Responses of Visual-Tracking Neurons from Cortical Area MST-I to Visual, Eye and Head Motion.

Thirty-one neurons which exhibited ocular pursuit-related activity [visual-tracking (VT) neurons] were found clustered within area MST-I (the lateral part of area MST) of two rhesus monkeys. Their responses were studied to determine whether this activity was correlated only with pursuit eye movement or with head movement as well. The latter hypothesis appeared to be preferable since visual, eye movement and head movement inputs were found to be mapped in register onto most of these cells. First, in each cell tested (n=19) the pursuit response persisted even in the absence of retinal image motion, offering clear evidence for non-visual input. Second, 22 of the 31 cells were directionally responsive to moving visual stimuli and in 20 of these the preferred directions for the visual motion and pursuit responses agreed closely. Responses were also obtained from many of the same cells during suppression of both the horizontal and the vertical vestibulo-ocular reflex (VOR). In each case, where directional visual, pursuit and VOR suppression responses were each obtained, vector addition of responses during suppression of the horizontal and vertical VOR resulted in an estimated preferred direction for head rotation which was closely aligned with the preferred direction previously obtained for eye motion or visual motion. In addition, the preferred direction of head movement was conserved even when the VOR was elicited by passive head rotation in complete darkness, although the responses in this instance were, on average, only 62% of those obtained during VOR suppression. Our interpretation is that, at present, MST-I VT neurons are best described as encoding the direction of target motion in space-centred coordinates by integrating inputs reflecting retinal image motion plus eye and head movement.

The response of area MT and V1 neurons to transparent motion.

An important use of motion information is to segment a complex visual scene into surfaces and objects. Transparent motions present a particularly difficult problem for segmentation because more than one velocity vector occurs at each local region in the image, and current machine vision systems fail in these circumstances. The fact that motion transparency is prevalent in natural scenes, and yet artificial systems display an inability to analyze it, suggests that the primate visual system has developed specialized methods for perceiving transparent motion. Also, the currently prevalent model of physiological mechanisms for motion-direction selectivity employs inhibitory interactions between neurons; such interactions would silence neurons under transparent conditions and render the visual system blind to transparent motion. To examine how the primate visual system solves this transparency problem, we recorded the activity of direction-selective cells in the first (area V1) and in a later (area MT) stage in the cortical motion-processing pathway in behaving monkeys. The visual stimuli consisted of random dot patterns forming single moving surfaces, transparent surfaces, and motion discontinuities. We found that area V1 cells responded to their preferred direction of movement even under transparent conditions, whereas area MT cells were suppressed under the transparent condition. These data suggest a simple solution to the transparency problem at the level of area V1. More than one motion vector can be represented at a single retinal location by different subpopulations of neurons tuned to different directions of motion; these subpopulations may represent the early stage for segmenting different, transparent surfaces. The results also suggest that facilitatory mechanisms, which unlike inhibitory interactions are largely unaffected by transparent conditions, play an important role in direction selectivity in area V1. The inhibitory interactions for different motion directions for area MT neurons may contribute to a mechanism for smoothing or averaging the velocity field, computations thought to be necessary for reducing noise and interpolating moving surfaces from sparse information.

A neuronal correlate of spatial stability during periods of self-induced visual motion.

Motion of background visual images across the retina during slow tracking eye movements is usually not consciously perceived so long as the retinal image motion results entirely from the voluntary slow eye movement (otherwise the surround would appear to move during pursuit eye movements). To address the question of where in the brain such filtering might occur, the responses of cells in 3 visuo-cortical areas of macaque monkeys were compared when retinal image motion of background images was caused by object motion as opposed to a pursuit eye movement. While almost all cells in areas V4 and MT responded indiscriminately to retinal image motion arising from any source, most of those recorded in the dorsal zone of area MST (MSTd), as well as a smaller proportion in lateral MST (MST1), responded preferentially to externally-induced motion and only weakly or not at all to self-induced visual motion. Such cells preserve visuo-spatial stability during low-velocity voluntary eye movements and could contribute to the process of providing consistent spatial orientation regardless of whether the eyes are moving or stationary.

Foveal tracking cells in the superior temporal sulcus of the macaque monkey.

Visual responses were recorded from neurons in the superior temporal sulcus (STS) of awake, behaving cynomolgus monkeys trained to fixate a small spot of light. Visual receptive fields, directionality, and responses during visual tracking were examined quantitatively for 50 cells in the foveal portion of the middle temporal (MT) visual area and surrounding cortex. Directionality indices and preferred directions for tracked and nontracked stimuli were compared. Eighteen cells (18/50 = 36%) were found to respond preferentially during tracking (tracking cells), 7 within MT, 9 in area FST on the floor of the STS, and 2 in unidentified areas. Three distinctly different tracking response profiles (VTS, VTO, and T) were observed. VTS and VTO cells had foveal receptive fields and gave directionally selective visual responses. VTS cells (3 in foveal MT, 6 in FST, 1 in an unidentified area) had a preferred visual direction that coincided with the preferred tracking direction, and began responding 50-100 ms before the onset of tracking. VTO cells (4 in foveal MT, 0 in FST, 1 in an unidentified area) had a preferred visual direction opposite to the preferred tracking direction, and began responding 0-100 ms after the onset of tracking. T cells (0 in MT, 3 in FST) had no visual responses and began responding simultaneously with the onset of tracking. It is suggested that this region of the brain could be the primary location for converting direction-specific visual responses into signals specifying at least the direction of an intended pursuit movement.

