The activity of a border ownership selective (BOS) neuron indicates where a foreground object is located relative to its (classical) receptive field (RF). A population of BOS neurons thus provides an important component of perceptual grouping, the organization of the visual scene into objects. In previous theoretical work, it has been suggested that this grouping mechanism is implemented by a population of dedicated grouping ("G") cells that integrate the activity of the distributed feature cells representing an object and, by feedback, modulate the same cells, thus making them border ownership selective. The feedback modulation by G cells is thought to also provide the mechanism for object-based attention. A recent modeling study showed that modulatory common feedback, implemented by synapses with N-methyl-D-aspartate (NMDA)-type glutamate receptors, accounts for the experimentally observed synchrony in spike trains of BOS neurons and the shape of cross-correlations between them, including its dependence on the attentional state. However, that study was limited to pairs of BOS neurons with consistent border ownership preferences, defined as two neurons tuned to respond to the same visual object, in which attention decreases synchrony. But attention has also been shown to increase synchrony in neurons with inconsistent border ownership selectivity. Here we extend the computational model from the previous study to fully understand these effects of attention. We postulate the existence of a second type of G-cell that represents spatial attention by modulating the activity of all BOS cells in a spatially defined area. Simulations of this model show that a combination of spatial and object-based mechanisms fully accounts for the observed pattern of synchrony between BOS neurons. Our results suggest that modulatory feedback from G-cells may underlie both spatial and object-based attention.

Autonomous robots can rely on attention mechanisms to explore complex scenes and select salient stimuli relevant for behaviour. Stimulus selection should be fast to efficiently allocate available (and limited) computational resources to process in detail a subset of the otherwise overwhelmingly large sensory input. The amount of processing required is a product of the amount of data sampled by a robot's sensors; while a standard RGB camera produces a fixed amount of data for every pixel of the sensor, an event-camera produces data only for where there is a contrast change in the field of view, and does so with a lower latency. In this paper, we describe the implementation of a state-of-the-art bottom-up attention model, based on structuring the visual scene in terms of proto-objects. As an event-camera encodes different visual information compared to frame-based cameras, the original algorithm must be adapted and modified. We find that the event-camera's inherent detection of edges removes the need for some early stages of processing in the model. We describe the modifications, compare the event-driven algorithm to the original, and validate the potential for use on the iCub humanoid robot.

We will demonstrate a real-time implementation of a protoobject based neuromorphic visual saliency model [1] on an embedded processing board. Visual saliency models are difficult to implement in hardware for real-time applications due to their computational complexity. The conventional implementation is not optimal because of the requirement of a large number of convolution operations for filtering on several feature channels across multiple image pyramids. Our current implementation considers the dynamic temporal motion change by convoluting along time efficiently by parallelly processing them. We have implemented the model on an NVIDIA Jetson TX1 board (Fig. 1), which has NVIDIA Maxwell GPU with 256 NVIDIA CUDA Cores, hosted on an Ubuntu environment. The board has a 5 MP fixed focus MIPI CSI camera through which the frames are fetched using a Quad-core ARM Cortex-A57 MPCore Processor with 4 GB LPDDR4 Memory. The camera module fetches the frames to the application for processing, and the result is then displayed through the HDMI port. The application is written in Tensorflow, Cuda, and Python and uses several Python libraries. For further analysis, the user can also save the output onto a file.

A key question in decision-making is how people integrate amounts and probabilities to form preferences between risky alternatives. Here we rely on the general principle of integration-to-boundary to develop several biologically plausible process models of risky-choice, which account for both choices and response-times. These models allowed us to contrast two influential competing theories: i) within-alternative evaluations, based on multiplicative interaction between amounts and probabilities, ii) within-attribute comparisons across alternatives. To constrain the preference formation process, we monitored eye-fixations during decisions between pairs of simple lotteries, designed to systematically span the decision-space. The behavioral results indicate that the participants' eye-scanning patterns were associated with risk-preferences and expected-value maximization. Crucially, model comparisons showed that within-alternative process models decisively outperformed within-attribute ones, in accounting for choices and response-times. These findings elucidate the psychological processes underlying preference formation when making risky-choices, and suggest that compensatory, within-alternative integration is an adaptive mechanism employed in human decision-making.

