Smooth pursuit eye movements function to keep moving targets foveated. Behavioral studies have shown that pursuit is particularly effective for predictable target motion. There is evidence that both the frontal eye field (FEF) and supplementary eye field (SEF) (also known as the dorsomedial frontal cortex) contribute to pursuit control. The goal of the current experiment was to determine whether these 2 areas made different contributions to the initiation of pursuit in response to predictable compared with unpredictable target motion. Transcranial magnetic stimulation (TMS) was used in 5 healthy human participants to temporarily disrupt each area around the time of target motion onset. TMS over the FEF delayed contraversive pursuit markedly more than ipsiversive pursuit and this direction-dependent difference was more deeply modulated during pursuit of unpredictable than predictable target motion. By contrast, TMS over the SEF resulted in a much more muted modulation of pursuit latency that was similar across both predictable and unpredictable conditions. Taken together, we conclude that the human FEF, but not the SEF, makes a significant contribution to the processing required during the preparation of contraversive pursuit responses to unpredictable target motion and this contribution is less vital during pursuit to predictable target motion.
Smooth pursuit eye movements function to keep the image of a moving object on the fovea. The pathway by which visually guided smooth pursuit is generated has its origin in the motion processing centers in the cortex (Tychsen and Lisberger 1986; Lisberger et al. 1987; Komatsu and Wurtz 1989; Beutter and Stone 1998; Watamaniuk and Heinen 1999). This specialized subdivision of the visual system includes 2 areas near the superior temporal sulcus. The first area, the middle temporal (MT) area (Baker et al. 1981; Maunsell and Van Essen 1983a; Albright 1984) projects to the frontal eye field (FEF) located in the arcuate sulcus (Ungerleider and Desimone 1986; Tian and Lynch 1996a). A small area in the fundus of the arcuate sulcus has been shown to contain neurons that are active before and during smooth pursuit (MacAvoy et al. 1991; Gottlieb et al. 1994), and microstimulation of these neurons elicits smooth eye movements (Gottlieb et al. 1993; Tian and Lynch 1996b).
The second area, the medial superior temporal (MST) area (Maunsell and Van Essen 1983b; Desimone and Ungerleider 1986) also contains neurons active during pursuit (Newsome et al. 1988) and sends projections to both the FEF (Churchland and Lisberger 2005) as well as the dorsomedial frontal cortex (Huerta and Kaas 1990; Maioli et al. 1998), an area that corresponds anatomically with the supplementary eye field (SEF). In the macaque monkey, SEF neurons are active during smooth pursuit (Heinen 1995; Heinen and Liu 1997), and microstimulation of this area can elicit smooth eye movements (Tian and Lynch 1995; although see Russo and Bruce 2000). Although the cortical mechanisms underlying pursuit control may be different in monkeys compared with man, human imaging studies have identified the FEF, SEF, intraparietal sulcus, and the MT/MST complex as cortical areas active when performing smooth pursuit eye movements (Petit and Haxby 1999; O'Driscoll et al. 2000; Schmid et al. 2001).
In the present study, we investigated 2 aspects of FEF and SEF function relative to smooth pursuit responses. The first was the putative hemispheric specialization for contraversive versus ipsiversive pursuit. Previous research in both humans and nonhuman primates has demonstrated that large unilateral lesions to the FEF impair the ability to generate ipsiversive pursuit (Lynch 1987; Heide et al. 1996). In addition, microstimulation of the FEF tends to result in ipsiversive pursuit, although other pursuit directions can also be elicited (Gottlieb et al. 1993). By contrast, although single-unit recording studies have generally demonstrated directionally selective pursuit-related activity in the FEF; across populations of FEF neurons, all directions appear to be represented with no particular specificity for either ipsiversive or contraversive pursuit (MacAvoy et al. 1991; Tanaka and Lisberger 2001). This lack of directional coding is also apparent in the activity of the human FEF as assessed with functional magnetic resonance imaging (Rosano et al. 2002). Individual SEF neurons also possess directionally selective pursuit activity, but as in FEF, as a population there is no specificity for ipsiversive or contraversive pursuit. Thus, the evidence for directional selectivity within populations of FEF and SEF neurons is equivocal.
