Basic properties of saccadic eye movements

Thus, FEF provides a major efferent pathway from the neocortex for commanding voluntary saccadic eye movements, and so understanding this highly specialized aspect of the foveation system is key to understanding FEF function. Saccades are brief (usually <75 msec) and fast (usually >500 deg/sec) eye movements. Humans average about two saccades per second while awake, and so our day is spent in brief fixations of different small parts of the visual world, constantly, but only briefly, interrupted by saccades. This unremitting procession, the processing of foveal visual information during a fixation, as well as the planning of the next saccade, engages much of the human brain, as exemplified by brain-imaging studies of humans doing both simple and complex visuomotor tasks (e.g., Luna et al., 2001; O'Driscoll et al., 2000).

Saccades are open-loop (i.e., ballistic, preprogrammed) movements. In fact, less than 50 to 100 msec prior to its start, a saccade generally cannot be canceled or redirected on the basis of new sensory information. The saccadic waveform and its main parameters (duration, peak velocity, etc.) are largely determined by the dimensions of the movement being accomplished, and thus one cannot simply choose to make faster or slower saccades.1 The saccadic objective is to effect the desired ocular displacement as quickly as possible and immediately return the eye to stationary fixation in the new direction. In fact, vision during the saccade is poor or absent (saccadic suppression), meaning that the subject is nearly blind during saccades—more reason for their high speed and brief duration.

Saccadic launch is under voluntary control, and saccades have relatively long latencies (i.e., reaction times). For example, saccade latency, as measured from the time of the appearance of a visual target until the start of the saccadic movement to foveate the target, is usually 100 to 400 msec, with 200 msec being typical. By contrast, the vestibulo-ocular reflex has a 10 msec latency. Although saccades to quickly foveate suddenly appearing visual stimuli are somewhat reflexive; one can readily forgo making saccades to even highly salient stimuli. Furthermore, one can instead make accurate and advantageous “predictive” or “anticipatory” saccades to the location where a stimulus is likely to appear in the future or even antisaccades in the opposite direction as a stimulus.

Thus, saccades are guided by our experience, guesses, objectives, and strategies,2 as well as by overt visual stimulation, even though their motoric details are largely independent of their purpose. Overall, the saccadic plan is to consider judiciously whether to disengage the current foveated stimulus, select a new stimulus of interest, plan a movement based on the retinal locus that will foveate that new stimulus, execute the plan in a ballistic fashion, and then process the new foveal image for at least a few hundred milliseconds before considering a subsequent saccade.3 Reading is the prime example of the saccadic system fully engaged under voluntary control.

Neural basis of saccadic eye movements

Figure 96.3 summarize the brain's distributed network for triggering and carrying out voluntary saccades. All saccades are ultimately fashioned by a dedicated neural circuit in the brainstem called the saccade generator (e.g., Scudder et al., 2002). This network of several distinctive types of neurons in the pontine and mesencephalic reticular formation originally evolved to produce the quick-phase movements of vestibulo-optokinetic nystagmus that reflexively stabilize images on the retina. In primates, however, it serves as the final common pathway for all rapid eye movements, and so if the saccade generator is damaged, then all rapid eye movements, reflexive and voluntary, are disabled. Thus, the voluntary saccades of primates involve triggering a part of the phylogenetically older image stabilization circuitry for the purpose of foveation.

The output of the saccade generator is a displacement vector of the current direction of gaze and a neural integrator downstream of the saccade generator computes the tonic eye position signal needed to hold the eye steady in its new position following each saccade. Because it serves all eye movement commands, smooth/slow eye movements as well as rapid/saccadic movements, it is called the common neural integrator. Figure 96.3 details the brainstem locale of the saccade generator and neural integrator and illustrates how all saccadic eye movements rely on these two vital brainstem circuits.

Figure 96.3..  

Pathways for voluntary and visually guided saccades and smooth pursuit. This diagram shows anatomical connections between cortical and subcortical structures involved in the control and generation of saccadic eye movements. The FEF receives visual information via pathways originating in the striate cortex, as do the PEF and the SEF. Note that all visual cortical areas project to the SC, with FEF, PEF, and SEF projecting primarily to its intermediate layers. By contrast, the superficial layers of SC receive direct visual projections from the retina and indirect visual projections from the striate and extrastriate cortices. Connections to locations marked with a star principally carry visual signals, whereas those marked by a saccade and burst are principally motoric.

