A model of binocular stereopsis including a global consistency constraint

 The binocular correspondence problem was solved by implementing the uniqueness constraint and the continuity constraint, as proposed by Marr and Poggio [Marr D, PoggioT (1976) Science 194: 283–287]. However, these constraints are not sufficient to define the proper correspondence uniquely. With these constraints, random-dot stereograms (RDSs), consisting of the periodic textures in each image, are treated as a correspondence of surfaces composed of patches of alternating values of disparity. This is quite different from the surface we perceive through the RDSs, that is a surface characterized by a single depth. Because these constraints are local, they cannot produce the global optimum of correspondence. To obtain the global optimum of correspondence, we propose a model of binocular stereopsis in which a global measure of correspondence is explicitly employed. The model consists of two hierarchical systems. First, the lower system processes various correspondences based on the uniqueness constraint. Second, the higher system provides a global measure of correspondence for the disparity in question. The higher system uniquely determines the global optimum of correspondence in the lower system through the recurrent loop between hierarchical systems. The convergence of the recurrent loop is determined by the consistency between the hierarchical systems. The condition is termed the `global consistency constraint. Received: 27 August 1998 / Accepted in revised form: 8 November 1999     

Geometry and dynamics of activity-dependent homeostatic regulation in neurons

To maintain activity in a functional range, neurons constantly adjust membrane excitability to changing intra- and extracellular conditions. Such activity-dependent homeostatic regulation (ADHR) is critical for normal processing of the nervous system and avoiding pathological conditions. Here, we posed a homeostatic regulation problem for the classical Morris-Lecar (ML) model. The problem was motivated by the phenomenon of the functional recovery of stomatogastric neurons in crustaceans in the absence of neuromodulation. In our study, the regulation of the ionic conductances in the ML model depended on the calcium current or the intracellular calcium concentration. We found an asymptotic solution to the problem under the assumption of slow regulation. The solution provides a full account of the regulation in the case of correlated or anticorrelated changes of the maximal conductances of the calcium and potassium currents. In particular, the solution shows how the target and parameters of the regulation determine which perturbations of the conductances can be compensated by the ADHR. In some cases, the sets of compensated initial perturbations are not convex. On the basis of our analysis we formulated specific questions for subsequent experimental and theoretical studies of ADHR.     

Comparison of the Intrinsic Dynamics of Aminoacyl-tRNA Synthetases

Aminoacyl-tRNA synthetases (AARSs) are an important family of enzymes that catalyze tRNA aminoacylation reaction (Ibba and Soll in Annu Rev Biochem 2000, 69:617–650) [1]. AARSs are grouped into two broad classes (class I and II) based on sequence/structural homology and mode of their interactions with the tRNA molecule (Ibba and Soll in Annu Rev Biochem 2000, 69:617–650) [1]. As protein dynamics play an important role in enzyme function, we explored the intrinsic dynamics of these enzymes using normal mode analysis and investigated if the two classes and six subclasses (Ia–c and IIa–c) of AARSs exhibit any distinct patterns of motion. The present study found that the intrinsic dynamics-based classification of these enzymes is similar to that obtained based on sequence/structural homology for most enzymes. However, the classification of seryl-tRNA synthetase was not straightforward; the internal mobility patterns of this enzyme are comparable to both IIa and IIb AARSs. This study revealed only a few general mobility patterns in these enzymes—(1) the insertion domain is generally engaged in anticorrelated motion with respect to the catalytic domain for both classes of AARSs and (2) anticodon binding domain dynamics are partly correlated and partly anticorrelated with respect to other domains for class I enzymes. In most of the class II AARSs, the anticodon binding domain is predominately engaged in anticorrelated motion with respect to the catalytic domain and correlated to the insertion domain. This study supports the notion that dynamic-based classification could be useful for functional classification of proteins.     