The second aspect of FEF and SEF function that was examined in the present study was that related to target motion predictability. It is known that prior information regarding when an object is going to appear, or in what direction it is going to move, allows us to better localize and respond to that object. This has been shown to be true for movements of the body (Thaut et al. 2002) as well as for eye movements (Dorris et al. 1999). Additionally, when targets being tracked with smooth pursuit eye movements are suddenly extinguished, subjects continue to produce pursuit for a short period of time (Von Noorden and Mackensen 1962; Becker and Fuchs 1985). This smooth pursuit output in the absence of a moving visual object suggests an interaction between oculomotor pathways and information regarding predicted future target motion. Thus, there is evidence for a relationship between cognitive factors, such as predicting when and where a target will be at a given time, and the processing within the pursuit system that is responsible for generating eye movements. Indeed, several recent human brain imaging studies have demonstrated that target predictability modulates the oculomotor activity in the FEF and SEF (Schmid et al. 2001; Gagnon et al. 2002).
Transcranial magnetic stimulation (TMS) is a noninvasive tool for the electrical stimulation of neural tissue, including the cerebral cortex, spinal roots, and cranial and peripheral nerves. TMS allows one to transiently alter neural function at a specific time relative to the processing of information related to the task of interest (Pascual-Leone et al. 1999). By measuring the resulting changes in performance, one can make strong inferences regarding the role played by the stimulated area. TMS has been used in only 2 previous studies to investigate the putative contributions of different brain regions during pursuit responses. Ohtsuka and Enoki (1998) demonstrated increased pursuit acceleration ipsilateral to the side of stimulation following a single pulse of TMS to the cerebellum. More recently, Gagnon et al. (2006) showed that single-pulse TMS over the FEF and SEF led to different changes in ongoing pursuit velocity depending upon when the pulse was delivered during the tracking of a predictable sinusoidal target motion. Our goal in the current study was to use single-pulse TMS to examine how the human FEF and SEF contributed to the initiation, as opposed to maintenance, of smooth pursuit under predictable and unpredictable conditions and whether this contribution was modulated by target motion direction.
Materials and Methods
Eight healthy young subjects were initially recruited into the current study from the University of Oregon community. Exclusion criteria included any one of the following: history of epilepsy or other neurological dysfunction, presence of any nonremovable ferrous material in or around the cranium, use of any cardioregulatory device, or consistent blinks in response to the delivery of TMS. The consistent blinkers were excluded because of the need for artifact-free eye movement recordings around the time of TMS delivery. Blinking was assessed with a brief procedure by recording eye movements during delivery of TMS over frontal cortex. Three of the subjects were excluded based on this criterion, and the remaining 5 completed the experiment. All 5 were male (mean age 30.8 years) with normal or corrected-to-normal vision including binocular stereoscopic vision. Each participant signed an informed consent form prior to taking part in the study, and the local university human subjects compliance committee approved the experimental protocol.
Our goal was to deliver single pulses of TMS over the left FEF and SEF. To ensure that this was accomplished as accurately as possible, prior to the experiment, anatomical magnetic resonance scans were collected on each subject using a 3-T magnetic resonance imaging scanner (Siemens Allegra, Malvern, PA). For this purpose, the head of the subject was positioned comfortably within the head coil, and the head motion was minimized. The subject also wore earplugs and headphones to protect their hearing. Whole-brain anatomical scans were collected using a T1-weighted magnetization-prepared rapid gradient-echo sequence (time repetition = 2500 ms, echo time = 4.38 ms, flip angle = 8°, field of view = 256 × 256 mm; 176 slices per slab at 1 mm slice thickness). A Brainsight Frameless Stereotaxic System was subsequently used to coregister anatomical landmarks on the skull with the underlying anatomy from the brain images so that the locations of the FEF and SEF could be identified for each subject. In particular, the FEF was localized to the junction of the precentral sulcus and the superior frontal sulcus (Fig. 1) (Paus 1996; Petit et al. 1997; Luna et al. 1998; Berman et al. 1999; Gagnon et al. 2002), and the SEF was localized along the midline adjacent to the upper region of the paracentral sulcus lying anterior to the Supplementary Motor Area (SMA) and descending into the interhemispheric fissure (Fig. 2) (Luna et al. 1998; Grosbras et al. 1999; Gagnon et al. 2006). Marks were placed on the scalp corresponding to these sites for subsequent positioning of the TMS coil in the experiment. An additional control site positioned over the vertex was used to test for any nonspecific effects of TMS on task performance.
