Complexity and diversity in sparse code priors improve receptive field characterization of Macaque V1 neurons

System identification techniques—projection pursuit regression models (PPRs) and convolutional neural networks (CNNs)—provide state-of-the-art performance in predicting visual cortical neurons’ responses to arbitrary input stimuli. However, the constituent kernels recovered by these methods are often noisy and lack coherent structure, making it difficult to understand the underlying component features of a neuron’s receptive field. In this paper, we show that using a dictionary of diverse kernels with complex shapes learned from natural scenes based on efficient coding theory, as the front-end for PPRs and CNNs can improve their performance in neuronal response prediction as well as algorithmic data efficiency and convergence speed. Extensive experimental results also indicate that these sparse-code kernels provide important information on the component features of a neuron’s receptive field. In addition, we find that models with the complex-shaped sparse code front-end are significantly better than models with a standard orientation-selective Gabor filter front-end for modeling V1 neurons that have been found to exhibit complex pattern selectivity. We show that the relative performance difference due to these two front-ends can be used to produce a sensitive metric for detecting complex selectivity in V1 neurons.     

A network that uses few active neurones to code visual input predicts the diverse shapes of cortical receptive fields

Computational models of primary visual cortex have demonstrated that principles of efficient coding and neuronal sparseness can explain the emergence of neurones with localised oriented receptive fields. Yet, existing models have failed to predict the diverse shapes of receptive fields that occur in nature. The existing models used a particular "soft" form of sparseness that limits average neuronal activity. Here we study models of efficient coding in a broader context by comparing soft and "bard" forms of neuronal sparseness. As a result of our analyses, we propose a novel network model for visual cortex. The model forms efficient visual representations in which the number of active neurones, rather than mean neuronal activity, is limited. This form of hard sparseness also economises cortical resources like synaptic memory and metabolic energy. Furthermore, our model accurately predicts the distribution of receptive field shapes found in the primary visual cortex of cat and monkey.     

Pooling strategies in V1 can account for the functional and structural diversity across species

Neurons in the primary visual cortex are selective to orientation with various degrees of selectivity to the spatial phase, from high selectivity in simple cells to low selectivity in complex cells. Various computational models have suggested a possible link between the presence of phase invariant cells and the existence of orientation maps in higher mammals’ V1. These models, however, do not explain the emergence of complex cells in animals that do not show orientation maps. In this study, we build a theoretical model based on a convolutional network called Sparse Deep Predictive Coding (SDPC) and show that a single computational mechanism, pooling, allows the SDPC model to account for the emergence in V1 of complex cells with or without that of orientation maps, as observed in distinct species of mammals. In particular, we observed that pooling in the feature space is directly related to the orientation map formation while pooling in the retinotopic space is responsible for the emergence of a complex cells population. Introducing different forms of pooling in a predictive model of early visual processing as implemented in SDPC can therefore be viewed as a theoretical framework that explains the diversity of structural and functional phenomena observed in V1.     

A model for the temporal organization of X- and Y-type receptive fields in the primate retina

A model is proposed for the temporal characteristics of X-and Y-type responses of ganglion cells in the primate retina. The main suggestions of the model are: (I) The X-type temporal response is determined primarily by the delay between center and surround contributions. (II) The Y-type response is generated in the inner plexiform layer by a derivativelike operation on the bipolar cell's input, followed by a rectification in the convergence of these inputs onto the Y-ganglion-cell. (III) The derivative-like operation is obtained by recurrent inhibition in the dyad synaptic structure.The X-and Y-type responses predicted by the model, for a variety of stimuli, were examined and compared with available electrophysiological recordings. Finally, certain predictions derived from the model are discussed.     

A linear model for symmetric receptive fields: implications for classification tests with flashed and moving images

The purpose of this study was to explore the effects of spatial and temporal properties on the expected responses of visual neurons that have linear receptive fields (RFs), particularly those having a mirror symmetric distribution of spatial subregions. Receptive fields that are symmetric in at least one spatial dimension occur in neurons of the retina, the lateral geniculate nucleus (LGN), and the visual cortex of mammals. Responses to flashing bars, moving bars, and moving edges were studied for different configurations of an analog RF model in which spatial and temporal aspects were varied independently. Responses of the model at intermediate stimulus speeds were found to agree with responses in the literature for X and Y units of the LGN and often for simple units of the visual cortex. In particular, having separated regions of response to light and dark edges, an identifying property of simple cells, was found to be a linear consequence of RF regions responding inversely to stimuli of opposite polarity. Model differences from responses of cortical complex units show that a linear model cannot mimic their responses, and imply that complex units employ major nonlinearities in coding image polarity (light vs dark), which signifies a nonlinearity in coding intensity. Because sudden flux changes inherent in flashing bars test mainly temporal RF properties, and slowly moving edges test mainly spatial properties, these two tests form a useful minimal set with which to describe and classify RFs. The usefulness of this set derives both from its sensitivity to spatial and temporal variables, and from the correlation between the linearity of a cell's processing of stimulus intensity and its RF classification.     

