The Journal of Neurophysiology Vol. 77 No. 2 February 1997, pp. 554-561
Copyright ©1997 by the American Physiological Society

Simulated Optic Flow and Extrastriate Cortex. I. Optic Flow Versus Texture

Jong-Nam Kim, Kathleen Mulligan, and Helen Sherk

Department of Biological Structure, University of Washington, Seattle, Washington 98195-7420

	ABSTRACT

Abstract Introduction Methods Results Discussion References

Kim, Jong-Nam, Kathleen Mulligan, and Helen Sherk. Simulated optic flow and extrastriate cortex. I. Optic flow versus texture. J. Neurophysiol. 77: 554-561, 1997. A locomoting observer sees a very different visual scene than an observer at rest: images throughout the visual field accelerate and expand, and they follow approximately radial outward paths from a single origin. This so-called optic flow field is presumably used for visual guidance, and it has been suggested that particular areas of visual cortex are specialized for the analysis of optic flow. In the cat, the lateral suprasylvian visual area (LS) is a likely candidate. To test the hypothesis that LS is specialized for analysis of optic flow fields, we recorded cell responses to optic flow displays. Stimulus movies simulated the experience of a cat trotting slowly across an endless plain covered with small balls. In different simulations we varied the size of balls, their organization (randomly or regularly dispersed), and their color (all one gray level, or multiple shades of gray). For each optic flow movie, a "texture" movie composed of the same elements but lacking optic flow cues was tested. In anesthetized cats, >500 neurons in LS were studied with a variety of movies. Most (70%) of 454 visually responsive cells responded to optic flow movies. Visually responsive cells generally preferred optic flow to texture movies (69% of those responsive to any movie). The direction in which a movie was shown (forward or reverse) was also an important factor. Most cells (68%) strongly preferred forward motion, which corresponded to visual experience during locomotion.

	INTRODUCTION

Abstract Introduction Methods Results Discussion References

During locomotion, images move throughout the visual field in what Gibson et al. (1955) termed an optic flow pattern. Each object's image follows an approximately radial-outward trajectory originating at the heading point, and expands and accelerates as it moves. Gibson (1950) suggested that locomoting observers might make use of the predictability of cues in optic flow for visual guidance, and psychophysical work is consistent with this idea (Crowell and Banks 1993; Priest and Cutting 1985; Van den Berg 1992; Vishton and Cutting 1995; Warren 1976; Warren and Hannon 1988; Warren et al. 1988).

During the past decade, investigators have begun to explore the neural basis of optic flow analysis. A primary question is whether one or more specific visual cortical areas are specialized for such analysis. The middle superior temporal area (MST) in the macaque has been considered a candidate based on the observation that some cells in this area respond selectively to large fields of dots moving along radial-outward trajectories (Duffy and Wurtz 1991; Lagae et al. 1994; Orban et al. 1992; Saito et al. 1986; Tanaka and Saito 1989). In the cat, attention has focused on the lateral suprasylvian area (LS): directional preferences in this area reportedly are biased for radial-outward trajectories, which might match image motion in the cat's optic flow field (Hamada 1987; Rauschecker et al. 1987).

For neither cortical area, however, is a central role in visual guidance during locomotion well established. MST may be involved in the analysis of complex motion such as spiral trajectories as well as optic flow (Graziano et al. 1994). In LS, the evidence, although suggestive, is indirect. To test more directly whether LS may be specialized for the analysis of optic flow fields, we compared cell responses with two kinds of large stimulus display, one simulating an optic flow field, and the other, a textured field moving in a frontoparallel fashion. Our optic flow movies simulated the view of a cat trotting slowly across an endless plain covered with small balls (see Fig. 1A). Images of the balls followed appropriate trajectories and expanded and accelerated as appropriate for the cat's rate of travel. What we refer to as texture movies (Fig. 1B) were also composed of the same basic elements (multiple disks in varying shades of gray), but lacked "optic flow" cues: the disks followed parallel, not radial, trajectories, and they maintained a constant velocity and constant size.

