between the actual eye movement and a straight trajectory from the fixation spot to the final eye position (Fig. 6B). A blink increased the curvature of the horizontal and vertical components of visually guided saccades significantly [horizontal, t(89) = 6.2, P < .001; vertical, t(89) = 13.9, P < .001]. Horizontal and vertical component curvature significantly increased in memory-guided saccadic eye movements accompanied by a blink [horizontal, t(179) = 6.7, P < 0.001; vertical, t(179) = 5.9, P < 0.001]. The significant increase in vertical component curvature with memory-guided saccades was larger than the change seen in any other movement (Fig. 6B).

Fig. 6.

Blinks alter the trajectories of saccadic eye movement. A: 16 saccades from monkey 2 in which the solid lines show saccades with gaze-evoked blinks and the dashed lines show saccades without accompanying blinks. B: average curvature of the horizontal (open bars) and vertical (filled bars) components of visually and memory-guided saccades made with and without gaze-evoked blinks. Error bars are 1 SEM.

To consider further how gaze-evoked blinks affected saccadic eye movements, we estimated the slope of the amplitude-average velocity relationship by dividing saccade average velocity by saccade amplitude as a function of gaze-evoked blink amplitude. For all visually guided saccades without blinks, the estimated slope was 17.2°/s/° (Fig. 7A, blue circles) and 16.6°/s/° for memory-guided saccades (Fig. 7A, red circles). As gaze-evoked blink amplitude increased, the estimated slope of the amplitude average velocity relationship decreased until gaze-evoked blink amplitude exceeded 4°. For larger blink amplitudes, the average velocity slope reached an asymptote at 5.1°/s/°. Curvature, the average of the horizontal and vertical curvature, increased with blink amplitude (Fig. 7B), supporting the hypothesis that blink modifications of saccade trajectory are important in saccade slowing. If the neural generation of increased curvature contributed significantly to saccade slowing, then the estimated slope of the amplitude maximum velocity relationship should decrease as average saccade curvature increased. The data confirmed this proposal (Fig. 7C).

Fig. 7.

A: changes in saccade kinematics correlate with gaze-evoked blink amplitude. The estimated slope of the amplitude average velocity relationship plotted as a function of gaze-evoked blink amplitude for all memory- (red) and visually guided saccades (blue) are shown. B: average total saccade curvature plotted as a function of gaze-evoked blink amplitude for all memory- (red) and visually guided saccades (blue). C: the estimated slope of the amplitude-average velocity relationship plotted as a function of total saccade curvature for all memory- (red) and visually guided saccades (blue). Error bars are ± 1 SEM.

DISCUSSION

We investigated the basis for the slowing of memory-guided saccades relative to visually guided saccades. The typical explanation for this slowing is that the absence of a visual stimulus decreases the activity of superior colliculus neurons associated with a saccadic eye movement (Edelman and Goldberg 2001). Although this reduction in visual drive clearly contributes to the slowing of memory-guided saccades, our data indicate that the absence of a visual stimulus is a much smaller factor than the cooccurrence of a gaze-evoked blink in saccade slowing. For visually and memory-guided saccades without concomitant blinks, the mean average velocity of memory-guided saccades is only 12% slower than that of visually guided saccades. In contrast, the mean average velocity of memory-guided saccades with blinks is 74% slower than that of visually guided saccades without blinks.

Gaze-evoked blinks do not simply slow saccades; they fundamentally alter the kinematics of saccadic eye movements. Regardless of the range of saccadic peak velocities among species, multiplying 1.9 times the average velocity yields peak velocity (Becker 1989; Evinger et al. 1981). Without accompanying gaze-evoked blinks, memory- and visually guided saccades exhibit this relationship (Fig. 4). A gaze-evoked blink, however, disrupts this pattern with both memory- and visually guided saccades. Our data indicate that blink-evoked eye movements modify saccade kinematics and alter the trajectory of saccadic eye movements. As blink amplitude increases, saccadic eye movement curvature increases (Fig. 7B). Concomitantly, the average velocity of saccadic eye movements decreases (Fig. 7C). Thus the neural processes that create gaze-evoked blinks slow concomitant saccades, revealing themselves through the alterations in saccade trajectory.

