1
|
Exploitation of image statistics with sparse coding in the case of stereo vision. Neural Netw 2020; 135:158-176. [PMID: 33388507 DOI: 10.1016/j.neunet.2020.12.016] [Citation(s) in RCA: 5] [Impact Index Per Article: 1.3] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/28/2020] [Revised: 12/06/2020] [Accepted: 12/14/2020] [Indexed: 11/23/2022]
Abstract
The sparse coding algorithm has served as a model for early processing in mammalian vision. It has been assumed that the brain uses sparse coding to exploit statistical properties of the sensory stream. We hypothesize that sparse coding discovers patterns from the data set, which can be used to estimate a set of stimulus parameters by simple readout. In this study, we chose a model of stereo vision to test our hypothesis. We used the Locally Competitive Algorithm (LCA), followed by a naïve Bayes classifier, to infer stereo disparity. From the results we report three observations. First, disparity inference was successful with this naturalistic processing pipeline. Second, an expanded, highly redundant representation is required to robustly identify the input patterns. Third, the inference error can be predicted from the number of active coefficients in the LCA representation. We conclude that sparse coding can generate a suitable general representation for subsequent inference tasks.
Collapse
|
2
|
Silencing "Top-Down" Cortical Signals Affects Spike-Responses of Neurons in Cat's "Intermediate" Visual Cortex. Front Neural Circuits 2017; 11:27. [PMID: 28487637 PMCID: PMC5404610 DOI: 10.3389/fncir.2017.00027] [Citation(s) in RCA: 4] [Impact Index Per Article: 0.6] [Reference Citation Analysis] [Abstract] [Key Words] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/27/2016] [Accepted: 04/04/2017] [Indexed: 11/13/2022] Open
Abstract
We examined the effects of reversible inactivation of a higher-order, pattern/form-processing, postero-temporal visual (PTV) cortex on the background activities and spike-responses of single neurons in the ipsilateral cytoarchitectonic area 19 (putative area V3) of anesthetized domestic cats. Very occasionally (2/28), silencing recurrent “feedback” signals from PTV, resulted in significant and reversible reduction in background activity of area 19 neurons. By contrast, in large proportions of area 19 neurons, PTV inactivation resulted in: (i) significant reversible changes in the peak magnitude of their responses to visual stimuli (35.5%; 10/28); (ii) substantial reversible changes in direction selectivity indices (DSIs; 43%; 12/28); and (iii) reversible, upward shifts in preferred stimulus velocities (37%; 7/19). Substantial (≥20°) shifts in preferred orientation and/or substantial (≥20°) changes in width of orientation-tuning curves of area 19 neurons were however less common (26.5%; 4/15). In a series of experiments conducted earlier, inactivation of PTV also induced upward shifts in the preferred velocities of the ipsilateral cytoarchitectonic area 17 (V1) neurons responding optimally at low velocities. These upward shifts in preferred velocities of areas 19 and 17 neurons were often accompanied by substantial increases in DSIs. Thus, in both the primary visual cortex and the “intermediate” visual cortex (area 19), feedback from PTV plays a modulatory role in relation to stimulus velocity preferences and/or direction selectivity, that is, the properties which are usually believed to be determined by the inputs from the dorsal thalamus and/or feedforward inputs from the primary visual cortices. The apparent specialization of area 19 for processing information about stationary/slowly moving visual stimuli is at least partially determined, by the feedback from the higher-order pattern-processing visual area. Overall, the recurrent signals from the higher-order, pattern/form-processing visual cortex appear to play an important role in determining the magnitude of spike-responses and some “motion-related” receptive field properties of a substantial proportion of neurons in the intermediate form-processing visual area—area 19.