View larger version (83K): [in this window] [in a new window]   FIG. 1. Frames from 3 different movies, each 27.8 × 19.6°. A: optic flow movie simulating 2.5-cm-diam balls. This display was centered 8.1° to the left of the vertical meridian, and 2.6° below the horizontal meridian. At the center, a ball subtended 1.1°. White rectangle shows a 10 × 10° receptive field. B: texture movie matched to the optic flow movie in part A. All elements moved at 9.1°/s, in a down/left direction, 29° clockwise from vertical. This speed and direction were identical to that of elements in the optic flow movie as they passed through the receptive-field center. C: regular-array optic flow movie. Balls behaved exactly as in A, but their starting locations on the simulated ground plane were regularly spaced.

We found that most cells responded to optic flow movies, and many distinguished between optic flow and texture movies. Cells preferring optic flow predominated in the sample. Although we tested a limited variety of optic flow displays, and explored only a limited extent of LS, the results are consistent with the idea that this area plays a role in visual analysis during locomotion.

	METHODS

Abstract Introduction Methods Results Discussion References

Animal preparation and maintenance

All experimental procedures were approved by the Animal Care Committee of the University of Washington. Data were collected from seven cats, except for four cells recorded from an additional animal used for an anatomic experiment. Cats were sedated with xylazine (Rompun; 0.1 ml/kg im), and anesthesia was induced with ketamine (Ketalar, 0.1 ml/kg im). After a foreleg vein had been cannulated and an endotracheal catheter inserted, anesthesia was maintained by administration of 0.75-1.0% halothane in nitrous oxide (70-75%) and oxygen. Electrocardiogram, expired CO₂ and, in some cats, electroencephalogram were monitored, and deep tendon reflexes were checked periodically before paralysis was initiated, to ensure that the level of anesthesia was satisfactory.

The animal was placed in a stereotaxic frame, and craniotomies were made over the left lateral gyrus and right suprasylvian sulcus. When all surgical procedures were complete, paralysis was induced by intravenous administration of pancuronium bromide (Pavulon, 0.1 mg·kg¹·h¹). Contact lenses were used to protect and focus the eyes, as judged by tapetal reflection (Pettigrew et al. 1979). The projected locations of the areae centrales and optic disks were drawn on paper attached to a computer monitor facing the cat.

Visual stimulation

Cats viewed stimulus displays binocularly, with images aligned in the two eyes using Risley prisms. Alignment was assessed by plotting the receptive fields of binocular cells recorded in area 17 in the left hemisphere. The positions of these receptive fields were checked periodically in order to detect eye drifts.

Receptive fields in LS were plotted by the use of a hand-held adjustable light slit, and the best direction of motion, speed, and length of bar was estimated. All cell responses were then tested quantitatively. Visual stimuli were generated on a 21-in. NeXT computer monitor placed 32 cm in front of the cat's eyes. The monitor had a resolution of 19 pixels/deg and a refresh rate of 68 Hz. Two different classes of stimuli were tested; the first, a bar moving against a blank background and the second, a large display containing multiple moving elements. In bar stimulus displays (described in detail in the following paper) a bar of optimal size and velocity, oriented orthogonally to its trajectory, was swept in eight directions through the receptive field.

Three varieties of multielement movie (optic flow, texture, and regular-array optic flow; see Fig. 1) were tested, each 27.8 × 19.6° in size. Movies were large enough to encompass receptive fields in LS, but only when centered on the receptive field, so different movies were needed for different receptive-field locations in order to maintain a constant heading point. The stimulus display was always centered on the receptive field. Movies were composed of prefabricated frames (TIFF images) and ran as endless loops without any discontinuity at a frame rate of 30 Hz. In one stimulus set we always included an optic flow movie, a matched texture movie, and a regular-array optic flow movie (described below). Each movie was shown in the forward direction and the reverse direction. In addition, the stimulus set included a black, expanding and accelerating bar shown against a blank gray background. The bar moved along the same trajectory and at the same rate as the central element in the movies and was oriented orthogonal to its direction of motion. Thus there were 7 different interleaved conditions in a given stimulus set, and they were repeated 5 times for a total of 35 presentations. Many cells were tested with two or three stimulus sets that differed in the size of balls simulated (see below). Within a stimulus set, a given movie ran for 6 s, followed by a 2-s interval when the display was stationary.

