JN Track the topics, authors and articles important to you
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

J Neurophysiol 80: 282-296, 1998;
0022-3077/98 $5.00
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Eifuku, S.
Right arrow Articles by Wurtz, R. H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Eifuku, S.
Right arrow Articles by Wurtz, R. H.

The Journal of Neurophysiology Vol. 80 No. 1 July 1998, pp. 282-296
Copyright ©1998 by the American Physiological Society

Response to Motion in Extrastriate Area MSTl: Center-Surround Interactions

Satoshi Eifuku and Robert H. Wurtz

Laboratory of Sensorimotor Research, National Eye Institute, National Institutes of Health, Bethesda, Maryland 20892

    ABSTRACT
Abstract
Introduction
Methods
Results
Discussion
References

Eifuku, Satoshi and Robert H. Wurtz. Response to motion in extrastriate area MSTl: center-surround interactions. J. Neurophysiol. 80: 282-296, 1998. The medial superior temporal area of the macaque monkey extrastriate visual cortex can be divided into a dorsal medial (MSTd) and a lateral ventral (MSTl) region. The functions of the two regions may not be identical: MSTd may process optic flow information that results from the movement of the observer, whereas MSTl may be related more closely to processing visual motion related specifically to the motion of objects. If MSTl were related to such object motion, one would expect to see mechanisms for the segregation of objects from their surround. We investigated one of these mechanisms in MSTl neurons: the effect of stimuli falling in the region surrounding the receptive field center on the response to stimuli falling in the field center. We found the effects of the surround stimulation to be modulatory with little response to the surround stimulus itself but a clear effect on the response to the stimulus falling on the receptive field center. The response to motion in the center in the direction preferred for the neuron usually increased when the surround motion was in the opposite direction to that in the center and decreased when surround motion was in the same direction as that in the center. Fifty-seven percent of the neurons showed a ratio of response for center motion with a surround moving in the opposite direction to that in the center for center motion alone that was >1. The response to motion in the center also increased when the surround stimulus was stationary, and this increase was sometimes larger than that with a moving surround. Nearly 70% of the neurons showed a ratio of response to center motion with a stationary surround to center motion alone that was >1. This is in contrast to the minimal effect of stationary surrounds in middle temporal area neurons. When the stimulus presentation was reversed so that the stimulus in the center was stationary and the surround moved, some MSTl neurons responded when the direction of motion in the surround was in the direction opposite to the preferred direction of motion in the center of the receptive field. Stimulation of the surround thus had a profound effect on the response of MSTl neurons, and this pronounced effect of the surround is consistent with a role in the segmentation of objects using motion.

    INTRODUCTION
Abstract
Introduction
Methods
Results
Discussion
References

The medial superior temporal area (MST) lies within the superior temporal sulcus of the extrastriate cortex of the macaque monkey and has a large fraction of neurons that are directionally selective (Desimone and Ungerleider 1986; Tanaka et al. 1986; Van Essen et al. 1981). Komatsu and Wurtz (1988a) suggested that this area might actually comprise two regions because they found that the receptive fields of neurons in the dorsal medial region (MSTd) and in the lateral ventral region (MSTl) of MST both approached or included the fovea, whereas neurons intermediate between these two did not. Furthermore, they found that the size of the receptive fields in both regions of MST were larger than in the middle temporal area (MT) but that MSTd neurons responded better to large patterns of moving stimuli, whereas neurons in MSTl responded as well or better to the motion of single spots of light. This hypothesized separation based on receptive field eccentricity, size of receptive fields, and type of preferred stimulus recently has been strengthened by the quantitative study of Tanaka et al. (1993) who showed that if the ratio of the field size to eccentricity is considered, the distinction between MSTd and MSTl becomes even clearer.

The functional characteristics of cells in these two areas also suggested that their contributions to behavior may differ. Saito et al. (1986) first noted that the neurons in MST that respond to rotating, expanding, or contracting stimuli were located in one region of MST that subsequently has become identified with MSTd (Komatsu and Wurtz 1988a; Tanaka et al. 1993). These properties might be appropriate for analyzing the motion that results from movement of an observer through the environment (Duffy and Wurtz 1991a; Tanaka et al. 1986), and such optic flow stimuli might contribute to the determination of heading, the control of posture, and the structure of the environment (Gibson 1950). In contrast, Tanaka et al. (1993) suggested that the neurons in the lateral ventral region of MST responding best to smaller stimuli were better suited for the analysis of object motion.

The relation of MSTd neuronal activity to optic flow stimulation has been extensively studied. MSTd neurons respond to planar, radial, and circular motion, which are components of optic flow (Andersen et al. 1990; Duffy and Wurtz 1991a,b; Saito et al. 1986; Sakata et al. 1986; Tanaka and Saito 1989; Tanaka et al. 1989; Wurtz et al. 1990); they respond to changes in stimulus position (Andersen et al. 1990; Duffy and Wurtz 1991b; Graziano et al. 1994; Lagae et al. 1994), to the speed of flow components (Duffy and Wurtz 1997; Orban et al. 1995; Tanaka and Saito 1989; Tanaka et al. 1989), and to combinations of flow components (Duffy and Wurtz 1996; Graziano et al. 1994; Lagae et al. 1994; Orban et al. 1995); they change their responses when the centers of motion of the optic flow are shifted to different parts of the visual field to simulate different headings of observer movement (Duffy and Wurtz 1993); they partially compensate for the effect of pursuit eye movements (Bradley et al. 1996). Taken together, these studies provide substantial evidence that the characteristics of MSTd neurons are appropriate for the analysis of motion generated by an observer's own movement and that this information could contribute to determining heading, posture, and environmental structure.

In contrast, a role for MSTl in the processing of object motion has been supported largely by two observations. First single-cell recording (Komatsu and Wurtz 1988a) indicated that the responses of neurons in MSTl were related closely to the generation of smooth pursuit movements, to the motion processing underlying such eye movements, and to the nonvisual input required by such a system (Erickson and Thier 1991; Newsome et al. 1988; Thier and Erickson 1992). Because these pursuit movements are made to follow moving objects, this view is consistent with the idea that MSTl might be more generally related to object motion. Second, Tanaka et al. (1993) more recently showed that some cells in MSTl respond best when one moving stimulus moves in front of and occludes a large background stimulus, which also is consistent with their role in object motion.