Collapse
|
3
|
Specificity of neuronal responses in primary visual cortex is modulated by interhemispheric corticocortical input. ACTA ACUST UNITED AC 2010; 20:2776-86. [PMID: 20211943 PMCID: PMC2978237 DOI: 10.1093/cercor/bhq024] [Citation(s) in RCA: 33] [Impact Index Per Article: 2.4] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 12/02/2022]
Abstract
Within the visual cortex, it has been proposed that interhemispheric interactions serve to re-establish the continuity of the visual field across its vertical meridian (VM) by mechanisms similar to those used by intrinsic connections within a hemisphere. However, other specific functions of transcallosal projections have also been proposed, including contributing to disparity tuning and depth perception. Here, we consider whether interhemispheric connections modulate specific response properties, orientation and direction selectivity, of neurons in areas 17 and 18 of the ferret by combining reversible thermal deactivation in one hemisphere with optical imaging of intrinsic signals and single-cell electrophysiology in the other hemisphere. We found interhemispheric influences on both the strength and specificity of the responses to stimulus orientation and direction of motion, predominantly at the VM. However, neurons and domains preferring cardinal contours, in particular vertical contours, seem to receive stronger interhemispheric input than others. This finding is compatible with interhemispheric connections being involved in horizontal disparity tuning. In conclusion, our results support the view that interhemispheric interactions mainly perform integrative functions similar to those of connections intrinsic to one hemisphere.
Collapse
|
4
|
Stimulus-Dependent Interaction between the Visual Areas 17 and 18 of the 2 Hemispheres of the Ferret (Mustela putorius). Cereb Cortex 2007; 18:1951-60. [DOI: 10.1093/cercor/bhm222] [Citation(s) in RCA: 37] [Impact Index Per Article: 2.2] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/14/2022] Open
|
5
|
Abstract
This study assessed the early mechanisms underlying perception of plaid motion. Thus, two superimposed gratings drifting in a rightward direction composed plaid stimuli whose global motion direction was perceived as the vector sum of the two components. The first experiment was aimed at comparing the perception of plaid motion when both components were presented to both eyes (dioptic) or separately to each eye (dichoptic). When components of the patterns had identical spatial frequencies, coherent motion was correctly perceived under dioptic and dichoptic viewing condition. However, the perceived direction deviated from the predicted direction when spatial frequency differences were introduced between components in both conditions. The results suggest that motion integration follows similar rules for dioptic and dichoptic plaids even though performance under dichoptic viewing did not reach dioptic levels. In the second experiment, the role of early cortical areas in the processing of both plaids was examined. As convergence of monocular inputs is needed for dichoptic perception, we tested the hypothesis that primary visual cortex (V1) is required for dichoptic plaid processing by delivering repetitive transcranial magnetic stimulation to this area. Ten minutes of magnetic stimulation disrupted subsequent dichoptic perception for approximately 15 min, whereas no significant changes were observed for dioptic plaid perception. Taken together, these findings suggest that V1 is not crucial for the processing of dioptic plaids but it is necessary for the binocular integration underlying dichoptic plaid motion perception.
Collapse
|
6
|
Abstract
The present study aims at evaluating the spatial disparity response profiles of binocular cells in the superficial layers of the superior colliculus of the cat using drifting light bars and phase-shifted spatial frequency gratings. Results show that a total of 64% of the cells were sensitive to phase disparities and had large tuning profiles. Similarly, a large proportion (75%) of those tested with position offsets showed one of the four classic disparity profiles, those of the tuned cells being rather coarse. When tested with both position and phase disparities, 54% of the cells showed sensitivity profiles to the two types of stimuli. The overall results suggest that the superior colliculus is involved in the analysis of coarse stereopsis and/or the planning and initiation of saccades during vergence eye movements and/or the control of fine adjustments to maintain fixation as the stimulus moves in depth.
Collapse
|
7
|
Binocular fusion/suppression to spatial frequency differences at the border of areas 17/18 of the cat. Neuroscience 2004; 124:121-36. [PMID: 14960345 DOI: 10.1016/j.neuroscience.2003.10.035] [Citation(s) in RCA: 2] [Impact Index Per Article: 0.1] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Accepted: 10/07/2003] [Indexed: 11/19/2022]
Abstract
As shown by various human psychophysical studies, interocular spatial frequency disparities can yield a variety of percepts. In order to examine how binocular fusion is affected by spatial frequency differences, we have recorded cells in the border region of areas 17/18 of anesthetized cats. The optic axes of the eyes were deviated onto cathode-ray screens, and the optimal spatial frequency of each eye was assessed by monocular stimulations using drifting sinusoidal gratings. The optimal relative phase using identical spatial frequencies in both eyes was first determined. Spatial frequency differences were then introduced by keeping the optimal spatial frequency constant in one eye and varying the spatial frequency in the other. Results indicate that cells (39%) responded with an increased firing rate (facilitation) to similar spatial frequencies in each eye and with a gradual attenuation (occlusion or suppression) when spatial frequency differences were increased. However, binocular facilitation did not always occur to the presentation of identical stimuli. For 16% of the cells, maximal responses were observed when lower spatial frequencies than the optimal one were presented in one eye while higher spatial frequencies produced suppression. The opposite pattern was observed only for two cells. These findings are discussed in terms of binocular fusion and suppression.