OPTIC FLOW MOVIES. These simulated the experience of a cat 30 cm high, maintaining a constant angle of gaze 12° below the heading point, and moving at 50-80 cm/s across a ground plane covered with small balls. Different movies were made simulating different sized balls, 1.5, 2.5, or 4 cm diam [different sizes were tested because the size of elements in a large, complex display may affect cell behavior in LS (Hamada 1987)]. One frame from a movie using 2.5-cm balls is shown in Fig. 1A. The image of each ball behaved as it would in an optic flow field. It originated at the cat's heading point, and followed an appropriate trajectory along a longitudinal path on an imaginary sphere surrounding the cat (such a trajectory is close to radial-outward near the heading point). The image expanded and accelerated as appropriate for its distance from the cat and for the cat's rate of travel.

View larger version (35K): [in this window] [in a new window]   FIG. 2. Poststimulus time histograms (PSTHs) for 3 different cells. Movies were shown in the order illustrated, and the whole sequence was repeated 5 times. Optic flow forward: optic flow movie composed of randomly dispersed balls run in the forward direction. Regular-array flow reverse: regular-array optic flow movie run in reverse. Texture forward: texture movie run in the forward direction. Optic flow reverse: optic flow movie run in reverse. Regular-array forward: regular-array optic flow movie run in the forward direction. Texture reverse: texture movie run in reverse. Expanding, accelerating bar: a black bar that behaved like a ball in a forward-going optic flow movie, following the same trajectory while accelerating and expanding. It was 6° long as it passed through the center of the receptive field; the background was uniform gray. A: cell that responded well only to optic flow in the forward direction. B: cell that responded best to forward-moving optic flow, but also quite well to forward-moving texture and regular-array optic flow movies. C: cell with best response to forward-moving texture movies. Note the feeble response to the solitary black bar (bottom PSTH, arrow). Cells in A and B did not respond to the bar, which passed through their receptive-field centers at the points marked with arrows. Scale bar in C applies to all PSTHs.

TEXTURE MOVIES. For each optic flow movie, a matching texture movie was made (e.g., Fig. 1B). It was composed of disks identical in size, velocity, and direction of motion to the disk passing through the center of the corresponding optic flow movie. All elements in a texture movie followed parallel paths and maintained constant size and speed.

REGULAR-ARRAY OPTIC FLOW MOVIES. This was a variant of the optic flow movie. All balls in these movies had the same color (dark gray or white), and they were positioned in a regular array on the ground plane (see Fig. 1C). Their behavior was identical to that of balls in randomly arrayed optic flow movies. We hypothesized that these movies would evoke at best weak responses because, being highly repetitive, they would rapidly habituate cell responses (e.g., Morrone et al. 1986). On the other hand, it is possible that cells are sensitive to motion cues but relatively indifferent to other stimulus details and therefore would give similar responses to regular-array optic flow and to optic flow simulations using randomly distributed and shaded balls.

For human observers, the two varieties of optic flow movie gave a strong sense of motion in depth, whether they were run in the forward or reverse direction. Texture movies never produced this illusion.