Visual saliency models are difficult to implement in hardware for real time applications due to their computational complexity. The conventional digital implementation is not optimal because of the requirement of a large number of convolution operations for filtering on several feature channels across multiple image pyramids [1], [2]. Here, we propose an alternative approach to implement a neuromorphic visual saliency algorithm [3] in digital hardware using stochastic computation, which can achieve very low power and small area. We show the real time implementation of important building blocks of the system and compare the overall system with its software implementation. Our implementation will be useful for facilitating high-fidelity selective rendering in computer graphics applications using the output of the saliency model, and for communications, where the non-salient parts of an image can be compressed more heavily than the salient parts. Our implementation will find several applications as a frontend co-processor for information triaging, compression and analysis in computer vision tasks. Our proposed SC-based convolution circuit could be a potential building block for implanting deep convolutional neural networks (CNN) on hardware.

The standard architecture of neocortex is a network with excitation and inhibition in closely maintained balance. These networks respond fast and with high precision to their inputs and they allow selective amplification of patterned signals. The stability of such networks is known to depend on balancing the strengths of positive and negative feedback. We here show that a second condition is required for stability which depends on the relative strengths and time courses of fast (AMPA) and slow (NMDA) currents in the excitatory projections. This condition also determines the response time of the network. We show that networks which respond quickly to an input are necessarily close to an oscillatory instability which resonates in the delta range. This instability explains the existence of neocortical delta oscillations and the emergence of absence epilepsy. Although cortical delta oscillations are a network-level phenomenon, we show that in non-pathological networks, individual neurons receive sufficient information to keep the network in the fast-response regime without sliding into the instability.

Making technological advances in the field of human-machine interactions requires that the capabilities and limitations of the human perceptual system be taken into account. The focus of this report is an important mechanism of perception, visually selective attention, which is becoming more and more important for multimedia applications. We introduce the concept of visual attention and describe its underlying mechanisms. In particular, we introduce the concepts of overt and covert visual attention, and of bottom-up and top-down processing. Challenges related to modeling visual attention and their validation using ad hoc ground truth are also discussed. Examples of the usage of visual attention models in image and video processing are presented. We emphasize multimedia delivery, retargeting and quality assessment of image and video, medical imaging, and the field of stereoscopic 3-D image applications.

An accurate calculation of the first passage time probability density (FPTPD) is essential for computing the likelihood of solutions of the stochastic leaky integrate-and-fire model. The previously proposed numerical calculation of the FPTPD based on the integral equation method discretizes the probability current of the voltage crossing the threshold. While the method is accurate for high noise levels, we show that it results in large numerical errors for small noise. The problem is solved by analytically computing, in each time bin, the mean probability current. Efficiency is further improved by identifying and ignoring time bins with negligible mean probability current.

When a neuronal spike train is observed, what can we deduce from it about the properties of the neuron that generated it? A natural way to answer this question is to make an assumption about the type of neuron, select an appropriate model for this type, and then choose the model parameters as those that are most likely to generate the observed spike train. This is the maximum likelihood method. If the neuron obeys simple integrate-and-fire dynamics, Paninski, Pillow, and Simoncelli (2004) showed that its negative log-likelihood function is convex and that, at least in principle, its unique global minimum can thus be found by gradient descent techniques. Many biological neurons are, however, known to generate a richer repertoire of spiking behaviors than can be explained in a simple integrate-and-fire model. For instance, such a model retains only an implicit (through spike-induced currents), not an explicit, memory of its input; an example of a physiological situation that cannot be explained is the absence of firing if the input current is increased very slowly. Therefore, we use an expanded model (Mihalas & Niebur, 2009 ), which is capable of generating a large number of complex firing patterns while still being linear. Linearity is important because it maintains the distribution of the random variables and still allows maximum likelihood methods to be used. In this study, we show that although convexity of the negative log-likelihood function is not guaranteed for this model, the minimum of this function yields a good estimate for the model parameters, in particular if the noise level is treated as a free parameter. Furthermore, we show that a nonlinear function minimization method (r-algorithm with space dilation) usually reaches the global minimum.

We present an electronic neuron that uses first-order log-domain low-pass filters to implement the Mihalas-Niebur model. The neuron consists of a leaky-integrate-and-fire core and building blocks to implement an adaptive threshold and spike induced currents. Simulation results show that this modular neuron can emulate different spiking behaviours observed in biological neurons.