Disulfide linkage of biotin identifies a 106-kDa Ca2+ release channel in sarcoplasmic reticulum

Reactive disulfide reagents (RDSs) with a biotin moiety have been synthesized and found to cause Ca2+ release from sarcoplasmic reticulum (SR) vesicles. The RDSs oxidize SH sites on SR proteins via a thiol-disulfide exchange, with the formation of mixed disulfide bonds between SR proteins and biotin. Biotinylated RDSs identified a 106-kDa protein which was purified by biotin-avidin chromatography. Disulfide reducing agents, like dithiothreitol, reverse the effect of RDSs and thus promoted active re-uptake of Ca2+ and dissociated biotin from the labeled protein indicating that biotin was covalently linked to the 106-kDa protein via a disulfide bond. Several lines of evidence indicate that this protein is not Ca2+, Mg2+-ATPase and is not a proteolytic fragment or a subunit of the 400-kDa Ca2+-ryanodine receptor complex (RRC). Monoclonal antibodies against the ATPase did not cross-react with the 106-kDa protein, and polyclonal antibodies against the 106-kDa did not cross-react with either the ATPase or the 400-kDa RRC. RDSs did not label the 400-kDa RRC with biotin. Linear sucrose gradients used to purify the RRC show that the 106-kDa protein migrated throughout 5-20% linear sucrose gradients, including the high sucrose density protein fractions containing 400-kDa RRC. Protease inhibitors diisopropylfluorophosphate used to prevent proteolysis of 400-kDa proteins did not alter the migration of 106-kDa in sucrose gradients nor the patterns of biotin labeling of the 106-kDa protein. Incorporation of highly purified 106-kDa protein (free of RRC) in planar bilayers revealed cationic channels with large Na+ (gNa+ = 375 +/- 15 pS) and Ca2+ (gCa2+ = 107.7 +/- 12 pS) conductances which were activated by micromolar [Ca2+]free or millimolar [ATP] and blocked by micromolar ruthenium red or millimolar [Mg2+]. Thus, the SR contains a sulfhydryl-activated 106-kDa Ca2+ channel with apparently similar characteristics to the 400-kDa "feet" proteins.     

Visual experience before eye-opening and the development of the retinogeniculate pathway

Visual experience before eye-opening is not usually thought to have any developmental significance. Here we show that naturalistic visual stimuli presented through unopened eyelids robustly activate neurons in the ferret dorsal lateral geniculate nucleus. Further, dark-rearing prior to natural eye-opening has striking effects upon geniculate physiology. Receptive field maps after dark-rearing show increased convergence of On- and Off-center responses, and neurons frequently respond to both bright and dark phases of drifting gratings. There is also increased selectivity for the orientation of the gratings. These abnormalities of On-Off segregation can be explained by the finding that the responses of immature On and Off cells to naturalistic stimuli are strongly anticorrelated.     

Restrained molecular dynamics simulations of HIV-1 protease: the first step in validating a new target for drug design

To test the anticorrelated relationship that was recently displayed in conventional molecular dynamics (MD) simulations, several different restrained MD simulations on a wild type and on the V82F/I84V drug-resistant mutant of HIV-1 protease were performed. This anticorrelated relationship refers to the observation that compression of the peripheral ear-to-cheek region of HIV protease (i.e., the elbow of the flap to the fulcrum and the cantilever) occurred as the active site flaps were opening, and, conversely, expansion of that ear-to-cheek region occurred as both flaps were closing. An additional examination of this anticorrelated relationship was necessary to determine whether it can be harnessed in a useful manner. Consequently, six different MD experiments were performed that incorporated pairwise distance restraints in that ear-to-cheek region (i.e., the distance between the alpha-carbons of Gly40 and Gln61 was restrained to either 7.7 or 10.5 A, in both monomers). Pushing the backbones of the ear and the cheek regions away from each other slightly did force the flaps that guard the active site to remain closed in both the wild type and the mutant systems-even though there were no ligands in the active sites. Thus, these restrained MD simulations provided evidence that the anticorrelated relationship can be exploited to affect the dynamic behavior of the flaps that guard the active site of HIV-1 protease. These simulations supported our hypothesis of the mechanism governing flap motion, and they are the first step towards validating that peripheral surface as a new target for drug design.     