One of the ways in which neurons can distinguish objects from the background is by comparing motion in the center of their receptive fields to motion in the surrounding region of the field. We reasoned that if MSTl neurons were involved in the analysis of object motion, they should have a clear center and surround structure that would allow the motion of an object to stand out against a background. In the present experiments, we tested this hypothesis and found modulation of the response in the center of the receptive field by surround motion. We also found that the effect of a stationary surround was sometimes stronger than that of a moving surround and that this effect of the stationary surround was substantially stronger in MSTl than in area MT.

Brief reports of these experiments have appeared previously (Eifuku and Wurtz 1995, 1996).

    METHODS
Abstract
Introduction
Methods
Results
Discussion
References

Physiological and behavioral procedures

We studied areas MSTl in two adult male rhesus monkeys (Macaca mulatta; identified as OR and SA) weighing 8-11 kg. The monkeys were prepared for recording in a single surgical session using procedures described previously (Duffy and Wurtz 1995). Under general anesthesia, scleral search coils were implanted bilaterally (Judge et al. 1980), recording cylinders were placed over parietal cortex bilaterally, and a head holder was embedded in a dental acrylic cap that covered the top of the skull. Postoperative analgesia was administered as judged appropriate by the attending veterinarian. All hardware was compatible with magnetic resonance imaging (MRI): cylinders and head holders were plastic and screws in the skull were titanium. Perforated titanium strips also were used to anchor the dental acrylic cap to the bone with three or four titanium screws in each strip. All experimental protocols were approved by the Institute Animal Care and Use Committee and complied with Public Health Service Policy on the humane care and use of laboratory animals.

During the experiment, the monkey sat in a primate chair with its eyes 58-cm away from the center of a 100 × 100° tangent screen. Each trial began with the appearance of a spot of light (0.3° in diameter) at the center of the screen. The monkey's task was to fixate the spot within 500 ms of the onset and maintain fixation within a 4 × 4° window during visual stimulation. We used a relatively large fixation window because on some trials with large field stimulus motion, the monkey had difficulty maintaining fixation due to ocular following (Miles et al. 1986), and while we excluded such data after the eye began to move (see Data analysis), we did not want the monkeys to be frustrated by repeatedly truncated fixation periods. Eye position was monitored using the magnetic search coil technique (Robinson 1963). If the monkey maintained fixation until the end of a trial, a reinforcing tone was sounded, and the monkey received a liquid reward on a variable ratio reinforcement schedule (20-100% probability). Failure to maintain fixation aborted the trial, and results of that trial were discarded. The task was adjusted to maintain a high success rate during training; the monkeys attained 95% correct performance after a few weeks. The monkeys performed the task for several hours per day and then were returned to their home cages. Records were kept of the weight and health status of the monkeys, and supplemental fruit and water were provided.

The visual stimuli projected onto the screen while the monkey fixated were random dot patterns that were generated on-line and displayed using a Pentium-based computer and a Texan graphics card with a resolution of 640 × 480 pixels. They were back projected onto a translucent screen by a television projector. Each random dot subtended 0.6°, and they were spaced 0.6° between centers. The dots were 1.8 cd/m2, and the background was 0.2 cd/m2, which was identical to the intensity used in recent studies of MSTd in this laboratory (Duffy and Wurtz 1995). The random dot pattern used for stimulation was generated for each session and had 90% dark and 10% light areas in the pattern. The dot pattern was static, and the whole pattern usually moved within an aperture. The fixation point was generated by the same projector. The computer generating the visual stimulation was controlled by the standard laboratory real time experimental system REX (Hays et al. 1982) running on a dedicated 80486-based computer.

Two TV projectors were used, initially an Electrohome ECP 4000 that used three cathode ray tubes to project the image and then a Sharp 850 that projected an image turned on and off by liquid crystal displays (LCD). Both were running at 60-Hz frame rates. The projectors were synchronized to our computer by the vertical retrace signal which for the Electrohome gave an accurate indication of stimulus onset and movement. In the LCD projector, however, there was a fixed phase lag of 4 ms in the onset of the projected image and a variable one of between 0 and 16 ms that produced a mean stimulus delay of 12 ms. Although we did not measure or report visual or response latencies, we did set the window for unit response measurement to include the earliest visual responses, and we shifted the responses studied using the LCD projector by 12 ms. This applied almost exclusively to the MT data; all figures are from experiments using the Electrohome projector.

The behavioral task, stimulus timing, storage of single cell activity, and eye position were controlled by REX. Single neuron activity was digitized using a window discriminator, sampled at 1 kHz, and stored with markers of stimulus and behavioral events. An on-line raster display showed the occurrence of single neuron discharges aligned on stimulus and behavioral events during the experiment. Eye position was monitored by REX for behavioral control during all experiments and also was stored.

Recordings were made in both hemispheres of the two monkeys from cylinders placed 16- to 17-mm lateral from midline and 2.0- to 3.5-mm posterior to earbar zero on the stereotaxic. Penetrations were made in the vertical plane. A grid was placed within the recording cylinders (Crist et al. 1988) to facilitate the insertion of stainless steel guide tubes through the dura to a depth ~10 mm above the superior temporal sulcus. At the beginning of each recording session, a guide tube stylet was removed and an epoxy coated tungsten microelectrode (Microprobe, 1.0-1.4 MOmega at 1 kHz) was inserted. The electrode was advanced using a stepping microdrive while neuronal activity was monitored to establish the relative depth of landmarks, including gray and white matter layers and neuronal response properties.

Experimental sequence

For each cell isolated, we did two preliminary tests. First, the size and location of the excitatory receptive field (RF) region, which we will refer to as the RF center, was mapped by a mouse-controlled stimulus during a visual fixation task. For this purpose, five kinds of stimuli were used: a 1.1° diam spot, a 3.6° diam spot, a 10.5 × 10.5° random dot field, a 21 × 21° random dot field, and a 33.6 × 33.6° random dot field. The RF center was drawn on a tracing made on a monitor that duplicated the stimulus seen by the monkey. In most cases, mapping was done using only the small spot, but this varied with the cell. Second, the optimal speed and direction of motion across the RF center was estimated using the computer-controlled motion of the best stimulus found in the mapping of the RF center. Eight directions (0, 45, 90, 135, 180, 225, 270, and 315°---0 to the right) and five speeds (6, 10, 20, 40, and 80°/s) were tested. Table 1 shows the optimal speeds preferred by the neurons for which a full range of speeds was tested.