Data collection and analysis

Penetrations were made with the use of tungsten and glass microelectrodes, directed down the medial bank of the suprasylvian sulcus in a coronal plane. Data were collected via an A-to-D converter and array processor (Tucker-Davis Technologies) on a host Intel 486 computer. Stimulus presentation and data collection were synchronized by a trigger pulse sent from the NeXT to the 486 computer at the start of each stimulus trial. Spikes were sorted by size and shape with the use of appropriate software. Data from one to three cells were collected simultaneously, and, to provide a check on how well spikes had been sorted, 20 spike waveforms for each cell were saved from the middle of every stimulus run. Responses were shown as raster plots on-line, and subsequently as peristimulus time histograms (PSTHs).

Response strength was measured as firing rate, determined over an interval that depended on the kind of stimulus used. For bar stimuli, these intervals were short (typically 300-600 ms) because the stimulus was only briefly present in the receptive field. For movies, intervals were longer (typically 1,000-2,000 ms) because responses were more tonic. The initial transient response, shown by a few cells, was excluded because it usually occurred nonselectively at the start of every trial regardless of stimulus condition. A cell was considered to respond to a stimulus if its response was at least 7 spikes/s above the spontaneous level and was significant with P < 0.05 using a one-tailed Mann-Whitney test. In this test, responses to five presentations of a stimulus were compared with five periods of spontaneous activity, measured over the same interval and taken from the same stimulus set. In this and the following paper, by "response" we mean average firing frequency minus the cell's spontaneous firing level.

Two numerical indexes were used to measure cell preferences for different movies.

1) Flow Index. A value comparing a cell's response to an optic flow movie and to the matching texture movie. If the cell responded more strongly to optic flow, Flow Index = 1  response to texture/response to optic flow. Otherwise, Flow Index = (1  response to optic flow/response to texture).

2) Movie Direction Index. A measure of the strength of a cell's preference for forward or reverse motion of a movie. If the cell's response to forward motion was greater, Movie Direction Index = 1  response to reverse motion/response to forward motion. If its response to reverse motion was greater, Movie Direction Index = (1  response to forward motion/response to reverse motion).

Histological processing and localization of recording sites

At the end of the experiment, the animal was killed with an overdose of barbiturate and perfused transcardially with saline followed by aldehyde fixative. Brains were blocked stereotaxically in a coronal plane, cryoprotected in 30% sucrose-phosphate buffer, and frozen in powdered dry ice. Frozen sections, 30-40 µm thick, were cut coronally, and a 1-in-10 series was mounted and stained with cresyl violet. Penetrations were reconstructed from camera lucida drawings, and the borders between cortical layers were determined with the use of the descriptions of Sanides and Hoffman (1969). Recording sites were located relative to cortical borders by reference to electrolytic lesions made at known depths along each electrode penetration. Lesions were recovered for all but 1 of 15 penetrations.

	RESULTS

Abstract Introduction Methods Results Discussion References

Cell locations and visual responsiveness

Recording sites were located in the lower field representation of LS in the right hemisphere. Penetrations were made between A2 and the posterior end of the suprasylvian sulcus, with the great majority of recording sites on the medial bank of the suprasylvian sulcus, within the posteromedial lateral suprasylvian area (PMLS) of Palmer et al. (1978). The remainder were in the fundus or lateral bank of this sulcus, and thus presumably would be in the posterolateral lateral suprasylvian area (PLLS) of these authors. This part of PLLS falls within LS as defined by connections with area 17 (Grant and Shipp 1991; Sherk 1986) and by retinotopic organization (Grant and Shipp 1991; Sherk and Mulligan 1993). Receptive fields were located close to or below the horizontal meridian in the left visual field, within 30° of the center of gaze.

Over 500 cells were tested, and of these, 454 responded to some visual stimulus within the computer's repertoire and were included in the data set. We have not tried to estimate the number of unresponsive cells encountered because such estimates are likely to be unreliable. Some apparently unresponsive cells may not have been tested with appropriate stimuli, whereas other truly unresponsive cells probably escaped detection altogether.