Reactive disulfide compounds induce Ca2+ release from cardiac sarcoplasmic reticulum

Reactive disulfide compounds (RDSs) with a pyridyl ring adjacent to a disulfide bond, 2,2'dithiodipyridine (2,2' DTDP) and 4,4' dithiodipyridine (4,4' DTDP), induce Ca2+ release from isolated canine cardiac sarcoplasmic reticulum (SR) vesicles. RDSs are absolutely specific to free sulfhydryl (SH) groups and oxidize SH sites of low pKa via a thiol-disulfide exchange reaction, with the stoichiometric production of thiopyridone in the medium. As in skeletal SR, this reaction caused large increases in the Ca2+ permeability of cardiac SR and the number of SH sites oxidized by RDSs was kinetically and quantitatively measured through the absorption of thiopyridone. RDS-induced Ca2+ release from cardiac SR was characterized and compared to the action of RDSs on skeletal SR and to Ca2(+)-induced Ca2+ release. (i) RDS-induced Ca2+ release from cardiac SR was dependent on ionized Mg2+, with maximum rates of release occurring at 0.5 and 1 mM Mg2+free for 2,2' DTDP and 4,4' DTDP, respectively. (ii) In the presence of adenine nucleotides (0.1-1 mM), the oxidation of SH sites in cardiac SR by exogenously added RDS was inhibited, which, in turn, inhibited Ca2+ release induced by RDSs. (iii) Conversely, when the oxidation reaction between RDSs and cardiac SR was completed and Ca2+ release pathways were opened, subsequent additions of adenine nucleotides stimulated Ca2+ efflux induced by RDSs. (iv) Sulfhydryl reducing agents (e.g., dithiothreitol, DTT, 1-5 mM) inhibited RDS-induced Ca2+ efflux in a concentration-dependent manner. (v) RDSs elicited Ca2+ efflux from passively loaded cardiac SR vesicles (i.e., with nonfunctional Ca2+ pumps in the absence of Mg-ATP) and stimulated Ca2(+)-dependent ATPase activity, which indicated that RDS uncoupled Ca2+ uptake and did not act at the Ca2+, Mg2(+)-ATPase. These results indicate that RDSs selectively oxidize critical sulfhydryl site(s) on or adjacent to a Ca2+ release channel protein channel and thereby trigger Ca2+ release. Conversely, reduction of these sites reverses the effects of RDSs by closing Ca2+ release channels, which results in active Ca2+ reuptake by Ca2+, Mg2(+)-ATPase. These compounds can thus provide a method to covalently label and identify the protein involved in Ca2+ release from cardiac SR.     

The relationship between the longitudinal pressure gradient and the blood flow velocity in the canine superior vena cava

In the superior vena cava of anaesthetized open chest dogs the axial pressure gradient (delta P) was measured simultaneously with the blood flow velocity (V) under a variety of preload conditions. Both delta P and V curves showed distinct systolic and diastolic waves. Peak delta P ranged between 26 and 93 P/cm (0.2-0.7 mm Hg/cm) and V varied between 0.095 and 0.19 m/s. Peak systolic delta P, but not peak diastolic delta P was significantly linearly correlated to respectively peak systolic V and peak diastolic V. The shape of delta P and V curves corresponded fairly well but variations of delta P preceded the variations of V. Both the shape correspondence and the phase lag between delta P and V were evaluated by means of the normalized cross-correlation technique. During volume expansion the shape correspondence improved and the phase lag decreased. It is concluded that the transient vena caval blood velocity variations are directly related to the pulsatile axial pressure gradient.     

End-stopping and the aperture problem: two-dimensional motion signals in macaque V1

Our perception of fine visual detail relies on small receptive fields at early stages of visual processing. However, small receptive fields tend to confound the orientation and velocity of moving edges, leading to ambiguous or inaccurate motion measurements (the aperture problem). Thus, it is often assumed that neurons in primary visual cortex (V1) carry only ambiguous motion information. Here we show that a subpopulation of V1 neurons is capable of signaling motion direction in a manner that is independent of contour orientation. Specifically, end-stopped V1 neurons obtain accurate motion measurements by responding only to the endpoints of long contours, a strategy which renders them largely immune to the aperture problem. Furthermore, the time course of end-stopping is similar to the time course of motion integration by MT neurons. These results suggest that cortical neurons might represent object motion by responding selectively to two-dimensional discontinuities in the visual scene.     