 
View this table:
[in this window] [in a new window]
  		TABLE 1. Number of neurons preferring each tested speed

We studied the center-surround interaction of each neuron by presenting stimuli to the RF center in combination with large surround stimuli. For the center stimulus, we used a random dot field that filled the RF center that had been mapped. When the dots within this field moved, they moved en bloc and at the optimal speed. The stimuli moved in the direction that gave the best response (preferred direction) and in the opposite direction to the preferred direction (anti-preferred direction). The surround stimulus was a random dot pattern of the same density as the center stimulus (60 × 60° or 100 × 100°) and was moved in the same two directions as the center stimuli. We then presented combinations of stationary stimuli and moving stimuli in the center and surround. The presentation of all these stimuli was randomly interleaved.

Figure 1 shows the sequence of stimulus presentations. After looking at the fixation point (FP in Fig. 1, A1 and B1) for 400-800 ms, the visual stimulus appeared as a stationary random dot field (Fig. 1, A2 and B2), and after 800 ms the dots moved for 400 or 600 ms (Fig. 1, A3 and B3). Inserting the delay between the stationary and moving stimuli dissociated any response to stimulus motion from that to stimulus onset.


View larger version (48K):
[in this window]
[in a new window]
  		FIG. 1. Visual stimulation procedures. A: 3 periods of each fixation trial: fixation (1), stationary stimuli (2), and moving stimuli (3). In this and subsequent figures, the dark area indicates the background illumination level of the television projector, and the white areas show the bright regions of the stimulus. Size and eccentricity of the center stimuli was set to equal the receptive field (RF) center, and the surround was 60-100° on a side. Random dot pattern used for illustration does not exactly replicate the pattern described in METHODS. B: time sequence of the task. After looking at the fixation point (FP) for 400-800 ms, the stationary visual stimuli appeared. After 800 ms of the stationary stimulation (to separate stimulus onset from stimulus motion), the stimulus moved for 400 or 600 ms. Period for counting spikes was the 100-ms period that began 70 ms after motion onset for MSTl neurons (black-square).

The neurons studied were those that had the characteristics of cells in MSTl described previously in this laboratory (Komatsu and Wurtz 1988a). Neurons responded preferentially to moving stimuli, were directionally selective, had RF centers with a medial edge close to the fovea, had relatively large receptive fields (see Fig. 2), and responded to single spots of light as well or better than to the motion of random dot patterns. We used this latter characteristic, their preference for the motion of small spots rather than random dot patterns, to differentiate MSTl neurons from MSTd neurons. We also studied a few MT neurons for comparison with those in MSTl. We usually could identify the shift in the recording as the electrode moved from MST on the anterior bank of the superior temporal sulcus to MT on the posterior bank. The MT neurons were also distinguishable from MSTl neurons by the smaller range of sizes of their RF centers at the same eccentricity. Figure 2A shows the size-eccentricity relation of the RF center for the MSTl neurons in this sample, and Fig. 2B shows the same for the smaller sample of MT neurons.


View larger version (22K):
[in this window]
[in a new window]
  		FIG. 2. Size-eccentricity relation of excitatory RFs. Horizontal axis is the eccentricity of the RF which is defined as the distance between the center of RF and the FP, and the vertical axis is the size of RF, which is defined by the square root of the RF area.

We initially located the general region for recording within the superior temporal sulcus with the aid of a MRI of the brain after the cylinder was implanted but before experiments began. The images were made with several tungsten electrodes (but not the guide tubes) cemented into place on the grid while the monkey was held in a nonferrous stereotaxic instrument under narcotic and ketamine anesthesia. We used parasagital MRI sections at 0.5-mm intervals to locate the superior temporal sulcus and the approximate locations of MST and MT. The criteria for categorizing the neurons studied were the physiological ones described above. In addition, at the end of the experiments, the location of the electrode tracks were determined from the histological sections of both monkeys using the histological procedures described previously (Duffy and Wurtz 1995). We found that the electrode penetrations passed through the regions of the superior temporal sulcus indicated by the MRI.

Data analysis

Because the monkeys in this experiment were awake, large field stimulus motion produced the ocular following responses reported previously (Miles et al. 1986). The presence of a fixation point on all trials reduced this eye movement, but even this did not eliminate motion produced by eye movements as shown by the experiments in MSTd (Komatsu and Wurtz 1988b). To remove any possible effect of such eye movements in our data analysis, we quantified the response of the neurons before the earliest time that we observed the eye to move in our experiments. Using the velocity of the eye movements, we estimated this earliest movement to occur about 100 ms following stimulus motion of the size and contrast used in our experiments. This is considerably longer than that reported by Miles et al. (1986), and this was probably due to the different conditions in our experiment, which included motion of less than a full visual field while the monkey fixated instead of motion of the full field at the time of maximum sensitivity for ocular following just after a saccade. We allowed an additional time for visual latency that we also obtained from our sample of neurons using our stimuli. For MSTl neurons, this was a minimum of ~70 ms, and for MT, it was a minimum of ~60 ms. We therefore counted the spikes in the period between 70 and 170 ms after stimulus onset for MSTl and between 60 and 160 ms for MT. The disadvantage of this procedure was that it produced short sampling periods and therefore higher variance in the response magnitude, but the advantage was that in the analysis we could be confident that we were seeing the visual response to the stimuli we presented uncontaminated by the visual consequences of the monkey's eye movements. For comparison of neuronal response magnitudes to different visual stimulus configurations, a two-tailed t-test was used with a significance level of P < 0.01. Off-line data analysis used spike density histograms that were created by replacing the spikes with Gaussian pulses with a width corresponding to a standard deviation of 10 ms using the method of MacPherson and Aldridge (1979) as implemented by Richmond et al. (1987).