Spiking neurons and spiking neural circuits are finding uses in a multitude of tasks such as robotic locomotion control, neuroprosthetics, visual sensory processing, and audition. The desired neural output is achieved through the use of complex neuron models, or by combining multiple simple neurons into a network. In either case, a means for configuring the neuron or neural circuit is required. Manual manipulation of parameters is both time consuming and non-intuitive due to the nonlinear relationship between parameters and the neuron's output. The complexity rises even further as the neurons are networked and the systems often become mathematically intractable. In large circuits, the desired behavior and timing of action potential trains may be known but the timing of the individual action potentials is unknown and unimportant, whereas in single neuron systems the timing of individual action potentials is critical. In this paper, we automate the process of finding parameters. To configure a single neuron we derive a maximum likelihood method for configuring a neuron model, specifically the Mihalas-Niebur Neuron. Similarly, to configure neural circuits, we show how we use genetic algorithms (GAs) to configure parameters for a network of simple integrate and fire with adaptation neurons. The GA approach is demonstrated both in software simulation and hardware implementation on a reconfigurable custom very large scale integration chip.

In this paper we present the circuits and simulation results for a silicon neuron which is based on a modified version of the Mihalas-Niebur neural model [1]. This silicon neuron produces 15 of the 20 known neural spiking and bursting behaviors. It has low complexity and reliable matching and can thus be easily integrated into more complex neuromorphic systems. Implemented in a 0.15um 1.5V CMOS process, each neuron consumes about 7.5nW of power at 1kHz and occupies an area of 70um by 70um.

For simulations of neural networks, there is a trade-off between the size of the network that can be simulated and the complexity of the model used for individual neurons. In this study, we describe a generalization of the leaky integrate-and-fire model that produces a wide variety of spiking behaviors while still being analytically solvable between firings. For different parameter values, the model produces spiking or bursting, tonic, phasic or adapting responses, depolarizing or hyperpolarizing after potentials and so forth. The model consists of a diagonalizable set of linear differential equations describing the time evolution of membrane potential, a variable threshold, and an arbitrary number of firing-induced currents. Each of these variables is modified by an update rule when the potential reaches threshold. The variables used are intuitive and have biological significance. The model's rich behavior does not come from the differential equations, which are linear, but rather from complex update rules. This single-neuron model can be implemented using algorithms similar to the standard integrate-and-fire model. It is a natural match with event-driven algorithms for which the firing times are obtained as a solution of a polynomial equation.

We provide analytical solutions for mean firing rates and cross-correlations of coincidence detector neurons in recurrent networks with excitatory or inhibitory connectivity, with rate-modulated steady-state spiking inputs. We use discrete-time finite-state Markov chains to represent network state transition probabilities, which are subsequently used to derive exact analytical solutions for mean firing rates and cross-correlations. As illustrated in several examples, the method can be used for modeling cortical microcircuits and clarifying single-neuron and population coding mechanisms. We also demonstrate that increasing firing rates do not necessarily translate into increasing cross-correlations, though our results do support the contention that firing rates and cross-correlations are likely to be coupled. Our analytical solutions underscore the complexity of the relationship between firing rates and cross-correlations.

Recent technological advances as well as progress in theoretical understanding of neural systems have created a need for synthetic spike trains with controlled mean rate and pairwise cross-correlation. This report introduces and analyzes a novel algorithm for the generation of discretized spike trains with arbitrary mean rates and controlled cross correlation. Pairs of spike trains with any pairwise correlation can be generated, and higher-order correlations are compatible with common synaptic input. Relations between allowable mean rates and correlations within a population are discussed. The algorithm is highly efficient, its complexity increasing linearly with the number of spike trains generated and therefore inversely with the number of cross-correlated pairs.

We provide an analytical recurrent solution for the firing rates and cross-correlations of feedforward networks with arbitrary connectivity, excitatory or inhibitory, in response to steady-state spiking input to all neurons in the first network layer. Connections can go between any two layers as long as no loops are produced. Mean firing rates and pairwise cross-correlations of all input neurons can be chosen individually. We apply this method to study the propagation of rate and synchrony information through sample networks to address the current debate regarding the efficacy of rate codes versus temporal codes. Our results from applying the network solution to several examples support the following conclusions: (1) differential propagation efficacy of rate and synchrony to higher layers of a feedforward network is dependent on both network and input parameters, and (2) previous modeling and simulation studies exclusively supporting either rate or temporal coding must be reconsidered within the limited range of network and input parameters used. Our exact, analytical solution for feedforward networks of coincidence detectors should prove useful for further elucidating the efficacy and differential roles of rate and temporal codes in terms of different network and input parameter ranges.