Single pulses of TMS lasting ∼100 μs were delivered using a 2-T Magstim 200 delivered through a figure 8 coil. Immediately prior to each experimental session, TMS stimulation intensity was determined for each subject. This was accomplished by localizing the hand area of the motor cortex. The lowest current intensity at which an observable twitch of the first dorsal interosseus of the contralateral hand could most reliably be evoked by TMS was defined as the motor threshold. The stimulator output at this threshold was then increased by 10% during the experimental sessions when the FEF, SEF, or vertex were targeted.
Subjects were seated in a dimly illuminated room 57 cm in front of the display screen viewed binocularly. On each trial, a target (a plus sign “+”) subtending ∼0.5 degree was initially presented at the center of the screen for 1000 ms. Following this delay, it started moving at a constant speed to the left or right for 600 ms then disappeared. In the different conditions, the target direction (left or right) and speed (3, 5, or 10 degrees/s) were either predictable or unpredictable (see below). These speeds were similar to those used in our previous studies examining the interaction between attention and smooth pursuit (van Donkelaar 1999; van Donkelaar and Drew 2002). The subject was instructed to react to the onset of target motion as quickly as possible and to track the target motion as accurately as possible using smooth pursuit eye movements. After the target disappeared at the end of each trail, the screen remained blank for 2 s during which time the subject was instructed to make a saccade back to the starting position and wait for the beginning of the next trial.
Predictable Smooth Pursuit
In the predictable smooth pursuit condition, the target always moved in one direction (left or right) and at one speed (5 degrees/s). Thus, prior to every trial within a particular block, the subject knew both the direction and speed of the upcoming target motion. Because of this, subjects sometimes anticipated the onset of target motion. When this occurred (∼2% of the trials), the trial was discarded from further analysis, and the subject was verbally reminded to wait for the onset of target motion to initiate their responses. The subject completed 4 blocks of trials (2 with leftward target motion and 2 with rightward target motion) consisting of a total of 95 trials during which TMS was delivered 25% of the time. During TMS trials, the stimulation was given at one of 5 times: 60 or 120 ms prior to, at the time of, or 60 or 120 ms after target motion onset. The ordering of TMS trials within each block was pseudorandom to ensure that subjects were not aware if the upcoming trial involved TMS or, if it did, when the stimulation would be delivered. In separate counterbalanced sessions separated by at least 7 days, the FEF, SEF, or vertex were targeted for stimulation.
Unpredictable Smooth Pursuit
In the unpredictable smooth pursuit condition, the speed (3, 5, or 10 degrees/s) and direction of target motion (left or right) were unpredictable from trial to trial. All other aspects of target motion and TMS delivery were identical to those in the predictable condition. It should be noted, however, that only trials in which target speed was 5 degrees/s contained the possibility of including a TMS pulse. The completion of the 4 blocks of 95 trials resulted in a total of 20 trials of 5 degrees/s pursuit at each of the TMS onsets and 20 nonTMS trials at 5 degrees/s. In separate counterbalanced sessions separated by at least 7 days, the FEF, SEF, or vertex were targeted for stimulation.