    RESULTS
Abstract
Introduction
Methods
Results
Discussion
References

Effect of surround stimuli on the response to center motion for MSTl neurons

We recorded from 191 MSTl cells in four hemispheres of two monkeys. For each neuron, we compared the response to motion in the center of the excitatory RF alone with the response to such motion coupled with visual stimulation in the surrounding visual field. Figure 3 shows an example of a MSTl cell the response of which was modified by surround stimuli. The neuron responded to motion of random dots moving in the optimal direction and at the optimal speed of the cell (Fig. 3A1). With the addition of surround motion in the same direction as that in the center, the visual response was reduced (Fig. 3A2), but with surround motion in the opposite direction to that of the center, the response was increased (Fig. 3A3). This is similar to many of the center-surround interactions reported by Allman et al. for area MT of owl monkeys (Allman et al. 1985a,b). But the response of this MSTl neuron also increased even if the surround was stationary (Fig. 3A4), and the increased response was as great with the stationary surround as with the moving surround.


View larger version (73K):
[in this window]
[in a new window]
  		FIG. 3. An example of the increased response to center motion with both moving and stationary surrounds for a MSTl neuron. A: center motion in the preferred direction. Top row: visual stimulus configuration; the area of the visual field shown was 60 × 60°, and the fixation point is indicated by FP. From left, center motion only (1), center motion with the surround moving in the same direction as the center (2), center motion with the surround moving in the opposite direction (3), and center motion with the surround stationary (4). All responses were significantly different from the response to center stimulus motion alone (Student's t-test, P < 0.01). Location and size of the center is that estimated by initial mapping. The center stimulus was 11.5 × 11.5°, and the surround stimulus was 60 × 60°. Preferred direction of the cell was 315° (0° to the right), and the optimal speed was 40°/s. Second row: spike density function (SD = 10 ms); the height of the vertical line at stimulus onset (0) was 40 spikes·s-1·trial-1. Dark bar below the spike density plot indicates the stimulus period. Third row: rasters of unit firing for 14 trials. Fourth and fifth rows: horizontal and the vertical eye traces, respectively. Records are aligned on motion onset; the height of the vertical line at stimulus onset is 5°; upward to the right, downward to the left. Response to center motion increases both with the surround moving in the opposite direction preferred by the center and with the stationary surround. B: center motion in the antipreferred direction. Note the increased response with the stationary surround. C: presentations of center and surround separately. From left, the center motion in the preferred only (1), the surround motion in the same direction (2), the center motion in the anti-preferred direction (3), and the surround motion in the same direction (4). Note the lack of response to the surround motion alone.

Figure 3B shows the neuronal responses when the center was moving in the antipreferred direction. There were only small responses, but the responses to the center motion with the stationary surround were increased slightly compared with the center motion alone (Fig. 3, B4 compared with B1). Figure 3C shows separate center or surround stimulation in each direction and shows that the surround stimulation alone elicits minimal if any response. The increased or decreased response of this neuron with surround stimulation therefore results from modulation of the response to motion in the RF center by a silent surround.

Figure 4 shows another example of a MSTl cell with increased response to center motion in the presence of a stationary surround. In this case, when the center moved in the preferred direction and the surround moved in the opposite direction (Fig. 4A3), there was a slight but significant increase in the response as was the case in the example in Fig. 3. But when the surround was stationary (Fig. 4A4), the increase in the visual response was greater than when the surround was moving so that the stationary surround was more effective than the moving surround.


View larger version (69K):
[in this window]
[in a new window]
  		FIG. 4. Increased response to center motion with stationary surround for a MSTl neuron. Figure descriptions are the same as in Fig. 3. Center stimulus was ~6 × 6°, and the surround stimuli was 60 × 60°. Preferred direction of the cell was 270°, and the optimal speed was 40°/s. Visual response to the center motion in the preferred direction (A) was greater with the stationary surround than the moving surround, although both were statistically significant (P < 0.01)

Note that clear ocular following can be seen from the eye position traces in Fig. 4A following motion of the large surround stimulus. In this case, the eye moved ~2° during 400 ms and generated eye movements of ~5°/s. Actual motion of the stimulus was therefore close to 35°/s rather than 40°/s, and this motion would also shift the random dot pattern off the receptive field somewhat. As indicated in METHODS, we used only the period before the eye started to move for quantifying the responses. Note also that the eye movement record shows that the vertical eye position trace is displaced downward about a degree when the monkey was fixating against a random dot background (Fig. 4A, 2-4---moving or stationary backgrounds) compared with fixation against a uniform dark background (Fig. 4A1, left column). This was consistently the case for this monkey (as can be seen in Fig. 3 as well), but such a shift did not occur for the other monkey. Because the effects of the surround stimulus were the same in both monkeys and because the effects of the moving and stationary surrounds were compared quantitatively to each other rather than to the dark background, we do not think this offset of eye position materially affects the results.

The relative strength of the modulation by moving and stationary surrounds varied across the cells, and Fig. 5A shows the strength of this modulation for each MSTl neuron. The scatter plot compares the ratio of the responses with and without the surround (response to center and surround/response to center only) when the surround was stationary (abscissa) and when it was moving in the direction opposite to that in the center (ordinate). Neurons falling below the dashed diagonal line had stronger responses with stationary surrounds than with moving surrounds.


View larger version (13K):
[in this window]
[in a new window]
  		FIG. 5. Comparison of modulation by stationary and moving surrounds for MSTl neurons (A) and middle temporal area (MT) neurons (B). Abscissa shows the ratio: response to center motion in the preferred direction with stationary surround/response to center motion in the preferred direction only. Ordinate shows the same ratio but with the surround moving in the opposite direction to that of the center. Each symbol represents a neuron. Diagonal dashed line in A indicates equal responses to both moving and stationary surrounds for the MSTl neurons. Vertical dashed line in B shows the values when the stationary surround has no effect. Modulation by the stationary surround is common for MSTl neurons as indicated by the number of neurons above the diagonal dashed line in A but is unusual for the MT neurons as indicated by the number of points falling near the dashed vertical line.