Reactive disulfides trigger Ca2+ release from sarcoplasmic reticulum via an oxidation reaction

Reactive disulfide compounds (RDSs) with a pyridyl ring adjacent to the S-S bond such as 2,2'-dithiodipyridine (2,2'-DTDP), 4,4'-dithiodipyridine, and N-succinimidyl 3(2-pyridyldithio)propionate (SPDP) trigger Ca2+ release from sarcoplasmic reticulum (SR) vesicles. They are known to specifically oxidize free SH sites via a thiol-disulfide exchange reaction with the stoichiometric production of thiopyridone. Thus, the formation of a mixed S-S bond between an accessible SH site on an SR protein and a RDS causes large increases in SR Ca2+ permeability. Reducing agents, glutathione (GSH) or dithiothreitol reverse the effect of RDSs and permit rapid re-uptake of Ca2+ by the Ca2+, Mg2+-ATPase. The RDSs, 2,2'-DTDP, 4,4'-dithiodipyridine and SPDP displaced [3H]ryanodine binding to the Ca2+-receptor complex at IC50 values of 7.5 +/- 0.2, 1.5 +/- 0.1, and 15.4 +/- 0.1 microM, respectively. RDSs did not alter the rapid initial phase of Ca2+ uptake by the pump, stimulated ATPase activity, and induced release from passively loaded vesicles with nonactivated pumps; thus they act at a Ca2+ release channel and not at the Ca2+, Mg2+-ATPase. Efflux rates increased in 0.25-1.0 mM [Mg2+]free then decreased in 2-5 mM [Mg2+]free. Adenine nucleotides inhibited the oxidation of SHs on SR protein by RDSs and thus reduced Ca2+ efflux rates. However, once RDSs oxidized these SH sites and opened the Ca2+ release pathway, subsequent additions of nucleotides stimulated Ca2+ efflux. In skinned fibers, 2,2'-dithiodipyridine elicited rapid twitches which were blocked by ruthenium red. These results indicate that RDSs trigger Ca2+ release from SR by oxidizing a critical SH group, and thus provide a method to covalently label the protein(s) involved in causing these changes in Ca2+ permeability.     

Cytological observations on the ventromedial hypothalamic nucleus

Two sorts of neurons are recognized in Golgi impregnations of the rat ventromedial hypothalamic nucleus (HVM). The two cell types, category I and II neurons, are differentiated on the basis of their somatic, dendritic, and axonal characteristics. Category I neurons form most of the neuronal population and are located throughout HVM. The small number of category II neurons that have been studied occur in lateral HVM. Two varieties of neuronal profile, "common" and "uncommon cells", are seen in thin sections of HVM. The "uncommon cells", in comparison with the "common ones", appear to have a larger soma, a more electron-dense cytoplasmic matrix, an abundance of Nissl bodies, and a population of dense-cored vesicles (100--130 nm in diameter). Some of the somata and proximal dendrites of "common", but not "uncommon" cells, are wrapped in multiple layers of astrocytic processes. Although the correlation is tentative, it is argued that category I neurons correspond to "common cells" and category II, to "uncommon cells". One possible implication of this correspondence is discussed regarding neuronal alteration in response to change in the endocrinological environment of the brain.     

Inferring network activity from synaptic noise

During intense network activity in vivo, cortical neurons are in a high-conductance state, in which the membrane potential (V(m)) is subject to a tremendous fluctuating activity. Clearly, this "synaptic noise" contains information about the activity of the network, but there are presently no methods available to extract this information. We focus here on this problem from a computational neuroscience perspective, with the aim of drawing methods to analyze experimental data. We start from models of cortical neurons, in which high-conductance states stem from the random release of thousands of excitatory and inhibitory synapses. This highly complex system can be simplified by using global synaptic conductances described by effective stochastic processes. The advantage of this approach is that one can derive analytically a number of properties from the statistics of resulting V(m) fluctuations. For example, the global excitatory and inhibitory conductances can be extracted from synaptic noise, and can be related to the mean activity of presynaptic neurons. We show here that extracting the variances of excitatory and inhibitory synaptic conductances can provide estimates of the mean temporal correlation-or level of synchrony-among thousands of neurons in the network. Thus, "probing the network" through intracellular V(m) activity is possible and constitutes a promising approach, but it will require a continuous effort combining theory, computational models and intracellular physiology.     