We present a solution for the steady-state output rate of an ideal coincidence detector receiving an arbitrary number of excitatory and inhibitory inputspike trains. All excitatory spike trains have identical binomial count distributions (which includes Poisson statistics as a special case) and arbitrary pairwise cross correlations between them. The same applies to the inhibitory inputs, and the rates and correlation functions of excitatory and inhibitory populations may be the same or different from each other. Thus, for each population independently, the correlation may range from complete independence to perfect correlation (identical processes). We find that inhibition, if made sufficiently strong, will result in an inverted U-shaped curve for the output rate of a coincidence detector as a function of input rates for the case of identical inhibitory and excitory input rates. This leads to the prediction that higher presynaptic (input) rates may lead to lower postsynaptic (output) rates where the output rate may fall faster than the inverse of the input rate, and shows some qualitative similarities to the case of purely excitatory inputs with synaptic depression. In general, we find that including inhibition invariably and significantly increases the behavioral repertoire of the coincidence detector over the case of pure excitatory input.

We derive analytically the solution for the output rate of the ideal coincidence detector. The solution is for an arbitrary number of input spike trains with identical binomial count distributions (which includes Poisson statistics as a special case) and identical arbitrary pairwise cross-correlations, from zero correlation (independent processes) to complete correlation (identical processes).

In this letter, we extend our previous analytical results (Mikula & Niebur, 2003) for the coincidence detector by taking into account probabilistic frequency-dependent synaptic depression. We present a solution for the steady-state output rate of an ideal coincidence detector receiving an arbitrary number of input spike trains with identical binomial count distributions (which includes Poisson statistics as a special case) and identical arbitrary pairwise cross-correlations, from zero correlation (independent processes) to perfect correlation (identical processes). Synapses vary their efficacy probabilistically according to the observed depression mechanisms. Our results show that synaptic depression, if made sufficiently strong, will result in an inverted U-shaped curve for the output rate of a coincidence detector as a function of input rate. This leads to the counterintuitive prediction that higher presynaptic (input) rates may lead to lower postsynaptic (output) rates where the output rate may fall faster than the inverse of the input rate.

Gaze-contingent variable resolution display techniques allocate computational resources for image generation preferentially to the area around the center of gaze where visual sensitivity to detail is the greatest. Although these techniques are computationally efficient, their behavioral consequences with realistic tasks and materials are not well understood. The behavior of human observers performing visual search of natural scenes using gaze-contingent variable resolution displays is examined. A two-region display was used where a high-resolution region was centered on the instantaneous center of gaze, and the surrounding region was presented in a lower resolution. The radius of the central high-resolution region was varied from 1 to 15 degrees while the total amount of computational resources required to generate the visual display was kept constant. Measures of reaction time, accuracy, and fixation duration suggest that task performance is comparable to that seen for uniform resolution displays when the central region size is approximately 5 degrees.

Synchronized neuronal firing has been reported in many neural systems and may play a role in the representation of sensory stimuli and the modification of sensory representations by both experience and attention. In this report we describe a bootstrap procedure for computing the statistical significance of changes in the degree of synchrony and apply it to recordings from the second somatosensory (SII) cortex of Macaques performing tactile and visual discrimination tasks. A majority (68%) of neuron pairs in SII fire synchronously in response to a tactile stimulus. In a fraction of those pairs (17.5%), the degree of synchrony covaries with the focus of attention.