Figure 6, A-C, shows the strength of modulation of these MSTl neurons with the different surround stimuli. With stationary surrounds, nearly 70% of the neurons showed a modulation ratio (center with surround/center only) >1, whereas with the surround moving in the opposite direction, 57% of the neurons had a ratio >1. Thus the MSTl neurons showed more frequent response increases with stationary surrounds than with moving surrounds. When the surround moved in the same direction as the center, only ~20% of the neurons had ratios >1 and >60% had <1.


View larger version (27K):
[in this window]
[in a new window]
  		FIG. 6. Frequency of modulation effect on MSTl (A-C) and MT (D-F) neurons. On the abscissa is a modulation ratio indicating the mean neuronal activity to the center motion in the preferred direction with different motions in the surround. Columns show percent of cells for each ratio. A and D: stationary surround. B and E: surround moving in the opposite direction to the center. C and F: surround moving in the same direction as the center. With moving surrounds (B, C, E, and F), both MSTl and MT neuron samples behaved similarly, but with stationary surrounds (A and D), MSTl neurons were modulated more clearly than were MT neurons.

Stationary surround effect for MT neurons

The increase in the response to center motion with stationary surround stimulation in the MSTl neurons appears to be stronger than has been reported previously for the effect of such surrounds in MT. Addition of motion in the surround in the direction opposite to the direction preferred by the center in MT neurons frequently led to larger responses than those to center motion alone, but stationary surround stimuli had minimal effect (Olavarria et al. 1992). To verify that this difference in the effect of the stationary surround between MSTl and MT was also present in the awake monkey, we recorded from 41 neurons in MT in two monkeys using the same visual stimulation procedures that we used in MSTl. The major difference was that the center stimulus sizes were smaller because the receptive field centers were substantially smaller in MT (see Fig. 2).

Figure 7 shows an example of a MT neuron that showed an increased response to motion in the center with motion in the surround opposite to the direction preferred by the center (Fig. 7A, 1 and 3). In contrast, when the surround was stationary, there was not such a clear increase (Fig. 7A, 1 and 4). This was a consistent finding as shown in the scatter plot in Fig. 5B. A surround moving in the direction opposite to the preferred direction did modify the response to motion in the center of the field (Fig. 5B, ordinate) but a stationary surround had little effect (Fig. 5B, abscissa; the dashed vertical line is shown for reference). This is in contrast to the nearly equal modulation of many MSTl neurons by moving and stationary surrounds shown in Fig. 5A.


View larger version (84K):
[in this window]
[in a new window]
  		FIG. 7. Lack of a stationary surround effect on a MT cell. Figure descriptions are the same as in Fig. 3. Center stimulus was ~4.5 × 4.5°, and the surround stimuli was 60 × 60°. Preferred direction of the cell was 45°, and the optimal speed was 40°/s. Response to motion in the center was affected minimally by the stationary surround, but motion in the surround opposite to that preferred in the center had a clear effect.

This comparison between the surround effects in MSTl and MT becomes clearer with the comparison of the percent of neurons that show different modulation effects with different surrounds (Fig. 6, D-F). For stationary surrounds (Fig. 6D), MT cells were centered close to a ratio of 1, whereas MSTl neurons showed more positive ratios (Fig. 6A). With moving surrounds, MSTl and MT neurons did not show such a difference (Fig. 6, B and C compared with E and F). We conclude that there is less modulation of the response to motion in the center by a stationary surround stimulus in MT neurons than in MSTl neurons.

Effect of the center stimulus size on MSTl neurons

One factor affecting the interaction of center and surround motion might be the extent to which the center of a neuron's RF was filled by the center stimulus, the extent to which the center stimulus overlapped the surround region or the blurring of the edges between the center and surround regions. To determine whether the placement of the center stimulus was a critical factor in the results of these experiments, we used center stimuli of several different sizes while keeping the size of the outside dimensions of the surround stimulus the same. For 41 MSTl cells that gave significant increases in the response to stationary surrounds (t-test, P < 0.01), we used three to five sizes of center stimuli. Figure 8A shows an example of a MSTl cell that showed an increased response with a stationary surround when the center motion filled the estimated size of the RF center. Four stimulus sizes were used for the center motion: the size that filled the excitatory RF as well as one-fourth, one-half, and twice that size. When the center stimulus presented alone equaled the size of the estimated excitatory RF, the neuron gave the largest response, indicating that the original estimate of the RF center size was reasonable (Fig. 8Aa). With the addition of the stationary surround (Fig. 8Ab) or the moving surround (Fig. 8Ac), there was a clear increase in response that was greatest when the center stimulus matched the estimated size of the RF center. Modulation was greater with the stationary surround. The increased response with the stationary stimulus persisted even when the center stimulus was twice or half the estimated center size. Figure 8B quantifies these responses and shows that the stationary surround was strongest when the center stimulus approximated the estimated size of the excitatory receptive field center, but it was not abolished when that size increased or decreased somewhat.


View larger version (29K):
[in this window]
[in a new window]
  		FIG. 8. Influence of changes in the center size on neuronal responses in MSTl cells. A: in each column, the neuronal activity is shown by a spike density function (SD = 10 ms) aligned on start of motion. Each row corresponds to the stimulus configuration depicted on the left. Aa: center only; Ab: center with the stationary surround; Ac: center with the surround moving in the opposite direction. In each row, the center size was, from left, one-fourth the size of the excitatory RF region (×1/4), one-half the size of the excitatory RF region (×1/2), the size of the excitatory RF region (×1), and double the size (×2). Preferred direction of the cell was upward, and the optimal speed was 40°/s. All center stimuli regardless of size were centered on the estimated center of the RF. B: quantification of the responses shown in A. Note both modulation by the stationary surround and modulation by the surround moving in the antipreferred direction had their peak around ×1. C: mean of the MSTl cells tested with the 3 sizes of the center stimulus (×1/2, ×1, and ×2, n = 32). Three curves had their peaks around size ×1.

We tested the effect of varying center stimulus size between one-fourth and four times the estimated RF center size for 41 MSTl neurons, and Fig. 8C shows the results for the 32 neurons that had at least the one-half and two-times size tests. The stationary surround was most effective when the stimulus size most closely approximated the estimated size of the RF center, whether the surround was stationary or moving in the direction opposite that preferred by the center. We conclude from this sample that errors in setting the size of the center stimulus are unlikely to account for the effects of the surround since making the size either larger or smaller only decreased the surround effect.