Responses to movies

Most cells in our sample (333 of 454) responded to at least one kind of movie. Figure 2 illustrates the responses of three cells with varying degrees of selectivity for different movies. The first cell (Fig. 2A) was strongly selective for optic flow movies run in the forward direction. Compared with other cells in our sample, this cell's response was unusually sustained, remaining well above background throughout the 6-s optic flow presentation. The second cell (Fig. 2B) also preferred optic flow, but in addition responded to texture movies and to regular-array optic flow movies, all run in the forward direction. The third cell (Fig. 2C) responded best to texture, and also fairly well to optic flow, both run in the forward direction. Unlike the other two cells, it responded, albeit weakly, to an expanding, accelerating bar (bottom PSTH, arrow).

View larger version (17K): [in this window] [in a new window]   FIG. 3. Flow Index values for all cells responding to movies. This measure showed how strongly a cell preferred either an optic flow movie or the matched texture movie. Values greater than 0 show a preference for optic flow, values less than 0, a preference for texture. Some cells had values greater than 1 because they were inhibited by texture movies.

View larger version (18K): [in this window] [in a new window]   FIG. 4. Responses of a cell that responded well to optic flow movies in the forward direction, weakly to other movies in the forward direction, and not significantly to other stimuli, including bars in any direction. Top PSTH shows the maximal response to a bar (arrow).

As these examples suggest, most cells did distinguish between different kinds of movie. The majority of cells responded best to optic flow movies. We quantified this preference by computing a Flow Index value for each cell (see METHODS), with 0 signifying no preference, positive values a preference for optic flow movies, and negative values a preference for texture movies. Figure 3 shows that for most cells this index was positive, reflecting the bias for optic flow. The average value for cells that preferred optic flow was 0.54, whereas the average for cells preferring texture was lower, 0.35. The strength of preference for optic flow was not correlated with either receptive-field location or size.

A few cells (8/333) appeared to be driven exclusively by optic flow displays. These responded well to optic flow movies shown in the forward direction, poorly or not at all to texture movies, and not to any moving bar stimulus. An example is illustrated in Fig. 4. The only stimulus that elicited a significant response was the optic flow movie run in the forward direction. A moving bar stimulus, which was usually a potent stimulus for LS cells, evoked no significant response (arrow, top PSTH).

Regular array optic flow movies (illustrated in Fig. 1C) were less effective stimuli than either of the other two varieties. Figure 5 compares responses to these movies with responses to the standard optic flow movies (Fig. 5A) and responses to texture (Fig. 5B). Responses to regular-array optic flow were clearly weaker for a substantial majority of cells. The reason for this difference is not clear. We had expected that responses to regular-array optic flow would rapidly habituate because of the highly repetitive nature of the display, but in fact responses were generally weak from the outset, with a duration similar to that elicited by other movies. Curiously, 49 cells instead preferred regular-array optic flow movies to either of the other varieties (white dots in Fig. 5). Nothing about their responses to single bars, such as a preference for white over black, explained this behavior.

View larger version (23K): [in this window] [in a new window]   FIG. 5. Comparison of responses to regular-array optic flow movies with responses to standard optic flow movies (A), or with responses to texture movies (B). White dots represent cells that responded better to regular-array movies than to both the other kinds.

There was a strong bias for movies shown in the forward rather than the reverse direction, evident in 68% of the cells. We quantified this preference by computing a Movie Direction Index that compares responses to a given movie shown forward and in reverse. This index is similar to the commonly used direction index but differs in that preferences for both directions are included on one axis. Preferences for forward directions have positive values; preferences for reverse directions have negative values. The distribution for our sample was bimodal and strongly skewed toward the positive end (see Fig. 6).

View larger version (18K): [in this window] [in a new window]   FIG. 6. Values of Movie Direction Index for cells responding to movies. Positive values show a preference for forward motion, negative values, a preference for reverse motion. A value of 0 indicates equal responses for the 2 directions.