The brainstem saccade generator is located in the paramedian pontine reticular formation (PPRF) and in the rostral interstitial nucleus of the medial longitudinal fasciculus (riMLF). The brainstem neural integrator is located in the medial vestibular nucleus (MVN), the adjacent nucleus prepositus hypoglossi (NPH), and the interstitial nucleus of Cajal (INC). The cerebellum controls the gains of both the saccade generator and the common integrator, in addition to being a critical link in the pursuit pathway. The smooth pursuit pathways shown with dashed boxes, dashed lines, and italicized text is very basic; for example, V5 and other temporal and parietal areas, as well as the pursuit zone of FEF, project to the pursuit-related pontine nuclei. SNpr, substantia nigra pars reticulata; NRTP, nucleus reticularis tegmenti pontis; LIP, lateral intraparietal sulcus; Suppl, Supplementary; Ø, eye position; dØ/dt, eye velocity; (dØ/dt)_sac, saccadic eye velocity; (dØ/dt)_slow, slow (or smooth) eye velocity; VOR, vestibulo-ocular reflex; OKR, optokinetic reflex; SP, smooth pursuit; (−), inhibitory synapse; A *ast; (Ø − Ø₀) + B * dØ/dt, equation for spike rate of the general oculomotor neuron as a function of eye position and velocity; A and B, constants; Ø, eye coordinate along the pulling direction of the extraocular muscle that is innervated; dØ/dt, eye velocity along the same direction; Ø₀, eye position at which the oculomotor neuron is recruited.

Thus, FEF does not directly control the extraocular musculature, but instead achieves its purposes by triggering and modulating the brainstem circuitry of the much older image stabilization system, that is, the circuits underlying the reflexive eye movements that serve to hold the whole visual world stationary on the retina. Without oculomotor stabilization during movements of the head or body, the retinal image is smeared and all vision, foveal and peripheral, is severely compromised. The stabilization system generally functions reflexively in parallel with the more voluntary foveation system (for a review, see Bruce and Friedman, 2002).

The superior colliculus and the visual-grasp response

The SC is the vertebrate brain's prototype sensorimotor structure and occupies a key position in the saccade circuitry of Figure 96.3 above the saccade generator. The primate SC receives topographic visual projections from corresponding parts of both retinas that portray a precise map of the contralateral hemifield on each hemisphere's SC. This visual map is found in the superficial layer of the SC and is in register with the map of the movement vectors for saccade-related neurons in the intermediate layer. This remarkable alignment of the sensory and motor maps underscores the principal role of the primate SC, which is to move the eyes so as to foveate quickly a stimulus seen anywhere in the visual periphery by commanding the saccade generator to fashion a rapid eye movement of the appropriate size and direction. This prepotent visually guided saccade has been called the visual grasp reflex and is the basic building block of the foveation system.

Probably the most important output from FEF for effecting saccades is its massive topographic projection to the intermediate layers of the SC (Huerta et al., 1986b; Stanton et al., 1988). Hanes and Wurtz (2001) found that FEF-elicited saccades were eliminated or distorted by temporary inactivation of parts of the SC map. Nevertheless, visually guided saccades and FEF-elicited saccades survive complete ablation of SC in primates because the primate also has a substantial neocortical network for saccades with access to the saccade generator independent of SC. Figure 96.3 also shows these cortical pathways and clarifies some effects of such lesions. For example, the sparing of visually guided saccades following SC lesions is mediated by FEF projections to the brainstem saccade generator that bypass the SC. The SC, however, normally provides a faster and more accurate pathway for visually guided saccades via its direct retinal projections and accurate retinotopic map, which explains why an increase in the latency of visually guided saccades is a lasting deficit of SC damage. Likewise, FEF lesions alone generally spare visually guided saccades; however, FEF lesions combined with SC lesions eliminate most visually guided saccades (Schiller et al., 1980). Thus, the primate SC and FEF provide both serial and parallel pathways for visual triggering of the brainstem saccade generator for the purpose of foveating saccades.

Low-threshold FEF and presaccadic burst activity

Bruce et al. (1985) define FEF as cortex where saccades are elicited by microstimulation at 50 µA or less with the electrode tip in gray matter. In the macaque monkey, this low-threshold (saccadic) FEF lies primarily in the anterior bank and anterior lip of the arcuate sulcus (see Figs. 96.1 and 96.2) and has a distinctive cytoarchitecture (see also Stanton et al., 1989).