Response to stationary center and moving surrounds in MSTl neurons

Tanaka et al. (1993) showed that neurons in MSTl responded with motion of the surround stimulus even when the stimulus falling on the center of the RF was stationary. We also tested this effect of a stationary stimulus in the center of the field on 101 neurons, and we also found such modulation in some neurons. Figure 9 shows an example of such modulation. Like the neurons that we have already considered, this neuron responded to motion in the center of the RF, and this response was stronger in the presence of a stationary surround (Fig. 9A, 1 compared with 4). It did not respond to motion of the surround only that was in the same direction as that preferred in the center (Fig. 9A6). The most interesting responses of the neuron, however, were when the motion in the surround was in the direction opposite to that preferred for motion in the center (Fig. 9B). When the center motion was in the antipreferred direction, there was little response (Fig. 9B4), but when the surround motion was in this antipreferred direction while the stimulus in the center remained stationary, there was a clear response (Fig. 9B5). This response was even present when the center had no textured stimulus but was just at the background level of luminance (Fig. 9B6). Thus a moving stimulus in the surround, which produced little response by itself, gave a strong response when combined with a stationary stimulus in the center, which also produced little response by itself (indicated by the lack of response to the left of the trigger line indicating onset of stimulus motion).


View larger version (60K):
[in this window]
[in a new window]
  		FIG. 9. Response of a MSTl neuron to a moving surround stimulus and a stationary center stimulus. Figure descriptions are the same as in Fig. 3. Neuron responded when a stimulus moved upward in the center of the RF (A1-A4) but also when the center stimulus was stationary and the surround moved downward (B5 and B6). A: center with motion in the preferred direction or with no motion. From the left: center motion only (1, dark surround); center motion with random dot surround motion in the same preferred direction as the center (2); center motion with the surround motion in the opposite direction (3); center motion with stationary surround (4); center stationary with motion of the surround opposite to the preferred direction of center motion (5); like 5 but with no pattern (background luminance) in the center (6). Center stimulus was 9.0 × 9.0°, and the surround stimulus was 60 × 60°. Preferred direction of the cell was 90°, and the optimal speed was 5°/s. B: legends as in A. Center with motion in the antipreferred direction or with no motion.

We encountered neurons with such a clear response to stationary center and moving surround only infrequently. Figure 10 indicates this frequency in our sample of neurons by using a modulation ratio shown schematically (inset): the response to the stationary center with motion in the surround opposite to the preferred direction in the center divided by the response to motion in the center in the preferred direction. The larger the ratio, the larger the response to the stationary center and moving surround configuration. For both the textured center and dark uniform center, few neurons showed large ratios: only 4 and 3% of the neurons showed responses greater than one in Fig. 10, A and B, respectively. On the other hand, because we would expect virtually no response in the numerator, even the smaller fractional values of the ratio indicate some response to the center stationary and surround moving combination. If we consider modulation ratios with values above ~0.25 (those values below were influenced by the change of only a few spikes/second), a larger fraction of the neurons would be included (48 and 34% in Fig. 10, A and B, respectively).


View larger version (28K):
[in this window]
[in a new window]
  		FIG. 10. Frequency of neurons responding to stationary center stimulus with moving surround. Modulation ratio is the response of the stationary center with motion of the surround in the direction opposite to that preferred for motion in the center (illustrated in the inset) divided by the response to motion in the preferred direction in the center. The larger the positive ratio, the larger the relative response to the stationary center and moving surround configuration; a ratio of 1 indicates equal responses to both stimulus configurations A: frequency distribution of the modulation index when the center stimulus was a stationary random dot pattern. B: when the center stimulus was uniform at background luminance. N = 101 neurons.

Because we had determined the response of these neurons to both motion in the center with stationary surround and the reverse, we could compare the relative strengths of these two modulations of activity in the same neuron. The scatter plot in Fig. 11 compares the responses of each neuron to motion in the center with a stationary surround (ordinate) to the response to motion in the surround with a stationary stimulus in the center (abscissa). The plot shows relative motion since the motion in the center was in the preferred direction for each neuron at its optimal speed, and the motion in the surround was in the antipreferred direction and optimal speed for motion in the center. There was a slight tendency for neurons that had the strongest responses to motion in the center also to have the strongest response to motion in the surround. The effect of motion in the center was almost always stronger than that in the surround as indicated by the number of points falling above the line of equal response.


View larger version (46K):
[in this window]
[in a new window]
  		FIG. 11. Comparison of the responses of the same neuron to motion in the center with stationary surround (ordinate) with the response to motion in the surround with a stationary stimulus in the center (abscissa). Motion in the center was in the preferred direction and speed for each neuron, and motion in the surround was at the same speed in the center's antipreferred direction so that the relative motion between center and surround were the same. Note that the scales on the axes are different; line indicates values of equality on the 2 axes. Effect of motion in the center was always stronger than that in the surround.

    DISCUSSION
Abstract
Introduction
Methods
Results
Discussion
References

We determined the extent to which responses of neurons in the MSTl changed when a stimulus in the region surrounding the excitatory center of the RF was added to a stimulus falling in the center of the field. We found striking effects from adding either moving or stationary surrounds on the response to the stimuli falling on the receptive field center. We will compare these interactions with those in other visual areas, consider their contribution to object motion, and discuss what these interactions indicate about the differences between MSTl and MSTd.

Comparison of MSTl to other monkey visual areas

The nature of the surround effect in MSTl is a modulatory one; the response to a stimulus in the center of the RF is altered by the presence of the surround stimulus, but the surround stimulus by itself produces a minimal response. Modulation by surround stimulation (reviewed by Allman et al. 1985a) has been observed at many levels in the visual system including cat visual cortex and superior colliculus (Rizzolatti et al. 1973, 1974), pigeon optic tectum (Frost et al. 1981), and monkey visual areas V1, V2 (Allman et al. 1990), MT (Allman et al. 1985b; Tanaka et al. 1986), and superior colliculus (Davidson and Bender 1991; Wurtz et al. 1980). It shares with the contextural stimulation observations in V1 of the monkey the modulation of response in the center of the RF by stimulation in the surrounding field beyond the center of the classical receptive field (Gilbert and Wiesel 1990; Zipser et al. 1996).