Data Recording and Analysis
During each experimental session, the horizontal movement of the left or right eye was monitored using an infrared corneal reflection device (Iris Skalar, Kent, UK) attached to a semirigid adjustable band placed on the head of the subject. This system provided a signal proportional to the position of the eye with respect to the head with an optimal resolution of 2 min arc and linearity within 3% between −25 and +25 degrees. The system was calibrated by having the subject make saccades to targets at known eccentricities prior to each block of trials. A dental impression bite bar was used to stabilize the head throughout the experiment.
Data analysis was accomplished with a graphical user interface implemented in Matlab. Pursuit latency was the main dependent variable of interest. It was defined as the period of time from the onset of target motion to the onset of smooth eye motion. Additional variables that were measured included initial pursuit acceleration, and catch-up saccade latency and amplitude. Initial pursuit acceleration was defined as the mean rate of change in pursuit velocity during the initial 100 ms of smooth eye motion. This duration was chosen as representing the “open-loop” period of initial pursuit—that is, prior to any influence on the pursuit response from visual feedback (Lisberger and Westbrook 1985; Krauzlis and Lisberger 1994). The latency of the catch-up saccade was determined as the time at which the initial saccade occurred within the trial, and the catch-up saccade amplitude was the change in position of the eye that occurred during this saccade. Trials in which the initial response was a saccade or that contained a blink around eye motion onset were discarded, as were trials with anticipatory (latencies less than 70 ms) or sluggish pursuit onsets (latencies greater than 500 ms). Trials in which pursuit was initiated in the wrong direction were very rare and were equally likely to occur with and without TMS. Together, these anomalous trials accounted for less than 5% of the data overall.
For each of these variables, we computed a TMS effect by calculating the difference during nonTMS and TMS trials for each of the onsets. The TMS effect for each variable during the FEF, SEF, or vertex sessions were then separately submitted to 2 (target predictability [predictable, unpredictable]) × 2 (target direction [left, right]) × 6 (TMS onset [−120, −60, 0, 60, 120 ms, no TMS]) repeated measures analyses of variance. Where appropriate, significant main effects and interactions were further probed with post hoc Tukey's tests. Significance was set at P < 0.05.
Our goal was to examine whether the processes underlying the preparation of pursuit responses under predictable and unpredictable target motion conditions engaged the FEF and SEF differently. Toward this end, TMS was used to disrupt these areas during task performance under these 2 conditions, and the resulting changes in pursuit characteristics were assessed. We will first describe the changes in pursuit induced by FEF stimulation.
Figure 3 displays eye position traces during individual trials for a single subject in the unpredictable target motion condition. Overall, pursuit latency during trials without TMS (black lines) was slightly faster for leftward than rightward target motion for this individual. When TMS was delivered prior to target motion onset, pursuit responses were initiated more quickly overall (light gray lines) with the quickest responses generated during ipsiversive (leftward) target motion. In fact, for this particular individual, the quicker latencies induced by early TMS for pursuit responses to leftward target motion resulted in a reduction in the number of catch-up saccades. However, this response characteristic was much less common in the remaining participants. In contrast to the effect of early TMS, when TMS was delivered after target motion onset, pursuit was initiated more slowly overall (dark gray lines) with the slowest responses generated during contraversive (rightward) target motion. Figure 4 shows the group means for the TMS effect on pursuit latency as a function of TMS onset (A) and the interaction between target direction and predictability (B). Two main trends are apparent in the figures. First, there was a consistent modulation of pursuit latency across the different TMS onsets (Fig. 4A) (F5,120 = 16.632, P < 0.0001). Post hoc tests confirmed that latency was faster than baseline (i.e., during nonTMS trials) when TMS was delivered 60 ms prior to target motion onset and slower than baseline when TMS was delivered 60 and 120 ms after target motion onset. Second, this modulation in pursuit latency was most apparent for contraversive target motion (Fig. 4B) with this effect being more deeply modulated in the unpredictable condition (i.e., ∼20 ms latency delay in predictable condition and ∼50 ms delay in unpredictable condition) (F1,120 = 5.213, P = 0.025).