To determine whether cell responses were shaped mainly by the global motion cues in the stimulus display, or by the appearance of individual elements within the display, we also tried varying the size of balls within optic flow and texture movies. Many cells were tested with two sets of movies, one simulating medium-sized balls (2.5 cm diam), and the other either large balls (4 cm diam) or small balls (1.5 cm diam). Some cells were tested with all three. Ball size did not greatly affect the responses of most cells (see Fig. 7). However, about one-quarter of those tested (40/148) had significantly better responses to one size (P < 0.05; white dots in Fig. 7).

View larger version (16K): [in this window] [in a new window]   FIG. 7. Effect of simulated ball size (small, medium, or large) on responses to optic flow movies. A: responses to movies composed of medium balls versus ones composed of large balls. B: responses to movies composed of small balls vs. ones composed of medium balls. White dots indicate cells whose responses were significantly better to one condition than to the other (t-test, P < 0.05).

	DISCUSSION

Abstract Introduction Methods Results Discussion References

The main question addressed by this study was whether cells in LS distinguish between large-field simulations of optic flow, and large-field texture displays lacking optic flow cues. Of cells that responded to any large-field display, most were found to prefer optic flow movies. Furthermore, ~75% of cells that preferred optic flow movies had a marked preference for forward motion compared with reverse. Because forward motion simulated optic flow such as the cat would see during normal locomotion, it seems plausible that these cells are involved in visual analysis during locomotion.

The fact that cells generally did distinguish between optic flow and matched texture movies suggests that responses were strongly influenced by images far from the receptive-field center. Images passing through the center of the receptive field were identical for the two movies. The differences between them increased only gradually with distance away from the center, so that even at the margins of a typical receptive field, the two kinds of display remained similar. Within the receptive field illustrated in Fig. 1, for example, the maximum difference between the directions of motion in optic flow movies and in texture movies was only 15°. Acceleration might have been a more effective cue for distinguishing between the two kinds of movie; in this example, velocities in the optic flow movie increased from 4 to 16.7°/s across the width of the receptive field, whereas images in the texture movie moved at a constant 10.3°/s. However, cells did not distinguish between solitary bars moving at constant velocity and the same bars when accelerating (see companion paper), suggesting that they were relatively indifferent to acceleration. The same argument applies to image expansion. The disks in optic flow movies expanded and those in texture movies did not, but cells did not appear very sensitive to this cue because they did not distinguish between solitary moving bars that expanded and ones that maintained a constant size (see companion paper).

The similarity of cues within the receptive field in optic flow and texture movies implicates the receptive-field surround as a mechanism for differentiating between the two. Far from the receptive-field center, but certainly within the large surrounds described by von Grunau and Frost (1983) in LS, the differences between optic flow and texture movies became pronounced. In Fig. 1, directions of motion at the periphery of the display differed by up to 45° from the direction seen by the receptive-field center. Von Grunau and Frost's results suggest that a receptive-field surround would be sensitive to this difference when image motion through the field center is similar to the cell's preferred direction. They found that random dot motion in the surround suppressed responses to stimuli passing through the center when the dots moved in the cell's preferred direction. However, when the direction of random dot motion in the surround was reversed, suppression disappeared or changed to facilitation. In the case of optic flow movies, we might expect little suppression or possibly facilitation because the directions of motion throughout much of the putative receptive-field surround differed from the direction seen by the receptive-field center. Texture movies, however, would tend to evoke suppression because elements throughout the surround all moved in the same direction. The parallel with previous work is not exact, however. For most cells in the present study, the direction of motion seen at the receptive-field center did not match the cell's preferred direction (see companion paper). How random dot motion throughout the surround modulates cell responses when the entire display moves in an inappropriate direction has not been explored.