When recording is restricted to the saccadic FEF, the vast majority of neurons have robust task-related changes in their discharge rate when a monkey performs almost any purposive saccadic eye movement task, such as a visually guided saccade task (see below), and a full range of saccade sizes and directions are tested for each cell. With appropriate control experiments, these functional activities can be parsed into several types, such as visual responses, auditory responses, presaccadic movement responses, postsaccadic responses, tonic gaze-direction responses, predictive movement activity, mnemonic visual activity, and so on (see below; also see Bruce and Goldberg, 1985; Schall, 1991; Sommer and Wurtz, 2001).

Incidence and significance of presaccadic burst activity in FEF

One particular activity, a burst of increased spiking that starts immediately prior to saccades into a cell's movement field, is the signature activity of the FEF. Approximately one-third of FEF neurons have this presaccadic burst (PSB) activity.4 A representative PSB from an individual FEF neuron is shown in Figure 96.4. PSB activity begins just prior to saccade initiation, usually ends approximately when the saccade is completed, and is always tuned, albeit fairly broadly, for a particular saccade vector (i.e., its movent field).

In many respects, FEF cells with PSB activity are remarkably similar to saccade-related bursters (SRBs), which are abundant in the intermediate layers of the SC (e.g., Glimcher and Sparks, 1993) and are even similar to vectorial long-lead burst neurons found in the rostral pontine reticular formation (e.g., Hepp and Henn, 1983). But there are significant differences in timing (Segraves and Park, 1993). Moreover, referring to the PSB activity in FEF as generic motor or movement activity (as in presaccadic movement responses, e.g., Bruce and Goldberg, 1985), although not incorrect, implies, by analogy with true motoneurons in the spinal cord and cranial nerve nuclei, a stronger relation between the FEF activity and the realized saccadic eye movements than is truly the case.

Figure 96.4..  

FEF neuron tested with the memory-saccade task. This neuron combined PSB with phasic-tonic visual activity and mnemonic continuation of the visual activity. Top: task events. Shortly after the monkey fixates a central light, a peripheral cue briefly appears. Then, after a delay period, the fixation light is extinguished and the monkey must saccade to the location where the peripheral cue had been shown earlier. The location was varied from trial to trial but was in the response field (RF), except for several trials in the middle of testing, when it was deliberately moved to locations outside the RF (see rasters and Gaussian-fit plots). Top-middle: rasters (by trial) and histograms of spike activity aligned on cue onset (left) show the burst of activity elicited by the visual stimulus, and those aligned on the saccade onset (right) show the large burst of activity preceding the saccades. Note that the visual response has three components: a phasic high-rate burst to the initial appearance of the visual stimulus (latency ∼50 msec), a robust tonic visual discharge while the cue remained on, and then, starting ∼50 msec after the cue was extinguished, a medium-level tonic mnemonic response that was maintained above the cell's baseline level of activity throughout the delay interval. Bottom-middle: spike density histograms based on rasters. Blue and green indicate the intervals used for the Gaussian fits (bottom) of best direction for visual and PSB activity, respectively. (From M. S. Kraus et al., unpublished data.)

Most FEF neurons with PSB activity also have one or more other functional activities. Most conspicuously, approximately two-thirds of FEF cells with PSB also have significant visual activity in that they will respond to the appearance of simple visual stimuli, such as a spots of light, within a large peripheral receptive field (RF), even when no saccade is involved.5 Such was the case for the neuron of Figure 96.4. Conversely, a substantial fraction (∼20%) of FEF cells have visual responses but lack PSB activity.

Relationship of PSB activity to electrically elicited saccades

PSB activity in the FEF correlates very strongly with the electrically elicited saccade phenomenon that has been used to discover, define, and map FEF. Bruce et al. (1985) found that thresholds for eliciting saccades were lowest at the sites of cells with PSB activity. When testing stimulation at the site of cells lacking PSB activity (usually these other cells were visually responsive, had postsaccadic activity, or both), far more electrical current was usually needed to elicit saccades. For example, low thresholds (<50 µA) were found at the sites of 34 cells with both visual and PSB activity and 11 cells with PSB activity only, but only at the site of 3 cells with visual responses and no PSB activity. Conversely, high thresholds (>150 µA) were found at the sites of 11 VIS + PSB cells and 2 PSB-only cells, but at the site of 21 visual-only cells. Thus, recording PSB activity is highly indicative that electrically evoked saccades will be obtained with low currents.