Collapse
|
8
|
Abstract
Binocular interactions were investigated in area 19 of the anaesthetized cat using dichoptically presented phase-shifted static spatial frequency gratings that flickered at a fixed temporal rate. More than two-thirds of the binocular cells showed phase specificity to static phase disparities leading to either summation or facilitation interactions. This proportion of spatial disparity selectivity was higher than that shown for the same area (one-third of the units) when drifting light bars or drifting spatial frequencies were used to create disparities. The range of phase disparities encoded by binocular cells in area 19 is inversely related to the optimal spatial frequency of the dominant eye. Thus, cells in this area are tuned to coarse spatial disparities which, as supported by behavioural studies, could reflect its involvement in the analysis of stereoscopic pattern having gross disparities but devoid of motion cues. Because of the nature of its interconnections with numerous visual cortical areas, area 19 could serve as a way station where stereoscopic information could be first analysed and sent to other higher order areas for a complete representation of three-dimensional objects.
Collapse
|
9
|
Abstract
The posteromedial lateral suprasylvian area of the cat is known to be involved in the analysis of motion and motion in depth. However, it remains unclear whether binocular cells in the posteromedial lateral suprasylvian area rely upon phase or positional offsets between their receptive fields in order to code binocular disparity. The present study aims at clarifying more precisely the neural mechanisms underlying stereoperception with two objectives in mind. First, to determine whether cells in the posteromedial lateral suprasylvian area code phase disparities. Secondly, to examine whether the cells sensitive to phase disparity are the same as those which code for position disparities or whether each group represent a different sub-population of disparity-sensitive neurons. We investigated this by testing both types of disparities on single neurons in this area. The results show that the vast majority of cells (74%), in the posteromedial lateral suprasylvian area, are sensitive to relative interocular phase disparities. These cells showed mostly facilitation (95%) and a few (5%) summation interactions. Moreover, most cells (81%) were sensitive to both position and phase disparities. The results of this study show that most binocular cells in the posteromedial lateral suprasylvian area are sensitive to both positional and phase offsets which demonstrate the importance of this area in stereopsis.
Collapse
|
10
|
Abstract
Binocular cells in area 19 are tuned to positional disparities. In effect, up to one-third of the cells respond preferentially to small incongruities between the optimal bar stimuli presented within the receptive fields of each eye. The aim of the present study was to determine whether cells in area 19 are also sensitive to phase disparities. Both types of disparities have been proposed as mechanisms through which stereoperception is achieved. Results indicate that phase disparities produced coherent interactions in 38% of the binocular cells, resulting in facilitation or summation. The remaining cells were phase insensitive. The overall results suggest that cells in area 19 code phase disparities in a proportion comparable to stimulus disparities, confirming that this area is implicated in binocular integration, albeit in a relatively smaller proportion than some of the other visual areas.
Collapse
|
11
|
Abstract
Binocular disparity provides the visual system with information concerning the three-dimensional layout of the environment. Recent physiological studies in the primary visual cortex provide a successful account of the mechanisms by which single neurons are able to signal disparity. This work also reveals that additional processing is required to make explicit the types of signal required for depth perception (such as the ability to match features correctly between the two monocular images). Some of these signals, such as those encoding relative disparity, are found in extrastriate cortex. Several other lines of evidence also suggest that the link between perception and neuronal activity is stronger in extrastriate cortex (especially MT) than in the primary visual cortex.