We also asked whether cells respond similarly to any optic flow movie characterized by a particular set of motion cues, or whether instead they are affected by differences in the elements making up the display. In pilot experiments we had found that displays with a sparse distribution of elements were relatively ineffective stimuli: many cells responded to particular elements passing through the receptive field but appeared unaffected by the rest of the display (Sherk and Kim 1994). In the present experiments, we manipulated the elements within optic flow movies in two ways. In our standard optic flow movies, the gray levels and spacing of balls on the simulated ground plane were highly variable, whereas in regular-array optic flow movies, all balls had the same gray level and were spaced regularly across the ground plane. We also tested the effect of element size in optic flow and texture movies. Note that none of these variations affected motion cues (direction of motion, acceleration, and image expansion). Overall, cells clearly were sensitive to the nature of the elements within optic flow movies. The regular-array movies in particular evoked weaker responses than did the others in a large majority of cells. We do not know why. A modest number of cells also showed a marked preference for optic flow movies containing a particular size of ball. This outcome is consistent with Hamada (1987) and von Grunau and Frost's (1983) reports that the size of dots in wide-field random dot arrays is critical for evoking suppressive surround mechanisms: small elements (<1° diam) yielded suppression, but larger ones did not.

If we suppose that cells in LS are actively engaged in visual analysis during locomotion, what role might they play? We might be tempted to suggest that those preferring optic flow to texture movies function as "optic flow detectors." But we think it unlikely that any significant population in LS performs this function. The utility of optic flow detectors to a locomoting observer is doubtful, because the observer has abundant other cues, both efferent and reafferent, that he is locomoting. It seems more plausible that some cells respond to particular objects passing through their receptive fields within the context of an optic flow field, but not when the same objects are embedded in a large moving scene that lacks optic flow cues. During saccades or smooth pursuit eye movements the entire visual scene (excluding the fixated object in pursuit) moves in a frontoparallel fashion. It is possible that a preference for optic flow permits cells in LS to respond to images during locomotion, but precludes significant responses during eye movements.

	ACKNOWLEDGEMENTS

  We gratefully acknowledge the helpful comments of P. Abel, D. Dacey, B. O'Brien, and J. Olavarria on this manuscript.

  This work was supported by Grant RO1 4847 and by the Royalty Research Fund of the University of Washington.

	FOOTNOTES

  Address for reprint requests: H. Sherk, Dept. of Biological Structure, University of Washington, Box 35-7420, Seattle, WA 98195-7420.

  Received 1 August 1996; accepted in final form 17 October 1996.