Comparison of the effects of surrounds between MSTl and MT is particularly relevant because MST receives a prominent afferent input from MT (Ungerleider and Desimone 1986; Van Essen and Maunsell 1983). The modulation in MSTl in the awake monkey is strikingly similar to that reported for area MT in the anesthetized monkey (Allman et al. 1985a,b; Tanaka et al. 1986). Surround motion in the direction opposite to that preferred by the center produces an increased response to the response to the center motion. Surround motion in the same direction as that preferred in the center usually reduces the response to motion in the center. The point that appears to be different between MSTl and MT is the effect of stationary surrounds. There is no indication in the work of Allman et al. (1985a) that a stationary surround is an effective stimulus for MT neurons, and this is confirmed by Olavarria et al. (1992), who found minimal modulation by stationary surround stimuli. In contrast, we found for many MSTl neurons that stationary surround stimuli produced the same increase, and occasionally even larger increases, in the response to motion in the center of the RF than did the surround moving in the opposite direction. We thought this might be due to differences in the experiments including the use of anesthesia, so we repeated our experiments on a small sample of neurons in MT, and we confirmed the previous reports that the stationary surround had minimal effect on the response to motion in the center of the RF of MT neurons. Thus the effectiveness of a stationary surround in MSTl seems to be genuinely different from that observed in MT.

The modulatory effect of surround motion also has been reported in V1 and V2 of the owl monkey (Allman et al. 1990). Some neurons in V1, but not V2, did show a suppression of the response to bar motion in the center of the RF in the presence of a stationary surround stimulus. There is no report of increased responses in the center of the field with stationary surrounds, but these results in V1 indicate that the effects of stationary surrounds are not unique to MSTl.

Neurons in the superior temporal sulcus project to the superior colliculus (Ungerleider et al. 1984), and neurons in the superficial layers of the colliculus also show a modulation of the responses in the RF center by the motion of surrounding stimuli. The colliculus neurons are quite different from the MSTl neurons: they do not show directionality in the center of their RF in the absence of any surround stimulus (Goldberg and Wurtz 1972; Schiller and Koerner 1971), but they do show such directionality in the presence of a surround stimulus (Davidson and Bender 1991), and they show only a suppressive surround effect. All three of these characteristics are different from the modulation in both MSTl and MT. The directional surround suppression in the monkey superior colliculus is similar to that seen in the pigeon optic tectum (Frost and Nakayama 1983); this is consistent with the possibility that the suppression in the optic tectum and superior colliculus is not dependent on an input from cerebral cortex, at least not from MT and MST.

Role of MSTl in object motion

The function of modulatory surrounds has been recognized universally as one of a number of mechanisms for the segregation of an object from its background---a segmentation of one part of the visual field from another (for reviews see Allman et al. 1985a; Nakayama 1985). The surround effect in MSTl shown in the present studies could certainly fill that role as has been proposed previously (Tanaka et al. 1993).

The consistent observation throughout these experiments is that many MSTl neurons respond better when motion differs in the receptive field center and surround. The response of 57% of the neurons increased with surround motion in the direction opposite to that in the center, and the response increased in nearly 70% of the neurons when the surround was stationary (Figs. 5 and 6). A smaller fraction of the neurons showed increased responses when the center was stationary and the surround moved in the direction opposite to that preferred by the center; this confirms the earlier observation of Tanaka et al. (1993). The effect of the moving surround on a stationary center was rarely as large as the modulatory effect of a stationary surround on the response to motion in the center (Fig. 11).

These modulations of the response of MSTl neurons all have two characteristic in common: the neurons all responded to changes in the relative motion between the stimuli in the center of the field and the surround and the direction of the relative motion was always in the direction preferred by the center. Such sensitivity to relative motion is exactly what would be required for a system involved in segmentation of an object from its background.

Note that this relative motion, which could contribute to the segmentation of objects, need not be involved specifically in determining the velocity of the object. However, we know that lesions of the MSTl area do lead to deficits in the maintenance of speed during pursuit eye movements (Dürsteler and Wurtz 1988; Dürsteler et al. 1987) and that microstimulation of the region alters this speed (Komatsu and Wurtz 1989). Determination of both the velocity tuning and the optimal relative motion of center and surround on the same neuron will be necessary for evaluating the contribution of these neurons to object speed and object segmentation.

Comparison of regions within MST

The present experiments indicate that the receptive field organization of neurons in MSTd and MSTl differ. We have found that neurons in the more lateral ventral region of MST (MSTl) respond to planar motion and have clear center-surround RF organizations; visual stimulation in the surround region modifies the activity resulting from visual stimulation of the RF center. The receptive fields of neurons in the more dorsal region of MST (MSTd) respond to the components of optic flow (including expanding and rotating stimuli in addition to planar motion) and have large receptive fields with less prominent center-surround organizations. These differences in the RF organization are consistent with the hypothesized regional specialization within MST. The compelling test of this distinction would be the demonstration in one monkey of a double dissociation between MSTl and MSTd: the selective response of MSTd neurons to large field optic flow motion and the absence of surrounds coupled with the insensitivity of MSTl neurons to such larger field motion and the presence of modulatory surrounds.

The division of MST into MSTl and MSTd probably is not as distinct as the division recently reported for MT. Born and Tootel (1992) have shown that MT in the owl monkey has a columnar organization that is visible as bands with 2-deoxyglucose staining. Neurons found within band regions show no indication of suppressive surrounds, whereas those in interband regions show suppression for large stimuli invading a surround. If this spatial separation of neurons in MT with and without the surround suppression were to be maintained in the projection to different regions of MST, the interbands in MT with cells having suppressive surrounds should project primarily to MSTl, and those without the surround suppression in the band regions should project more to MSTd. A preliminary report (Born et al. 1997) of these projections, however, indicates that the projections in the owl monkey may be just the opposite of this prediction, suggesting that the organization of the receptive fields in MST may not be so simply related to the band/interband input from MT.