Intracortical connections between neuron groups in the somatosensory cortex studied by the retrograde horseradish peroxidase transport method in rats

Short corticocortical connections between specialized groups of neurons (so-called barrels) were studied in the somatosensory cortex. After microinjections of horseradish peroxidase into a definite "barrel" labeled neurons were found in nearby groups within a radius of up to 400 µ. Labeled neurons were located chiefly in cortical layers V and III; 90% of them were pyramidal cells. Intracortical connection of labeled neurons were 1.6 times more numerous than thalamocortical connections. It is postulated that connections between neighboring cortical neuron groups are effected through their output cells, i.e., through pyramidal neurons of layers V and III.     

Pre-attentive segmentation and correspondence in stereo

Traditional stereo grouping models have focused on the problem of stereo correspondence between monocular inputs. Recent physiological data revealed that the disparity selective V2 cells increase their responses when (random-dot stereograms) stimuli within their receptive fields are at or near the boundary of a depth surface. Such highlights to depth (non-luminance) edges are seemingly not computationally required for the correspondence problem. Computationally, these highlights make the boundaries of a depth surface more salient, serving pre-attentive segmentation (between depth planes) and attracting visual attention. In special cases, they enable the psychophysically observed perceptual pop-out of a target from a background of visually identical distractors at a different depth. To achieve the highlights, mutual inhibition between disparity selective cells that are tuned to the same or similar depths is required. However, such mutual inhibition would impede the computation for the correspondence problem, which requires mutual excitation between the same cells. In this work, I introduce a computational model that, I believe, is the first to address both stereo correspondence and pre-attentive stereo segmentation. The computational mechanisms in the model are based on intracortical interactions in V2. I will demonstrate that the model captures the following physiological and psychophysical phenomena: (i) depth-edge highlighting; (ii) disparity capture; (iii) pop-out; and (iv) transparency.     

Adjusted regularization of cortical covariance

It is now common to record dozens to hundreds or more neurons simultaneously, and to ask how the network activity changes across experimental conditions. A natural framework for addressing questions of functional connectivity is to apply Gaussian graphical modeling to neural data, where each edge in the graph corresponds to a non-zero partial correlation between neurons. Because the number of possible edges is large, one strategy for estimating the graph has been to apply methods that aim to identify large sparse effects using an \(L_{1}\) penalty. However, the partial correlations found in neural spike count data are neither large nor sparse, so techniques that perform well in sparse settings will typically perform poorly in the context of neural spike count data. Fortunately, the correlated firing for any pair of cortical neurons depends strongly on both their distance apart and the features for which they are tuned. We introduce a method that takes advantage of these known, strong effects by allowing the penalty to depend on them: thus, for example, the connection between pairs of neurons that are close together will be penalized less than pairs that are far apart. We show through simulations that this physiologically-motivated procedure performs substantially better than off-the-shelf generic tools, and we illustrate by applying the methodology to populations of neurons recorded with multielectrode arrays implanted in macaque visual cortex areas V1 and V4.     

Membrane potential synchrony in primary visual cortex during sensory stimulation

When the primary visual cortex (V1) is activated by sensory stimulation, what is the temporal correlation between the synaptic inputs to nearby neurons? This question underlies the origin of correlated activity, the mechanism of how visually evoked activity emerges and propagates in cortical circuits, and the relationship between spontaneous and evoked activity. Here, we have recorded membrane potential from pairs of V1 neurons in anesthetized cats and found that visual stimulation suppressed low-frequency membrane potential synchrony (0-10 Hz), and often increased synchrony at high frequencies (20-80 Hz). The increase in high-frequency synchrony occurred for neurons with similar orientation preferences and for neurons with different orientation preferences and occurred for a wide range of stimulus orientations. Thus, while only a subset of neurons spike in response to visual stimulation, a far larger proportion of the circuit is correlated with spiking activity through subthreshold, high-frequency synchronous activity that crosses functional domains.     