Lateral spike conduction velocity in the visual cortex affects spatial range of synchronization and receptive field size without visual experience: a learning model with spiking neurons

Classical receptive fields (cRF) increase in size from the retina to higher visual centers. The present work shows how temporal properties, in particular lateral spike velocity and spike input correlation, can affect cRF size and position without visual experience. We demonstrate how these properties are related to the spatial range of cortical synchronization if Hebbian learning dominates early development. For this, a largely reduced model of two successive levels of the visual cortex is developed (e.g., areas V1 and V2). It consists of retinotopic networks of spiking neurons with constant spike velocity in lateral connections. Feedforward connections between level 1 and 2 are additive and determine cRF size and shape, while lateral connections within level 1 are modulatory and affect the cortical range of synchronization. Input during development is mimicked by spike trains with spatially homogeneous properties and a confined temporal correlation width. During learning, the homogeneous lateral coupling shrinks to limited coupling structures defining synchronization and related association fields (AF). The size of level-1 synchronization fields determines the lateral coupling range of developing level-1-to-2 connections and, thus, the size of level-2 cRFs, even if the feedforward connections have distance-independent delays. AFs and cRFs increase with spike velocity in the lateral network and temporal correlation width of the input. Our results suggest that AF size of V1 and cRF size of V2 neurons are confined during learning by the temporal width of input correlations and the spike velocity in lateral connections without the need of visual experience. During learning from visual experience, a similar influence of AF size on the cRF size may be operative at successive levels of processing, including other parts of the visual system.     

Dimensional reduction of a V1 ring model with simple and complex cells

In this paper, we extend a framework for constructing low-dimensional dynamical systems models of mammalian primary visual cortex to a cortical network model that incorporates the full nonlinear effects of complex cells. The procedure consists of capturing the essential dynamics in a low-dimensional subspace using empirical methods, then recasting the equations in the reduced vector space. Previously, we considered visual cortical network models consisting of only simple cells with nearly linear responses to external stimuli. Here we show that fully nonlinear effects can be incorporated by examining the dimensional reduction of an idealized ring model of V1 with both simple and complex cells. We found it expedient to divide the subspace into four separate neuronal populations: excitatory simple, excitatory complex, inhibitory simple and inhibitory complex. In order to reproduce the fluctuation-driven dynamics in this reduced space, we incorporated (1) white noises with different intensities into individual neuronal populations, and (2) firing rate estimates to capture the probability of firing due to subthreshold fluctuations. With a more accurate, fitted connectivity, our modified dimensional reduced models can reproduce the firing rates, circular variances and modulation ratios observed in the original ring model.     

A model for the polarization of neurons by extrinsically applied electric fields.

A model is presented for the subthreshold polarization of a neuron by an applied electric field. It gives insight into how morphological features of a neuron affect its polarizability. The neuronal model consists of one or more extensively branched dendritic trees, a lumped somatic impedance, and a myelinated axon with nodes of Ranvier. The dendritic trees branch according to the 3/2-power rule of Rall, so that each tree has an equivalent cylinder representation. Equations for the membrane potential at the soma and at the nodes of Ranvier, given an arbitrary specified external potential, are derived. The solutions determine the contributions made by the dendritic tree and the axon to the net polarization at the soma. In the case of a spatially constant electric field, both the magnitude and sign of the polarization depend on simple combinations of parameters describing the neuron. One important combination is given by the ratio of internal resistances for longitudinal current spread along the dendritic tree trunk and along the axon. A second is given by the ratio between the DC space constant for the dendritic tree trunk and the distance between nodes of Ranvier in the axon. A third is given by the product of the electric field and the space constant for the trunk of the dendritic tree. When a neuron with a straight axon is subjected to a constant field, the membrane potential decays exponentially with distance from the soma. Thus, the soma seems to be a likely site for action potential initiation when the field is strong enough to elicit suprathreshold polarization. In a simple example, the way in which orientation of the various parts of the neuron affects its polarization is examined. When an axon with a bend is subjected to a spatially constant field, polarization is focused at the bend, and this is another likely site for action potential initiation.     

A cooperation and competition based simple cell receptive field model and study of feed-forward linear and nonlinear contributions to orientation selectivity

We present a model for development of orientation selectivity in layer IV simple cells. Receptive field (RF) development in the model, is determined by diffusive cooperation and resource limited competition guided axonal growth and retraction in geniculocortical pathway. The simulated cortical RFs resemble experimental RFs. The receptive field model is incorporated in a three-layer visual pathway model consisting of retina, LGN and cortex. We have studied the effect of activity dependent synaptic scaling on orientation tuning of cortical cells. The mean value of hwhh (half width at half the height of maximum response) in simulated cortical cells is 58° when we consider only the linear excitatory contribution from LGN. We observe a mean improvement of 22.8° in tuning response due to the non-linear spiking mechanisms that include effects of threshold voltage and synaptic scaling factor.     