Moreover, the dimensions of the evoked saccades at FEF sites are highly correlated, in both direction and amplitude, with the movement fields for PSB activity recorded from neurons at the same site (Bruce et al., 1985) (Figs. 96.3 and 96.4). Dimensions of both the PSB movement fields and the evoked saccades change together in a fairly systematic manner across the extent of the FEF (Bruce et al., 1985; Dias and Segraves, 1999).

Given this close relationship with the evoked saccades, it is not surprising that FEF cells with PSB activity are prevalent among those found to project to the SC and to the pontine reticular formation. Indeed, Segraves and Goldberg (1987) found that 59% of corticotectal neurons had presaccadic movement-related activity and that another 22% had foveal and fixation-related activity. The remaining 20% were miscellaneous FEF types. In fact, none of the 51 corticotectal cells in this study were primarily responsive to peripheral visual stimuli, even though this is an extremely common FEF cell type (∼20%, as noted earlier; also see Fig. 96.6). A similar result was later found for FEF cells projecting to oculomotor regions of the pons (Segraves, 1992), with 48% of the corticopontine neurons firing in association with saccadic eye movements and 28% responsive to visual stimulation of the fovea combined with activity related to fixation.6

We infer from these data that the electrically evoked saccade phenomenon principally reflects the excitation of FEF cells with PSB activity that, in turn, activates the brainstem saccade generator via the corticotectal and corticopontine projections of FEF (Huerta et al., 1986; Stanton et al., 1988b). Thus, the functional involvement of the FEF in saccadic eye movements seems to involve naturally occurring PSB activity being sent to the saccade generator over the same FEF efferent pathways responsible for elicited saccades.

PSB activity is FEF's signature neuronal activity

In summary, the PSBs are a distinctive FEF activity. PSBs are manifest in ∼33% of FEF neurons but are relatively rare in other neocortex, including the periprincipal prefrontal cortex anterior to the FEF/arcuate and the premotor cortex lying posterior to the arcuate/FEF. These two adjacent cortical regions both have abundant visual responses with large RFs fairly similar to those in FEF.7 Thus, although visual responses, both peripheral and central/foveal, are abundant in FEF, what truly distinguishes neural activity in FEF from activity in all other neocortical areas (except for the two other cortical eye fields, SEF and PEF) is FEF's abundance of neurons with robust PSBs. Moreover, the PSB activity is very strongly associated with the low-threshold elicited saccade phenomenon used to discover, define, and map FEF.

An extremely strong line of evidence for the fundamental importance of the FEF's PSB activity for the production of saccadic eye movements comes from countermanding experiments. Hanes et al. (1998) find that FEF neurons with eye movement–related activity basically control the production of gaze shifts in this special paradigm where a signal to stop the saccade being prepared is sometimes given. Saccades were initiated if and only if the PSB activity of FEF neurons reached their threshold activation level, and not if their PSB activity decayed in response to the stop signal before reaching that level. In contrast, the spiking of FEF neurons with only visual activity was not critically related to the success of the stop signal.

FEF activity in conjunction with memory saccades

Single-cell examples of strong FEF activity in conjunction with memory saccades have been shown in many reports. Here we present the aggregate FEF response. Figure 96.5 shows the aggregate histograms from a primary database of 341 FEF cells that were (1) all deemed to have a significant presaccadic response and (2) all had their presaccadic responses parsed using the memory saccade task. The spike-density histogram on correctly performed trials in which the memory saccade was directed into its response field was computed for each neuron and then averaged across all 341 cells. Figure 96.6 shows the same data, but segregated into three basic FEF neuron types: 58 neurons with purely visual responses, 190 with both visual and PSB activity, and 93 with PSB activity but few or no visual responses.

Figure 96.5..  