Collapse
|
12
|
Abstract
Previous studies of the visual system of Siamese cats have shown that binocular cells are scarce in areas 17, 18 and 19, yet significantly more abundant in suprasylvian areas such as the postero-medial lateral suprasylvian area (PMLS). The present study aims at evaluating the sensitivity to spatial disparity of PMLS binocular cells in paralyzed and anesthetized Siamese cats. Centrally located receptive fields were mapped, separated using prisms and then stimulated simultaneously using two luminous bars optimally adjusted to the size of the excitatory receptive fields. Delays were introduced in the arrival of the luminous bars in the receptive fields so as to create the desired spatial disparities. Results indicate that approximately a third of PMLS units are binocular and that these binocular cells can detect spatial disparity cues. Indeed, although the sample was relatively small, cells of the tuned excitatory (14/34), tuned inhibitory (2/34), near (6/34) and far (1/34) types were identified. The spatial selectivity, as measured by the width at half height of the tuning curves of the excitatory and inhibitory cells and the slopes of the near and far cells, was similar to that obtained in PMLS of normal cats but not as precise as that found for primary visual areas in these animals. This suggests that these cells might serve as a substrate for coarse stereopsis.
Collapse
|
13
|
Abstract
The present study examined the neuronal sensitivity in area 19 of the cat to a motion-defined bar and to texture. Sensitivity was tested in normal, lesioned (areas 17-18) and split-chiasm cats using a kinematogram, as well as a textured bar drifting on a uniform light background and a light bar drifting on a stationary textured background. Texture density was varied. The results indicate that almost all cells of area 19 recorded in the three groups of cats responded to a motion-defined bar or to its edges. Texture density influenced the responses in that the discharge rate increased as density decreased. However, the majority of cells were sensitive to the highest texture density kinematogram. Moreover, the neural responses of all cats were either independent of the density of the textured bar or background, or were modulated by it. These results show that cells in area 19 can signal the presence of a kinetic bar and that the density of either the textured bar, the background or both can influence figure-ground detection. The results are interpreted with respect to how various inputs influence the function of area 19.
Collapse
|
14
|
Spatial and temporal matching of receptive field properties of binocular cells in area 19 of the cat. Neuroscience 1998; 86:121-34. [PMID: 9692748 DOI: 10.1016/s0306-4522(98)00020-7] [Citation(s) in RCA: 16] [Impact Index Per Article: 0.6] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 02/08/2023]
Abstract
The spatial and temporal properties of single neurons were investigated in area 19 of the cat. We evaluated the matching of binocular receptive field properties with regard to the respective strength of the ipsilateral and contralateral inputs. Results indicate that most cells in area 19 are well tuned to spatial and temporal frequencies and exhibit relatively low contrast threshold (mean=6.8%) when assessed using optimal parameters and tested through the dominant eye. Spatial resolution (mean=0.75 c/degree), optimal spatial frequencies (mean=0.16 c/degree) were relatively low and spatial bandwidths (mean=2.1 octaves) were broader as compared to those of cells in area 17 but comparable to those of cells in other extrastriate areas. On the other hand temporal resolution (mean=10.7 Hz), optimal temporal frequency (mean=4.5 Hz) and temporal bandwidths (mean=2.9 octaves) were higher and broader than in primary visual cortex. A significant relationship exists between most of the cell's properties assessed through either eye. For some parameters, such as spatial and temporal resolution, ocular dominance was shown to be significantly related to the extent of matching between the two eyes. For these parameters, binocular cells that exhibited a balanced ocular dominance were generally well matched with regard to the receptive field properties of each eye whereas the largest mismatches were found in cells that were more strongly dominated by one eye. These results suggest that visual input contributes to the activation of cells in area 19 in a redundant manner, possibly attesting to the multiplicity of parallel pathways to this area in the cat.
Collapse
|
15
|
Abstract
The spatial disparity sensitivity of single units in the primary visual cortex (17-18 border), in extrastriate area 19 and in the superficial layers of the superior colliculus of the cat brain were compared in the present study. Unit recordings were performed in paralyzed and anesthetized animals. Centrally located receptive fields were mapped, separated using prisms and then stimulated simultaneously using two luminous bars optimally adjusted to the size of the excitatory receptive fields. In the three regions studied, cells selective to spatial disparity were found and four classes of disparity sensitivity profiles emerged. Although the disparity sensitivity profiles of the cells in the three regions appeared to have the same general shape, selectivity was clearly different. Cells at the 17-18 border were sharply tuned, those of area 19 were not only less numerous but also less well tuned and collicular cells exhibited coarse selectivity. These differences in selectivity appear to be linked to the projection pattern of the X, Y and W systems to these regions and the roles that these cells might play in vision.