A simple kinetic model for myeloma cell culture with consideration of lysine limitation

A simple kinetic model is developed to describe the dynamic behavior of myeloma cell growth and cell metabolism. Glucose, glutamine as well as lysine are considered as growth limiting substrates. The cell growth was restricted as soon as the extracellular lysine is exhausted and then intracellular lysine becomes a growth limiting substrate. In addition, a metabolic regulator model together with the Monod model is used to deal with the growth lag phase after inoculation or feeding. By using these models, concentrations of substrates and metabolites, as well as densities of viable and dead cells are quantitatively described. One batch cultivation and two fed-batch cultivations with pulse feeding of nutrients are used to validate the model.     

Electronic structure of polyenes related to the visual chromophore. A simple model for the observed band shapes.

A model is presented which attributes the widths of absorption bands in carotenoids to conformational disorder induced by the beta-ionylidene moiety. With reasonable parameter choices, the model gives a good quantitative fit to the spectra observed for four carotenoids and at the same time accounts for the lack of structure in the long-wavelength absorption of retinal.     

The involvement of gamma-aminobutyric acid in the organization of cat retinal ganglion cell receptive fields. A study with picrotoxin and bicuculline

The effects of picrotoxin and bicuculline upon the discharge pattern of center-surround organized cat retinal ganglion cells of X and Y type were studied. All experiments were carried out under scotopic or possibly low mesopic conditions; mostly but not exclusively on-center cells were studied. Stimuli were chosen so that responses were either; (a) "purely" central; (b) surround dominated; or (c) clearly mixed but center dominated. In each case a pre-drug control response was estaboished, the drug was administered intravenously, and its subsequent effect upon the response was observed. In Y cells both picrotoxin and bicucullin caused the center-driven component of the response to become somewhat reduced in magnitude, while the surround component was substantially reduced. There was thus a change in center- surround balance in favor of the center-driven component. Responses of X cells remained virtually unaffected by both picrotoxin and bicuculline.     

Telomere exchange and asymmetric segregation of chromosomes can account for the unlimited proliferative potential of ALT cell populations

Telomerase-negative cancer cells show increased telomere sister chromatid exchange (T-SCE) rates, a phenomenon that has been associated with an alternative lengthening of telomeres (ALT) mechanism for maintaining telomeres in this subset of cancers. Here we examine whether or not T-SCE can maintain telomeres in human cells using a combinatorial model capable of describing how telomere lengths evolve over time. Our results show that random T-SCE is unlikely to be the mechanism of telomere maintenance of ALT human cells, but that increased T-SCE rates combined with a recently proposed novel mechanism of non-random segregation of chromosomes with long telomeres preferentially into the same daughter cell during cell division can stabilize chromosome ends in ALT cancers. At the end we discuss a possible experiment that can validate the findings of this study.     

A calibrated EMG-informed neuromusculoskeletal model can appropriately account for muscle co-contraction in the estimation of hip joint contact forces in people with hip osteoarthritis

Abnormal hip joint contact forces (HJCF) are considered a primary mechanical contributor to the progression of hip osteoarthritis (OA). Compared to healthy controls, people with hip OA often present with altered muscle activation patterns and greater muscle co-contraction, both of which can influence HJCF. Neuromusculoskeletal (NMS) modelling is non-invasive approach to estimating HJCF, whereby different neural control solutions can be used to estimate muscle forces. Static optimisation, available within the popular NMS modelling software OpenSim, is a commonly used neural control solution, but may not account for an individual’s unique muscle activation patterns and/or co-contraction that are often evident in pathological population. Alternatively, electromyography (EMG)-assisted neural control solutions, available within CEINMS software, have been shown to account for individual activation patterns in healthy people. Nonetheless, their application in people with hip OA, with conceivably greater levels of co-contraction, is yet to be explored. The aim of this study was to compare HJCF estimations using static optimisation (in OpenSim) and EMG-assisted (in CEINMS) neural control solutions during walking in people with hip OA. EMG-assisted neural control solution was more consistent with both EMG and joint moment data than static optimisation, and also predicted significantly higher HJCF peaks (p < 0.001). The EMG-assisted neural control solution also accounted for more muscle co-contraction than static optimisation (p = 0.03), which probably contributed to these higher HJCF peaks. Findings suggest that the EMG-assisted neural control solution may estimate more physiologically plausible HJCF than static optimisation in a population with high levels of co-contraction, such as hip OA.     

Conditioned feedback and inhibitory reorganization of the receptive fields of the cortical neurons as a basis for the subconscious change in the thresholds of visual recognition and detection

In the paper three groups of facts are compared: significant adaptative and adaptational modification of receptive fields of cat's visual cortex neurones, conditioned selective subsensory change of the threshold of perception (detection and recognition) of a letter by man in relation to two control ones and the role of spatially specialized cortical inhibition in formation and adaptative modifications of receptive fields and detector properties of the visual cortex neurones. Interconnection is discussed of the phenomena described as well as community of their mechanisms.