A model for the neuronal implementation of selective visual attention based on temporal correlation among neurons

We propose a model for the neuronal implementation of selective visual attention based on temporal correlation among groups of neurons. Neurons in primary visual cortex respond to visual stimuli with a Poisson distributed spike train with an appropriate, stimulus-dependent mean firing rate. The spike trains of neurons whose receptive fields donot overlap with the focus of attention are distributed according to homogeneous (time-independent) Poisson process with no correlation between action potentials of different neurons. In contrast, spike trains of neurons with receptive fields within the focus of attention are distributed according to non-homogeneous (time-dependent) Poisson processes. Since the short-term average spike rates of all neurons with receptive fields in the focus of attention covary, correlations between these spike trains are introduced which are detected by inhibitory interneurons in V4. These cells, modeled as modified integrate-and-fire neurons, function as coincidence detectors and suppress the response of V4 cells associated with non-attended visual stimuli. The model reproduces quantitatively experimental data obtained in cortical area V4 of monkey by Moran and Desimone (1985).     

A laminar cortical model of stereopsis and 3D surface perception: closure and da Vinci stereopsis

A laminar cortical model of stereopsis and 3D surface perception is developed and simulated. The model describes how monocular and binocular oriented filtering interact with later stages of 3D boundary formation and surface filling-in in the LGN and cortical areas V1, V2, and V4. It proposes how interactions between layers 4, 3B, and 2/3 in V1 and V2 contribute to stereopsis, and how binocular and monocular information combine to form 3D boundary and surface representations. The model includes two main new developments: (1) It clarifies how surface-to-boundary feedback from V2 thin stripes to pale stripes helps to explain data about stereopsis. This feedback has previously been used to explain data about 3D figure-ground perception. (2) It proposes that the binocular false match problem is subsumed under the Gestalt grouping problem. In particular, the disparity filter, which helps to solve the correspondence problem by eliminating false matches, is realized using inhibitory interneurons as part of the perceptual grouping process by horizontal connections in layer 2/3 of cortical area V2. The enhanced model explains all the psychophysical data previously simulated by Grossberg and Howe (2003), such as contrast variations of dichoptic masking and the correspondence problem, the effect of interocular contrast differences on stereoacuity, Panum's limiting case, the Venetian blind illusion, stereopsis with polarity-reversed stereograms, and da Vinci stereopsis. It also explains psychophysical data about perceptual closure and variations of da Vinci stereopsis that previous models cannot yet explain.     

When correlation implies causation in multisensory integration

Inferring which signals have a common underlying cause, and hence should be integrated, represents a primary challenge for a perceptual system dealing with multiple sensory inputs [1-3]. This challenge is often referred to as the correspondence problem or causal inference. Previous research has demonstrated that spatiotemporal cues, along with prior knowledge, are exploited by the human brain to solve this problem [4-9]. Here we explore the role of correlation between the fine temporal structure of auditory and visual signals in causal inference. Specifically, we investigated whether correlated signals are inferred to originate from the same distal event and hence are integrated optimally [10]. In a localization task with visual, auditory, and combined audiovisual targets, the improvement in precision for combined relative to unimodal targets was statistically optimal only when audiovisual signals were correlated. This result demonstrates that humans use the similarity in the temporal structure of multisensory signals to solve the correspondence problem, hence inferring causation from correlation.     

An inference upon the neural network finding binocular correspondence

Previously, the authors proposed a model of neural network extracting binocular parallax (Hirai and Fukushima, 1975). It is a multilayered network whose final layers consist of neural elements corresponding to binocular depth neurons found in monkey's visual cortex. The binocular depth neuron is selectively sensitive to a binocular stimulus with a specific amount of binocular parallax and does not respond to a monocular one. As described in the last chapter of the previous article (Hirai and Fukushima, 1975), when a binocular pair of input patterns consist of, for example, many vertical bars placed very closely to each other, the binocular depth neurons might respond not only to correct binocular pairs, but also to incorrect ones. Our present study is concentrated upon how the visual system finds correct binocular pairs or binocular correspondence. It is assumed that some neural network is cascaded after the binocular depth neurons and finds out correct binocular correspondence by eliminating the incorrect binocular pairs. In this article a model of such neural network is proposed. The performance of the model has been simulated on a digital computer. The results of the computer simulation show that this model finds binocular correspondence satisfactorily. It has been demonstrated by the computer simulation that this model also explains the mechanism of the hysteresis in the binocular depth perception reported by Fender and Julesz (1967)This work has been done in the NHK Broadcasting Science Research Laboratories