Collapse
|
16
|
Cortical projections of the parvocellular laminae C of the dorsal lateral geniculate nucleus in the cat: an anterograde wheat germ agglutinin conjugated to horseradish peroxidase study. J Comp Neurol 1998; 392:439-57. [PMID: 9514509 DOI: 10.1002/(sici)1096-9861(19980323)392:4<439::aid-cne3>3.0.co;2-1] [Citation(s) in RCA: 23] [Impact Index Per Article: 0.9] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 02/06/2023]
Abstract
The areal and laminar distributions of the projection from the parvocellular part of laminae C of the dorsal lateral geniculate nucleus (Cparv) were studied in visual cortical areas of the cat with the anterograde tracing method by using wheat germ agglutinin conjugated to horseradish peroxidase. A particular objective of this study was to examine the central visual pathways of the W-cell system, the precise organization of which is still unknown. Because the Cparv in the cat is said to receive W-cell information exclusively from the retina and the superior colliculus, the results obtained would provide an anatomical substrate for the W-cell system organization in mammals. The results show that the cortical targets of the Cparv are areas 17, 18, 19, 20a, and 21a and the posteromedial lateral suprasylvian (PMLS) and ventral lateral suprasylvian(VLS) areas. In area 17, the projection fibers terminate in the superficial half of layer I; the lower two-thirds of layer III, extending to the superficial part of layer IV; and the deep part of layer IV, involving layer Va. These terminations form triple bands in area 17. The projection terminals in layer I are continuous, whereas those in layers III, IV, and Va distribute periodically, exhibiting a patchy appearance. In areas 18 and 19, the projection fibers terminate in the superficial half of layer I and in the full portions of layers III and IV, forming double bands. In these areas, the terminals in layer I are continuous, whereas those in layers III and IV distribute periodically, exhibiting a patchy appearance. In area 20a, area 21a, PMLS, and VLS, projection fibers terminate in the superficial part of layer I, in part of layer III, and in the full portion of layer IV, although they are far fewer in number than those seen in areas 17, 18, and 19. The present results demonstrate that the Cparv fibers terminate in a localized fashion in both the striate and the extrastriate cortical areas and that these W-cell projections are quite unique in their areal and laminar organization compared with the X- and Y-cell systems.
Collapse
|
17
|
Spatial resolution and contrast sensitivity of single neurons in area 19 of split-chiasm cats: a comparison with primary visual cortex. Eur J Neurosci 1997; 9:1929-39. [PMID: 9383216 DOI: 10.1111/j.1460-9568.1997.tb00760.x] [Citation(s) in RCA: 13] [Impact Index Per Article: 0.5] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 02/05/2023]
Abstract
Electrophysiological recordings were carried out in the callosal recipient zone of area 19 in normal and split-chiasm cats and, for comparison purposes, at the border of areas 17 and 18 of split-chiasm cats. The influences of retinothalamic and callosal inputs on a single cortical neurons were thereby evaluated. Extracellular recordings of single cells were made in anaesthetized and paralysed cats in the zone representing the central visual field. Receptive field properties were assessed using sine wave gratings drifting in optimal directions. Results showed that in area 19 and areas 17/18 one-third of the cells were binocularly driven after section of the optic chiasm. In area 19, the spatial resolution and contrast sensitivity of cells driven via the dominant eye were similar in the normal and split-chiasm groups. In areas 17/18 and area 19 of split-chiasm cats, binocular cells showed significant interocular matching of their receptive field properties (spatial resolution and contrast threshold), although small differences were observed. These small interocular differences were related to the cell's ocular dominance rather than to the signal transmission route (thalamic or callosal).
Collapse
|
18
|
Spatial and temporal frequency tuning and contrast sensitivity of single neurons in area 21a of the cat. Brain Res 1996; 716:219-23. [PMID: 8738243 DOI: 10.1016/0006-8993(96)00031-5] [Citation(s) in RCA: 19] [Impact Index Per Article: 0.7] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 02/01/2023]
Abstract
The spatial and temporal selectivities of single neurons in area 21a of the adult cat were investigated using sinusoidal gratings. Optimal spatial frequencies and visual acuity (high cut-off frequency) were fairly low and spatial bandwidth was mainly narrow. Contrast threshold was generally low but a substantial number of cells were only excited by high contrast stimuli. The temporal selectivity suggests that cells responded to a wide range of temporal frequencies.
Collapse
|