The only other dependent variable that showed any systematic modulation under these conditions was catch-up saccade latency. In particular, significant main effects of TMS onset (F5,120 = 3.448, P = 0.007) and predictability (F1,120 = 5.666, P = 0.02) as well as a significant interaction between TMS onset and predictability (F1,120 = 5.036, P = 0.001) were observed. These effects were due to a larger increase in saccadic latency relative to baseline in the unpredictable compared with predictable target motion conditions.
TMS delivered over the SEF produced much more muted effects than that observed during stimulation of the FEF (Fig. 5). The only significant effect that was observed was for TMS onset (F5,120 = 4.35, P = 0.036). Post hoc tests demonstrated that this was due to pursuit latencies being markedly slower than baseline when TMS was given at target motion onset only. All other main effects and interactions were not significant.
TMS delivered over the vertex produced no significant changes in any of the pursuit characteristics for either the predictable or the unpredictable target motion (Fig. 5). Thus, the alterations in task performance observed with stimulation at FEF and SEF were not due to any nonspecific effects of TMS.
The purpose of the present study was to examine the contribution of the human FEF and SEF to the preparation and initiation of smooth pursuit eye movements in response to the onset of target motion that varied in both direction and predictability. This was accomplished by delivering single pulses of TMS over the FEF and SEF around the time of target motion onset and characterizing the resulting changes in pursuit output. We found that TMS over the FEF produced a marked modulation of pursuit latency that was dependent on the timing of the TMS pulse relative to target motion onset as well as the direction and predictability of the target motion. By contrast, TMS over the SEF led to much more muted alterations in pursuit latency. Based on this data, we infer that the human FEF contributes to a much greater extent to the initiation of contraversive smooth pursuit, especially when the target motion that drives this response is unpredictable. By contrast, the SEF appears to play a much more minor role in pursuit initiation. In what follows, we discuss the current results in the context of previous findings related to the function of the FEF and SEF during smooth pursuit.
Although it is clear from previous research that both FEF and SEF contribute to the processing underlying pursuit, attempts to understand the details of this processing, especially with respect to target motion direction and predictability, have been much more common in the FEF. The results of the current study demonstrate a marked influence of TMS over the FEF on pursuit latency that was related to target motion direction relative to the side of stimulation and stimulation time relative to target motion onset. Previous human and nonhuman primates research has demonstrated that large lesions of the FEF impair ipsiversive pursuit (Lynch 1987; Heide et al. 1996), as do lesions to either area MT or MST (Dursteler and Wurtz 1988; Morrow and Sharpe 1993; Barton and Sharpe 1998). In addition, it has been shown that FEF microstimulation has a tendency to elicit ipsiversive pursuit (although all directions are represented) (Gottlieb et al. 1993). This tendency is much more robust in MT and MST where microstimulation during pursuit increases ipsiversive eye velocity and decreases contraversive eye velocity (Komatsu and Wurtz 1989). The current results are generally consistent with these observations and add a temporal component in that ipsiversive pursuit latencies sped up when stimulation was delivered prior to target motion onset, whereas contraversive pursuit latencies were more substantially increased when TMS was delivered after target motion onset. Thus, under the current conditions, TMS appeared to modulate pursuit output in a direction- and time-dependent manner.
Based on these observations, we speculate that prior to target motion onset, FEF cell activity is at or near baseline and TMS causes it to increase to a level closer to the threshold for eliciting the pursuit response. The information derived from the subsequent visual motion of the target leads to a pattern of activity that drives the cells over the threshold sooner than normal. By contrast, after the target has started moving and the pattern of cell activity is building up, TMS causes a disruption to the pattern resulting in latencies that are longer than normal. In other words, adding noise to the population signal with TMS when it is at baseline gets it closer to the threshold for activation, whereas adding noise when the population signal is evolving into a particular pattern is disruptive. The directional specificity of these effects implies that there is a balance of activity between the left and right FEF that leads to each making differing contributions to pursuit planning during the period surrounding target motion onset depending on the direction that the target moves. Given that the projections from MT and MST to the FEF are almost completely ipsilateral (Ungerleider and Desimone 1986), this balanced opposition appears more likely to result from local interactions between the FEFs in each hemisphere (Stanton et al. 2005).