Representation of the fovea in the superior temporal sulcus of the macaque monkey.

The response properties of 633 neurons from striate and prestriate cortex were recorded in 3 hemispheres of two awake cynomolgus monkeys while they fixated or tracked a small spot of light. Of 254 penetrations located at 1 mm intervals, 39% were identifiable from visible electrolytic lesions or electrode tracks and were used to reconstruct the positions of all recording sites. A total of 226 cells were located in the superior temporal sulcus and 81 cells in area V1. The location and visuotopic organization of the foveal portion of the middle temporal (MT) visual area were determined in three hemispheres. MT was defined physiologically on the basis of direction-selectivity, receptive field size, and retinotopic organization. Of 170 MT neurons, most were motion sensitive, and 65% had a directionality index, (best-opposite)/best, of 0.6 or higher. MT was defined anatomically on the basis of myelin staining within the superior temporal sulcus (STS). On the posterior bank of the STS the physiologically defined border corresponded closely to a myelin border visible on our sections. Distinct myelin borders were not consistently identifiable on the anterior bank. The representation of the central fovea (eccentricities of 0-1 deg) was located partly on the floor, but mostly on the posterior bank of the STS at the extreme postero-lateral edge of MT. In all three hemispheres foveal MT extended onto the roof of a cleft formed between the posterior bank and a wide flattened area on the floor of the STS. This region lies 10-12 mm below the brain surface, measuring along a line normal to the surface at a point 2-3 mm antero-lateral to foveal V1. The area of MT was 6-9 mm2 for the central fovea (0-1 deg), 15-24 mm2 for the entire fovea (0-3 deg), and 28-40 mm2 including the fovea and parafovea (0-10 deg). A visuotopic map of central foveal V1 (0-1 deg) was obtained in one animal. The measured area of this representation was 116 mm2. Using published estimates of the total areas of cynomolgus MT and V1 (73 and 1200 mm2 respectively) the ratio of central foveal to total area was calculated to be 0.10 for both MT (7.5/73) and V1 (116/1200), indicating that the relative magnification of the foveal versus the peripheral visual field is preserved in the mapping of V1 onto MT. A separate representation of the central visual field was found immediately adjacent to foveal MT.(ABSTRACT TRUNCATED AT 400 WORDS)

Physiological and anatomical identification of the nucleus of the optic tract and dorsal terminal nucleus of the accessory optic tract in monkeys.

Physiological and anatomical criteria were used to clearly establish the existence of a pretectal relay of visual information to the ipsilateral inferior olive in the macaque monkey. After injection of horseradish peroxidase into the inferior olivary nucleus, retrogradely labelled neurons were found in the nucleus of the optic tract (NOT) and the dorsal terminal nucleus of the accessory optic tract (DTN). The labelled cells were distributed in a sparse band arching below the margin of the brachium of the superior colliculus between the dorsal and lateral borders of the brainstem at the caudal edge of the pulvinar. Various types of cells could be distinguished. More superficially the cells were extremely spindle shaped, cells deeper within the midbrain had more compact somata. NOT-DTN neurons in the same region were also found to respond with short latencies to electrical stimulation of both the inferior olive and the optic chiasm. All neurons in the NOT-DTN which were antidromically activated from the inferior olive were also found to have direction specific binocular visual responses. Such neurons were excited by ipsiversive motion and suppressed by contraversive motion, regardless of whether large area random dot stimuli moved across the visual field or small single dots moved across the fovea. Direct retinal input to these neurons was via slowly conducting fibers (3-9 m/s) from the monkey's optic tract conduction velocity spectrum. As shown previously for non-primates, NOT-DTN cells may also in the monkey carry a signal representing the velocity error between stimulus and retina (retinal slip), and relay this signal into the circuitry mediating the optokinetic reflex.

Uncrossed retinal projections to the accessory optic nuclei in rabbits and cats.

Retinal projections to the accessory optic nuclei of rabbits and cats were demonstrated with standard autoradiographic techniques following intraocular injections of [35S]methionine and [3H]proline. In the pigmented rabbit, albino rabbit, normally pigmented domestic cat and Siamese cat the medial, lateral and dorsal terminal nuclei (MTN, LTN, and DTN, respectively) of the accessory optic system were densely labelled on the side contralateral to the injected eye. An ipsilateral projection, while clearly present in all but the Siamese cat, varied in the number of nuclei involved. In the albino rabbit, the ipsilateral projection ended in the MTN, while in the pigmented rabbit, it ended in the MTN, LTN and DTN, and in the normally pigmented domestic cat it ended in the MTN and LTN. These results indicate that the accessory optic system in rabbits and cats is more extensive than previously reported and that differences exist in the accessory optic system which may be related to genetic differences in normally pigmented and hypopigmented animals.