	REFERENCES

Abstract Introduction Methods Results Discussion References

• CROWELL, J. A., BANKS, M. S. Perceiving heading with different retinal regions and types of optic flow. Percept. Psychophys. 53: 325-337, 1993.[Medline]
• DUFFY, C. J., WURTZ, R. H. Sensitivity of MST neurons to optic flow stimuli. I. A continuum of response selectivity to large-field stimuli. J. Neurophysiol. 65: 1329-1345, 1991.[Abstract/Free Full Text]
• GIBSON, J. J. In: The Perception of the Visual World., . Boston, MA: Houghton-Mifflin, 1950
• GIBSON, J. J., OLUM, P., ROSENBLATT, F. Parallax and perspective during aircraft landings. Am. J. Psychol. 68: 372-385, 1955.[Medline]
• GRANT, S., SHIPP, S. Visuotopic organization of the lateral suprasylvian area and adjacent area of ectosylvian gyrus of the cat cortex: a physiological and connectional study. Visual Neurosci. 6: 315-338, 1991.[Medline]
• GRAZIANO, M. S. A., ANDERSEN, R. A., SNOWDEN, R. J. Tuning of MST neurons to spiral motions. J. Neurosci. 14: 54-67, 1994.[Abstract]
• HAMADA, T. Neural response to the motion of textures in the lateral suprasylvian area of cats. Behav. Brain Res. 25: 175-185, 1987.[Medline]
• LAGAE, L., MAES, H., RAIGUEL, S., XIAO, D. K., ORBAN, G. A. Responses of macaque STS neurons to optic flow components: a comparison of areas MT and MST. J. Neurophysiol. 71: 1597-1626, 1994.[Abstract/Free Full Text]
• MORRONE, M. C., DI SEFANO, M., BURR, D. C. Spatial and temporal properties of neurons of the lateral suprasylvian cortex of the cat. J. Neurophysiol. 56: 969-986, 1986.[Abstract/Free Full Text]
• ORBAN, G. A., LAGAE, L., VERRI, A., RAIGUEL, S., XIAO, D., MAES, H., TORRE, V. First-order analysis of optical flow in monkey brain. Proc. Natl. Acad. Sci. USA 89: 2595-2599, 1992.[Abstract/Free Full Text]
• PALMER, L. A., ROSENQUIST, A. C., TUSA, R. J. The retinotopic organization of lateral suprasylvian visual areas in the cat. J. Comp. Neurol. 177: 237-256, 1987.
• PETTIGREW, J. D., COOPER, M. L., BLASDEL, G. G. Improved use of tapetal reflection for eye-position monitoring. Invest. Ophthalmol. Visual Sci. 18: 490-495, 1979.[Abstract]
• PRIEST, H. F., CUTTING, J. E. Visual flow and direction of locomotion. Science Wash. DC 227: 1063-1064, 1985.[Free Full Text]
• RAUSCHECKER, J. P., VON GRUNAU, M. W., POULIN, C. Centrifugal organization of direction preferences in the cat's lateral suprasylvian visual cortex and its relation to flow field processing. J. Neurosci. 7: 943-958, 1987.[Abstract]
• SAITO, H., YUKIE, M., TANAKA, M., HIKOSAKA, K., FUKADA, Y., IWAI, E. Integration of direction signals of image motion in the superior temporal sulcus of the macaque monkey. J. Neurosci. 6: 145-157, 1986.[Abstract]
• SANIDES, F., HOFFMANN, J. Cyto- and myeloarchitecture of the visual cortex of the cat and of the surrounding integration cortices. J. Hirnforsch. 11: 79-104, 1969.[Medline]
• SHERK, H. Location and connections of visual cortical areas in the cat's suprasylvian sulcus. J. Comp. Neurol. 247: 1-31, 1986.[Medline]
• SHERK, H., KIM, J.-N. Responses to optic flow simulations in cat extrastriate cortex. Soc. Neurosci. Abstr. 20: 427, 1994.
• SHERK, H., MULLIGAN, K. A. A reassessment of the lower visual field map in striate-recipent lateral suprasylvian cortex. Visual Neurosci. 10: 131-158, 1993.[Medline]
• TANAKA, K., SAITO, H.-A. Analysis of motion of the visual field by direction, expansion/contraction, and rotation cells clustered in the dorsal part of the medial superior temporal area of the macaque monkey. J. Neurophysiol. 62: 626-641, 1989.[Abstract/Free Full Text]
• VAN DEN BERG, A. V. Robustness of perception of heading from optic flow. Visual Res. 32: 1285-1296, 1992.
• VON GRUNAU, M., FROST, B. J. Double-opponent process mechanism underlying RF-structure of directionally specific cells of cat lateral suprasylvian visual area. Exp. Brain Res. 49: 84-92, 1983.[Medline]
• VISHTON, P. M., CUTTING, J. E. Wayfinding, displacements, and mental maps: velocity fields are not typically used to determine one's aimpoint. J. Exp. Psychol. 21: 978-995, 1995.
• WARREN, R. The perception of egomotion. J. Exp. Psychol. Hum. Percept. Perform. 2: 448-456, 1976.[Medline]
• WARREN, W. H. JR, HANNON, D. J. Direction of self-motion is perceived from optical flow. Nature Lond. 336: 162-163, 1988.
• WARREN, W. H. JR, MORRIS, M. W., KALISH, M. Perception of translational heading from optical flow. J. Exp. Psychol. Hum. Percept. Perform. 14: 646-660, 1988.[Medline]

0022-3077/97 $5.00 Copyright ©1997 The American Physiological Society