Aggregate FEF response on the memory-saccade task. Average activity of 341 FEF neurons (from six monkeys) tested on the memory-saccade task. Histograms of spike activity aligned on the cue onset (left) show the burst of activity elicited by the visual stimulus. The histograms aligned on the saccade onset (right) show the large burst of activity preceding and during the saccades. This same PSB of activity is shown in relation to the signal to saccade in the middle histogram. Note that the visual response has three components: a phasic high-rate burst to the initial appearance of the visual stimulus (latency ∼50 msec), a robust tonic visual discharge while the cue remained on, and then, starting ∼50 msec after the cue was extinguished, a low-level tonic mnemonic response that was maintained throughout the delay interval. (Data from Friedman et al., 1997.)

Figure 96.6..  

Aggregate FEF response on the memory-saccade task segregated by neuron type. This is the same experiment and data as in the previous figure, and presented in the same way except that the 341 neurons are separated into three categories: Top: neurons with only a visual response; Middle: neurons with both a visual (VIS) and a presaccadic burst (PSB); Bottom: neurons with a PSB but no visual response. The red lines indicate the average baseline rate during fixation; the green lines are the average rate during the delay period after the cue is off. Note that PSB-only neurons do not have an elevated rate during the delay period and that VIS + PSB neurons have a higher elevation than VIS-only neurons. (Data from Friedman et al., 1997.)

These aggregate histograms indicate that overall spiking rate in the FEF is largest in conjunction with saccades, not with the visual target's appearance. Specifically, Figure 96.5 shows that the population PSB in conjunction with memory saccades has over twice as many spikes as the aggregate phasic response to the visual cues. The aggregate histograms in Figure 96.6 further make the case that the PSB is the more intense FEF activity: for the VIS + PSB cells, the PSB is much larger than the phasic visual burst. Moreover, the PSB for memory saccades of the PSB-only cells is larger than the phasic visual cue response of the VIS-only cells.

Memory-saccade PSBs versus PSBs for visually guided saccades

We compared the memory saccade activity of a subset of the cells shown in Figures 96.5 and 96.6 with their activity on two visually guided saccade tasks, namely, the deferred-saccade and step-saccade tasks. The deferred-saccade paradigm8 is identical to the memory-saccade paradigm except that the peripheral cue remains on and is thus still available to guide the saccade when the fixation light is eventually extinguished (which is the signal to saccade in both tasks). In the step-saccade task, the appearance of the peripheral spot coincides with the extinction of the fixation point (as if the fixation point quickly moved or “stepped”). Hence, there is no imposed wait (or delay) period, and the monkey can saccade as quickly as desired after the saccade target light appears.

Figure 96.7 shows that the aggregate PSB magnitude for the memory saccade is nearly identical to the PSB magnitude for the deferred saccades and is only slightly smaller than the PSB on the step-saccade task. This equality is particularly unexpected in light of a similar comparison recently carried out in the SC by Edelman and Goldberg (2001). They found that many SC burster cells did not respond in conjunction with memory saccades or responded much less than on their version of the deferred-saccade task. Thus, it would seem that FEF PSB activity is more reliable than SRB activity in the SC in conjunction with purposive memory-guided saccades performed without a visual target, even though the SRB activity in the SC is thought to be significantly more reliable than the PSB activity in FEF for most saccades, especially spontaneous saccades.

Figure 96.7..  

Comparison of PSB activity in FEF for visually guided saccades with PSB for memory-guided saccades. Histograms are aggregate activity of sets of FEF neurons recorded from six different monkeys during both tasks. Top: deferred-saccade task. Shortly after the monkey fixates a central light, a peripheral target appears in the neuron's response field. This target remains on for the remainder of the trial, but no saccade is permitted until the fixation light is extinguished at the end of the delay period. Thus, the target is an established presence at the time of the saccade. Bottom: standard step-saccade task. The fixation light is extinguished at the same time that the peripheral target appears. Left: histograms aligned on the fixation point's extinction. Note that the deferred-saccade PSB activity has the same latency as the memory-saccade PSB activity, but that the step-saccade response slightly leads the memory-saccade response; however, the memory saccades themselves have a longer latency than the step saccades. Right: histograms aligned on the saccade start. Note that the deferred-saccade PSB has the same magnitude as the memory-saccade PSB activity, but that the step-saccade PSB activity is larger than the memory-saccade response; presumably this reflects the phasic visual response to the appearance of the peripheral target of the VIS + PSB cells in this sample. The number of neurons is smaller than that in the previous figures because (1) only a subset of the 341 cells were tested on both of these two visually guided saccade tasks, and (2) VIS-only cells were omitted from this comparison; only the PSB-only and the VIS + PSB cells were used. (Data from Friedman et al., 1997.)