We began these experiments with the idea that MSTl might be specifically related to the motion of objects in the environment and that MSTd might be devoted to the motion that results from the movement of the observer through the environment. Although we think that the present experiments expand on the differences between these two regions of MST, such a crisp separation of function is probably a substantial oversimplification based on what we now know about these areas. The neurons in MSTd have many characteristics that would make them appropriate for the analysis of the large field optic flow resulting from movement of the observer, and these characteristics are appropriate for the determination of heading as well as for the control of posture, as already outlined in the INTRODUCTION. However, this same optic flow sensitivity of the MSTd neurons also might contribute to the segregation of objects in the field because of the motion parallax resulting from the relative motion of a stationary object and its more distant background that results from movement of the observer. In addition, the optic flow sensitive neurons in MSTd could be activated by the optic flow patterns in the motion of objects within the field (Buracas and Albright 1996; Geesaman and Andersen 1996; Zemel and Sejnowski 1995). For example, a shearing motion (such as that produced by different speeds of motion at different distances from the observer) may contribute to the recognition of the tilt of objects within the visual field. The recent demonstration that neurons in MT are sensitive to the distribution of stimulus speeds (Treue and Andersen 1996; Xiao et al. 1997) and the heterogeneity of their surrounds (Raiguel et al. 1995; Xiao et al. 1995) suggests that MST neurons receiving input from MT also might respond to such flow patterns that would result from the relative motion of surfaces tilted in depth.

Similarly, the sensitivity of MSTl neurons to motion may not link them to just one function. The present studies show that the center surround organization could contribute to the separation of objects from the background and that this function could be part of a more general one of object segregation rather than one specifically devoted to object motion. In addition, previous experiments have demonstrated that the discharge of these neurons changes during pursuit eye movements (Erickson and Thier 1991; Newsome et al. 1988; Thier and Erickson 1992), and such pursuit depends in part on the velocity of motion of an object. Although both of these characteristics of MSTl neurons are related to object motion, their specific contribution might be quite different.

Thus although there are clear differences in the activity of neurons in the MSTl and MSTd regions, these differences might not be related simply to observer movement as opposed to object motion. MSTd might be more involved in processing optic flow information. This optic flow would largely be motion relative to the observer and would contribute to determining heading, controlling posture, and determining the structure of the environment, but it also could include the motion of large objects moving independently of the observer. MSTl, in contrast, might be more involved in object motion both for segmentation of objects from the background and for the control of movement such as pursuit eye movements.

    ACKNOWLEDGEMENTS

  We are grateful to J. W. McClurkin, who developed the real time visual display system (VEX) and modified it for the purposes of these experiments, and to M. Smith for histology. The Laboratory of Diagnostic Radiology Research provided the MRIs of the monkeys as well as generous advice and assistance. We also are grateful to anonymous reviewers for insightful suggestions.

    FOOTNOTES

  Address for reprint requests: R. H. Wurtz, Lab. of Sensorimotor Research, National Institutes of Health, National Eye Institute, Bldg. 49, Room 2A50, Bethesda, MD 20892.

  Received 18 August 1997; accepted in final form 6 March 1998.

    REFERENCES
Abstract
Introduction
Methods
Results
Discussion
References
0022-3077/98 $5.00 Copyright ©1998 The American Physiological Society



This article has been cited by other articles:


Home page
 Cereb CortexHome page
S. Pitzalis, M.I. Sereno, G. Committeri, P. Fattori, G. Galati, F. Patria, and C. Galletti
Human V6: The Medial Motion Area
Cereb Cortex, February 1, 2010; 20(2): 411 - 424.
[Abstract] [Full Text] [PDF]

Home page
 J. Neurophysiol.Home page
K. Amano, B. A. Wandell, and S. O. Dumoulin
Visual Field Maps, Population Receptive Field Sizes, and Visual Field Coverage in the Human MT+ Complex
J Neurophysiol, November 1, 2009; 102(5): 2704 - 2718.
[Abstract] [Full Text] [PDF]

Home page
 J VisHome page
D. H. Baker and E. W. Graf
Equivalence of physical and perceived speed in binocular rivalry
J Vis, April 1, 2008; 8(4): 26 - 26.
[Abstract] [Full Text] [PDF]

Home page
 Cereb CortexHome page
W. K. Page and C. J. Duffy
Cortical Neuronal Responses to Optic Flow Are Shaped by Visual Strategies for Steering
Cereb Cortex, April 1, 2008; 18(4): 727 - 739.
[Abstract] [Full Text] [PDF]

Home page
 Cereb CortexHome page
D. J. Logan and C. J. Duffy
Cortical Area MSTd Combines Visual Cues to Represent 3-D Self-Movement
Cereb Cortex, October 1, 2006; 16(10): 1494 - 1507.
[Abstract] [Full Text] [PDF]

Home page
 J. Neurophysiol.Home page
F. V. Barthelemy, I. Vanzetta, and G. S. Masson
Behavioral Receptive Field for Ocular Following in Humans: Dynamics of Spatial Summation and Center-Surround Interactions
J Neurophysiol, June 1, 2006; 95(6): 3712 - 3726.
[Abstract] [Full Text] [PDF]

Home page
 J. Neurophysiol.Home page
A. L. Williams, K. D. Singh, and A. T. Smith
Surround Modulation Measured With Functional MRI in the Human Visual Cortex
J Neurophysiol, January 1, 2003; 89(1): 525 - 533.
[Abstract] [Full Text] [PDF]

Home page
 J. Neurophysiol.Home page
R. T. Born
Center-Surround Interactions in the Middle Temporal Visual Area of the Owl Monkey
J Neurophysiol, November 1, 2000; 84(5): 2658 - 2669.
[Abstract] [Full Text] [PDF]

Home page
 J. Neurophysiol.Home page
S. Eifuku and R. H. Wurtz
Response to Motion in Extrastriate Area MSTl: Disparity Sensitivity
J Neurophysiol, November 1, 1999; 82(5): 2462 - 2475.
[Abstract] [Full Text] [PDF]

Home page
 J. Neurosci.Home page
K. Zhang and T. J. Sejnowski
A Theory of Geometric Constraints on Neural Activity for Natural Three-Dimensional Movement
J. Neurosci., April 15, 1999; 19(8): 3122 - 3145.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Eifuku, S.
Right arrow Articles by Wurtz, R. H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Eifuku, S.
Right arrow Articles by Wurtz, R. H.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Visit Other APS Journals Online