Blinks other than gaze-evoked blinks also slow other types of eye movements. For example, a voluntary blink reduces the velocity of vergence eye movements (Rambold et al. 2002), and reflex blinks reduce smooth pursuit eye velocity (Rambold et al. 2005). Similar to their effect on smooth pursuit and vergence eye movements, reflex-evoked and voluntary blinks reduce the velocity and increase the duration of saccadic eye movements (Gandhi and Bonadonna 2005; Goossens and Van Opstal 2000a; Rambold et al. 2002). As all blinks create the same pattern of eye movements (Bour et al. 2000; Collewijn et al. 1985; Gandhi 2012; Goossens and Van Opstal 2000a; Goossens and Van Opstal 2010), the slowing of eye movements by other types of blinks is likely to reflect neuronal interactions between the circuits producing blink-evoked and saccadic eye movements.

Gaze-evoked blinks slow memory-guided saccades slightly more than visually guided saccades. A gaze-evoked blink slows the average velocity of memory-guided saccades by 70% and a visually guided saccade by 63%. This 7% larger effect of a gaze-evoked blink on memory-guided saccades probably reflects the stronger collicular activation associated with saccades made to visual targets compared with those made in the absence of visual targets (Edelman and Goldberg 2001). This difference raises the question of why memory-guided saccades are so much slower than visually guided saccades when compared across the entire population of saccades. Across all saccades with and without gaze-evoked blinks, the mean average velocity of memory-guided saccades is 26% slower than that of all visually guided saccades. This effect results from two differences in gaze-evoked blinks between visually and memory-guided saccades. First, gaze-evoked blinks occur more frequently with memory-guided than with visually guided saccades. Second, gaze-evoked blinks accompanying visually guided saccades are usually smaller than those occurring with memory-guided saccades (Fig. 7).

Our data demonstrate two mechanisms responsible for slowing memory-guided saccades relative to visually guided saccades. First, the absence of a visual target modestly reduces average saccade velocity (Figs. 2 and 5). Without a blink, the average velocity of a memory-guided saccade is 12% slower than a visually guided saccade. As previously suggested, this difference is best explained by a decrease in collicular activity in the absence of a visual target (Basso et al. 2000; Edelman and Goldberg 2001; Hikosaka and Wurtz 1985; Lee et al. 1988; Paré and Wurtz 2001). Second, the occurrence of a gaze-evoked blink dramatically slows memory-guided saccades. When a gaze-evoked blink occurs with a memory-guided saccade, average blink velocity decreases by 71% compared with memory-guided saccades without a blink. The discharge of superior colliculus neurons during visually guided saccades with or without gaze-evoked blinks was the same in the two neurons recorded (Fig. 4) (Goossens and Van Opstal 2000b). We hypothesize that the collicular discharge associated with memory-guided saccades is more susceptible to the disruptive effects of a gaze-evoked blink than is the collicular discharge associated with visually guided saccades. Thus, in contrast to a visually guided saccade, we would predict that a gaze-evoked blink associated with a memory-guided saccade would reduce collicular discharge in a manner similar to that seen with airpuff-evoked blinks (Goossens and Van Opstal 2000b). Based on our results and those of others recording in the colliculus, we conclude that the most important component of the slowing of memory-guided saccades relative to visually guided saccades is the likely effect of gaze-evoked blinks on collicular discharge.

GRANTS

This work was supported by National Eye Institute Grants EY13692 (M. A. Basso) and EY07391 (C. Evinger).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: A.S.P., M.A.B., and C.E. conception and design of research; A.S.P. and M.A.B. performed experiments; A.S.P., M.A.B., and C.E. analyzed data; A.S.P., M.A.B., and C.E. interpreted results of experiments; A.S.P., M.A.B., and C.E. drafted manuscript; A.S.P., M.A.B., and C.E. edited and revised manuscript; A.S.P., M.A.B., and C.E. approved final version of manuscript; C.E. prepared figures.

ACKNOWLEDGMENTS

We are grateful to Dr. Xiaobing Li for assistance with running some experiments.

REFERENCES

View Abstract