FEF visual activity

As mentioned earlier, over half of the neurons in FEF are visually responsive. Typically, they have large RFs entirely or predominantly in the contralateral visual hemifield and respond promptly to the appearance of almost any visual stimulus within their RF, with little selectivity for color or form. FEF visual responses do not require overt attention to the stimulus or to the stimulus location, or that the visual stimulus have any functional significance. In general, the only requirement is that the stimulus be in the cell's RF.

Phasic, tonic, and mnemonic visual activities

The strongest visual responses in FEF are the short-lived phasic responses to the initial appearance of visual targets (e.g., Figs. 96.4 and 96.5). The intense phasic responses can serve as both target and trigger for FEF PSB activity in the step-saccade task. This reflects the fact that the initial appearance of a visual target sometimes immediately garners an unwelcome saccade in the deferred-saccade and memory-saccade tasks, and also the fact that saccades in the step-saccade task have significantly shorter latencies relative to tasks in which the trigger to saccade is given long after the initial visual response (e.g., a mean of 236 msec for the step saccades vs. 270 msec for the deferred saccades for the behavior in Figure 96.5).

Some FEF cells respond phasically only to visual stimuli; however, assuming that a saccade to the stimulus's initial appearance does not occur, other visual cells then discharge tonically, usually at a relatively low rate, for as long as the stimulus remains in their RF (e.g., Fig. 96.4). This tonic FEF visual activity is available to target saccades to established visual targets on the deferred-saccade task when the signal to saccade is finally given.

After the visual cue is extinguished, many FEF cells still have significantly elevated activity during the delay period of the memory-saccade task (e.g., Figs. 96.4 and 96.5). This persistent mnemonic activity provides the targeting information that could underlie the spatially appropriate PSBs that materialize in FEF following the signal to saccade (usually extinction of the fixation light) at the end of the delay period. In Figure 96.6, note that the mnemonic/delay-period rate is most elevated for the combined visual-PSB cells, less elevated for the 58 visual-only cells, and nearly at the baseline level for the 93 PSB-only cells. Thus, FEF mnemonic activity is strongly associated with visual activity but is greatest in those visually responsive cells that also have PSB activity.

Alignment of visual and presaccadic movement fields

Visuomovement FEF cells have both visual and PSB activities, and their visual RF generally corresponds with the optimal saccade vector for their burst (i.e., movement field), as was shown for the cell of Figure 96.4. Moreover, the electrically elicited saccadic eye movement vector obtained at the cell's location generally matches its response fields as well. Thus, the FEF default is a foveating saccade. However, the PSB is independent of the location and/or presence of visual stimulation, particularly as shown by the antisaccade tasks and remapping experiments. Moreover, when carefully tested and fit as the cell in Figure 96.4, some visuomovement FEF cells turn out to be significantly discordant, with nonmatching, or even nonoverlapping, visual and movement fields (Friedman et al., 1998).

Remapped FEF auditory activity

Although auditory responses in FEF are infrequent, some cells in and near the medial FEF do respond to sounds (Azuma and Suzuki, 1984; Bruce, 1990; Bruce and Goldberg, 1985; Russo and Bruce, 1989). However, the sound source location is initially given in head-centered coordinates, and it would seem inappropriate to generate PSBs in an oculocentric framework. Interestingly, these auditory RFs in frontal cortex are partially remapped from a craniocentric to an oculocentric framework (Russo and Bruce, 1993), similar to the remapping of SC auditory fields found by Jay and Sparks (1987b).

Such remapping should facilitate the production of appropriate PSB activity for aurally guided saccades. However, this production is complicated by the fact that the monkey's ears move substantially, and these pinna movements have a systematic relation to voluntary eye movements (Bruce et al., 1988). Since pinna orientation strongly affects the magnitude of the sounds reaching the inner ear, accurate remapping of the sound source location to oculocentric coordinates would have to take pinna position, as well as eye position, into account. Interestingly, the macaque has a small pinna movement region of cortex adjacent to the medial (large-saccade) region of the FEF (Burman et al., 1988).

Figure 96.8..  