Unitary event analysis is a new method for detecting episodes of synchronized neural activity (Riehle, Grün, Diesmann, & Aertsen, 1997). It detects time intervals that contain coincident firing at higher rates than would be expected if the neurons fired as independent inhomogeneous Poisson processes; all coincidences in such intervals are called unitary events (UEs). Changes in the frequency of UEs that are correlated with behavioral states may indicate synchronization of neural firing that mediates or represents the behavioral state. We show that UE analysis is subject to severe limitations due to the underlying discrete statistics of the number of coincident events. These limitations are particularly stringent for low (0-10 spikes/s) firing rates. Under these conditions, the frequency of UEs is a random variable with a large variation relative to its mean. The relative variation decreases with increasing firing rate, and we compute the lowest firing rate, at which the 95% confidence interval around the mean frequency of UEs excludes zero. This random variation in UE frequency makes interpretation of changes in UEs problematic for neurons with low firing rates. As a typical example, when analyzing 150 trials of an experiment using an averaging window 100 ms wide and a 5 ms coincidence window, firing rates should be greater than 7 spikes per second.

A visual attention system, inspired by the behavior and the neuronal architecture of the early primate visual system, is presented. Multiscale image features are combined into a single topographical saliency map. A dynamical neural network then selects attended locations in order of decreasing saliency. The system breaks down the complex problem of scene understanding by rapidly selecting, in a computationally efficient manner, conspicuous locations to be analyzed in detail.

Visual space is represented by cortical cells in an orderly manner. Only little variation in the cell behavior is found with changing depth below the cortical surface, that is, all cells in a column with axis perpendicular to the cortical plane have approximately the same properties (Hubel and Wiesel 1962, 1963, 1968). Therefore, the multiple features of the visual space (e.g., position in visual space, preferred orientation, and orientation tuning strength) are mapped on a two-dimensional space, the cortical plane. Such a dimension reduction leads to complex maps (Durbin and Mitchison 1990) that so far have evaded an intuitive understanding. Analyzing optical imaging data (Blasdel 1992a, b; Blasdel and Salama 1986; Grinvald et al. 1986) using a theoretical approach we will show that the most salient features of these maps can be understood from a few basic design principles: local correlation, modularity, isotropy, and homogeneity. These principles can be defined in a mathematically exact sense in the Fourier domain by a rather simple annulus-like spectral structure. Many of the models that have been developed to explain the mapping of the preferred orientations (Cooper et al. 1979; Legendy 1978; Linsker 1986a, b; Miller 1992; Nass and Cooper 1975; Obermayer et al. 1990, 1992; Soodak 1987; Swindale 1982, 1985, 1992; von der Malsburg 1973; von der Malsburg and Cowan 1982) are quite successful in generating maps that are close to experimental maps. We suggest that this success is due to these principles, which are common properties of the models and of biological maps.

We study the dynamics of completely connected populations of refractory integrate-and-fire neurons in the presence of noise. Solving the master equation based on a mean-field approach, and by computer simulations, we find sustained states of activity that correspond to fixed points and show that for the same value of external input, the system has one or two attractors. The dynamic behavior of the population under the influence of external input and noise manifests hysteresis effects that might have a functional role for memory. The temporal dynamics at higher temporal resolution, finer than the transmission delay times and the refractory period, are characterized by synchronized activity of subpopulations. The global activity of the population shows aperiodic oscillations analogous to experimentally found field potentials.

To what extent do the mechanisms generating different receptive field properties of neurons depend on each other? We investigated this question theoretically within the context of orientation and direction tuning of simple cells in the mammalian visual cortex. In our model a cortical cell of the "simple" type receives its orientation tuning by afferent convergence of aligned receptive fields of the lateral geniculate nucleus (Hubel and Wiesel 1962). We sharpen this orientation bias by postulating a special type of radially symmetric long-range lateral inhibition called circular inhibition. Surprisingly, this isotropic mechanism leads to the emergence of a strong bias for the direction of motion of a bar. We show that this directional anisotropy is neither caused by the probabilistic nature of the connections nor is it a consequence of the specific columnar structure chosen but that it is an inherent feature of the architecture of visual cortex.

Horizontal long-range interactions are strongly involved in the generation of the receptive fields of visual cortical cells. Structurally imposed limitations of long-range interactions are demonstrated. In particular, it is shown that the cross-orientation inhibition scheme leads to inhomogeneous input for different cell populations which is experimentally not observed. This is not the case for circular inhibition, a new connection scheme proposed for long-range interaction. This is shown by computer simulation of the early visual system of the cat and by a simpler but analytically solvable model. The results are confirmed by applying the methods to the experimentally determined structure of the orientational hypercolumns in area 18 of the cat