In addition to pursuit direction, we also examined the contribution of the FEF to the modulation of pursuit responses by target motion predictability. Previous research has demonstrated that the FEF codes for the interaction between visual motion input and pursuit output because tasks such as tracking targets, anticipatory initiation, and predictive continuation of smooth pursuit are hindered following this manipulation (MacAvoy et al. 1991; Shi et al. 1998). Moreover, pursuit responses during microstimulation of the FEF display characteristics that are consistent with the reconstruction of predictive oculomotor signals (Fukushima 2003). Based on this and other data, the FEF has been postulated to be a site of pursuit gain control (Tanaka and Lisberger 2001) by which the pursuit response for a given visual motion stimulus is modulated in a context-dependent manner. Gagnon et al. (2006) have recently demonstrated that TMS over the FEF during ongoing pursuit systematically modifies eye velocity during the tracking of a predictable sinusoidal target motion. In particular, they showed that if TMS was delivered just prior to the reversal in target motion, eye velocity increased in the new direction. By contrast, if TMS was delivered at midcycle when target motion was fastest, eye velocity was decreased. This pattern of results was interpreted by the authors as reflecting modulation in the gain of predictive signals driving the ongoing pursuit responses.
The current data confirm and extend the results of Gagnon et al. (2006) by showing that TMS over the FEF modifies the processes underlying pursuit initiation in a manner that is dependent on target predictability and a combination of target motion direction and stimulation timing relative to target motion onset. The fact that the modulation of pursuit latency was larger in magnitude when target motion was unpredictable is consistent with the idea that the processes underlying pursuit initiation are more labile under these circumstances, and, thus, sensitive to the effects of TMS. By contrast, because the characteristics of the target motion were known in advance in the predictable condition, the initial response could be more thoroughly preplanned and simply released at the onset of target motion. As a result, TMS had a subtle yet systematically reduced influence on pursuit latency under these conditions.
The contribution of the SEF to smooth pursuit eye movements is less well understood than that of the FEF. There is clear evidence that this area contributes to the control of smooth pursuit output—neural activity in the SEF is modulated by pursuit output (Heinen 1995; Heinen and Liu 1997), and microstimulation during fixation can elicit smooth eye movements (Tian and Lynch 1995), whereas that during ongoing pursuit causes an omnidirectional increase in eye velocity (Missal and Heinen 2001). Human brain imaging studies have also consistently shown SEF activation during smooth pursuit (Berman et al. 1999; Petit and Haxby 1999; O'Driscoll et al. 2000; Schmid et al. 2001). In addition to stimulating the FEF, Gagnon et al. (2006) also delivered TMS over the SEF during ongoing pursuit of a sinusoidal target motion. They found that SEF stimulation increased eye velocity when delivered at target reversal, whereas it had no effect on eye velocity at the midcycle of the target trajectory. Based on this pattern of results, they concluded that SEF activation was particularly important for controlling pursuit velocity during the predictable changes that occur in target velocity at the target turnaround point of a sinusoidal trajectory and much less important during the maintenance of relatively steady pursuit velocity occurring at midcycle.
The results of the current study provide further insight into the conditions under which the SEF plays a role in the processes underlying pursuit responses. In particular, the demonstration that TMS over the SEF causes only a modest increase in pursuit latency when delivered at target motion onset that is not dependent on target predictability or direction implies that this area makes a much less significant contribution to controlling pursuit initiation than that of the FEF. When considered together with the results of Gagnon et al. (2006), it appears that the SEF contributes most substantially to predictable changes in ongoing pursuit and is much less involved in maintaining steady pursuit or initiating pursuit in response to either predictable or unpredictable target motion.