PSBs of FEF neurons in conjunction with aurally guided saccades. The rasters on the right show activity of an FEF neuron preceding visually guided saccades to an LED and preceding identical saccades guided to noise from a speaker at the LED location in the dark. The traces above the rasters show the waveform of representative visually and aurally guided saccades, and the raster lines are aligned on the saccade start. The scatter plot on the left shows, for all 22 FEF cells tested, PSB rates for aurally guided saccades versus those for visually guided saccades. (Adapted from Russo and Bruce, 1994.)

Fixation status signals (tonic foveation and eye position activity)

The FEF cells that respond in relation to fixation or foveal stimulation could trigger both the FDD and PSB. One such type is excited by fixation (foveal) stimulation; these cells could play a role in suppressing activity in FEF and elsewhere in the interest of maintaining fixation. Another class has the inverse activity, being suppressed by foveal stimulation and spiking in response to extinction of the fixation light. Hanes et al. (1998) found that the discharge of FEF fixation neurons exhibited saccade-control properties in the countermanding situation. Moreover, fixation cells in FEF often project to the SC (Segraves and Goldberg, 1987), where their signals could play a role in triggering and suppressing SRB activity there.

Some of these fixation responses are modulated by the current eye position, and a few FEF cells simply respond tonically as a linear function of absolute eye position, for example, azimuth or elevation (Bruce and Goldberg, 1985). Both of these phenomena could be created by an efferent copy signal derived from the common neural integrator in the brainstem. This signal of current eye position may allow the FEF to be somewhat biased against calling for centrifugal saccades that move the eye to extreme orbital positions, and to be biased for centripetal saccades that move the eye closer to a central position in the orbit. An eye position signal could be used to remap sensory locations (see below).

Remapped visual activity

Goldberg and Bruce (1990) hypothesized that a subtraction operation within FEF uses an efferent copy of the first saccade to remap FEF visual activity evoked by the second light to be suitable for guiding the second saccade (see also Bickle et al., 2000; Bruce, 1990; Umeno and Goldberg, 1997, 2001). In other words, visual activity in FEF is initially given in retinal coordinates but is remapped by saccades to maintain its oculocentric validity. Similar remapping of visual activity is prominent in PEF (Colby and Goldberg, 1999), and PEF may be critical for the double-step situation (e.g., Duhmael et al., 1992).

FEF postsaccadic activity codes the last saccade (efferent copy)

Postsaccadic activity in FEF was first described by Bizzi (1967, 1968), and ∼25% of FEF neurons are excited after particular saccadic eye movements (Bruce and Goldberg, 1985). This postsaccadic activity seems to represent an efferent copy of saccades actually executed, as it reliably follows all types of saccades made into the cell's postsaccadic movement field, even spontaneous saccades made in the dark or quick phases of nystagmus. An efferent copy of saccadic displacements, as coded by postsaccadic activity in FEF, is critical for several purposes. Interestingly, many FEF cells with presaccadic (visual, PSB, or both) activity often have postsaccadic activity for saccades directed opposite their presaccadic response fields. This reciprocity provides a mechanism for readily returning to the previous fixation (i.e., glances).

General suppression of presaccadic activities by saccade execution

A striking aspect of FEF presaccadic activity of all types (e.g., anticipatory, visual, mnemonic, and PSB) is that it quickly ends upon the execution of a saccade into the RF. Note in Figure 96.4 and in all the aggregate histogram figures that saccade execution actively ends the PSB shortly after the saccade, and often the postsaccadic rate goes well below baseline. Control experiments (Bruce, 1990) show postsaccadic suppression of both visual and mnemonic activities as well.

This suppression could come from the postsaccadic coding of prior saccades (efferent copy) just described. Such suppression is very important, because visual or mnemonic activity coding a peripheral cue location becomes misleading once the monkey foveates the peripheral location. Without prompt suppression, persistent activity could lead to multiple triggering of the same saccade, much like the “staircase” of saccades evoked by continued electrical stimulation in FEF. Bruce (1990) suggested that this suppression could be conceptualized as a single-step case of the efferent copy subtraction operation hypothesized to remap visual stimulus signals in FEF in the double-step paradigm, and thereby underlie double-step task behavior and physiology described earlier. Thus, the double-step behavior, the glance, and the virtue of not making multiple saccades in response to a single stimulus step may all reflect the same global transformation of FEF activity effected by the postsaccadic activity it receives following each saccade that is executed.