Computing with continuous attractors: stability and online aspects

Two issues concerning the application of continuous attractors in neural systems are investigated: the computational robustness of continuous attractors with respect to input noises and the implementation of Bayesian online decoding. In a perfect mathematical model for continuous attractors, decoding results for stimuli are highly sensitive to input noises, and this sensitivity is the inevitable consequence of the system's neutral stability. To overcome this shortcoming, we modify the conventional network model by including extra dynamical interactions between neurons. These interactions vary according to the biologically plausible Hebbian learning rule and have the computational role of memorizing and propagating stimulus information accumulated with time. As a result, the new network model responds to the history of external inputs over a period of time, and hence becomes insensitive to short-term fluctuations. Also, since dynamical interactions provide a mechanism to convey the prior knowledge of stimulus, that is, the information of the stimulus presented previously, the network effectively implements online Bayesian inference. This study also reveals some interesting behavior in neural population coding, such as the trade-off between decoding stability and the speed of tracking time-varying stimuli, and the relationship between neural tuning width and the tracking speed.     

Fisher and Shannon information in finite neural populations

The precision of the neural code is commonly investigated using two families of statistical measures: Shannon mutual information and derived quantities when investigating very small populations of neurons and Fisher information when studying large populations. These statistical tools are no longer the preserve of theorists and are being applied by experimental research groups in the analysis of empirical data. Although the relationship between information-theoretic and Fisher-based measures in the limit of infinite populations is relatively well understood, how these measures compare in finite-size populations has not yet been systematically explored. We aim to close this gap. We are particularly interested in understanding which stimuli are best encoded by a given neuron within a population and how this depends on the chosen measure. We use a novel Monte Carlo approach to compute a stimulus-specific decomposition of the mutual information (the SSI) for populations of up to 256 neurons and show that Fisher information can be used to accurately estimate both mutual information and SSI for populations of the order of 100 neurons, even in the presence of biologically realistic variability, noise correlations, and experimentally relevant integration times. According to both measures, the stimuli that are best encoded are those falling at the flanks of the neuron's tuning curve. In populations of fewer than around 50 neurons, however, Fisher information can be misleading.     

Difficulty of singularity in population coding

Fisher information has been used to analyze the accuracy of neural population coding. This works well when the Fisher information does not degenerate, but when two stimuli are presented to a population of neurons, a singular structure emerges by their mutual interactions. In this case, the Fisher information matrix degenerates, and the regularity condition ensuring the Cramér-Rao paradigm of statistics is violated. An animal shows pathological behavior in such a situation. We present a novel method of statistical analysis to understand information in population coding in which algebraic singularity plays a major role. The method elucidates the nature of the pathological case by calculating the Fisher information. We then suggest that synchronous firing can resolve singularity and show a method of analyzing the binding problem in terms of the Fisher information. Our method integrates a variety of disciplines in population coding, such as nonregular statistics, Bayesian statistics, singularity in algebraic geometry, and synchronous firing, under the theme of Fisher information.     

Multidimensional encoding strategy of spiking neurons

Neural responses in sensory systems are typically triggered by a multitude of stimulus features. Using information theory, we study the encoding accuracy of a population of stochastically spiking neurons characterized by different tuning widths for the different features. The optimal encoding strategy for representing one feature most accurately consists of narrow tuning in the dimension to be encoded, to increase the single-neuron Fisher information, and broad tuning in all other dimensions, to increase the number of active neurons. Extremely narrow tuning without sufficient receptive field overlap will severely worsen the coding. This implies the existence of an optimal tuning width for the feature to be encoded. Empirically, only a subset of all stimulus features will normally be accessible. In this case, relative encoding errors can be calculated that yield a criterion for the function of a neural population based on the measured tuning curves.     

Representational accuracy of stochastic neural populations.

Fisher information is used to analyze the accuracy with which a neural population encodes D stimulus features. It turns out that the form of response variability has a major impact on the encoding capacity and therefore plays an important role in the selection of an appropriate neural model. In particular, in the presence of baseline firing, the reconstruction error rapidly increases with D in the case of Poissonian noise but not for additive noise. The existence of limited-range correlations of the type found in cortical tissue yields a saturation of the Fisher information content as a function of the population size only for an additive noise model. We also show that random variability in the correlation coefficient within a neural population, as found empirically, considerably improves the average encoding quality. Finally, the representational accuracy of populations with inhomogeneous tuning properties, either with variability in the tuning widths or fragmented into specialized subpopulations, is superior to the case of identical and radially symmetric tuning curves usually considered in the literature.     

Population coding with correlation and an unfaithful model

This study investigates a population decoding paradigm in which the maximum likelihood inference is based on an unfaithful decoding model (UMLI). This is usually the case for neural population decoding because the encoding process of the brain is not exactly known or because a simplified decoding model is preferred for saving computational cost. We consider an unfaithful decoding model that neglects the pair-wise correlation between neuronal activities and prove that UMLI is asymptotically efficient when the neuronal correlation is uniform or of limited range. The performance of UMLI is compared with that of the maximum likelihood inference based on the faithful model and that of the center-of-mass decoding method. It turns out that UMLI has advantages of decreasing the computational complexity remarkably and maintaining high-level decoding accuracy. Moreover, it can be implemented by a biologically feasible recurrent network (Pouget, Zhang, Deneve, & Latham, 1998). The effect of correlation on the decoding accuracy is also discussed.     

Decoding a temporal population code

Encoding of sensory events in internal states of the brain requires that this information can be decoded by other neural structures. The encoding of sensory events can involve both the spatial organization of neuronal activity and its temporal dynamics. Here we investigate the issue of decoding in the context of a recently proposed encoding scheme: the temporal population code. In this code, the geometric properties of visual stimuli become encoded into the temporal response characteristics of the summed activities of a population of cortical neurons. For its decoding, we evaluate a model based on the structure and dynamics of cortical microcircuits that is proposed for computations on continuous temporal streams: the liquid state machine. Employing the original proposal of the decoding network results in a moderate performance. Our analysis shows that the temporal mixing of subsequent stimuli results in a joint representation that compromises their classification. To overcome this problem, we investigate a number of initialization strategies. Whereas we observe that a deterministically initialized network results in the best performance, we find that in case the network is never reset, that is, it continuously processes the sequence of stimuli, the classification performance is greatly hampered by the mixing of information from past and present stimuli. We conclude that this problem of the mixing of temporally segregated information is not specific to this particular decoding model but relates to a general problem that any circuit that processes continuous streams of temporal information needs to solve. Furthermore, as both the encoding and decoding components of our network have been independently proposed as models of the cerebral cortex, our results suggest that the brain could solve the problem of temporal mixing by applying reset signals at stimulus onset, leading to a temporal segmentation of a continuous input stream.     

Insights from a simple expression for linear fisher information in a recurrently connected population of spiking neurons

A simple expression for a lower bound of Fisher information is derived for a network of recurrently connected spiking neurons that have been driven to a noise-perturbed steady state. We call this lower bound linear Fisher information, as it corresponds to the Fisher information that can be recovered by a locally optimal linear estimator. Unlike recent similar calculations, the approach used here includes the effects of nonlinear gain functions and correlated input noise and yields a surprisingly simple and intuitive expression that offers substantial insight into the sources of information degradation across successive layers of a neural network. Here, this expression is used to (1) compute the optimal (i.e., information-maximizing) firing rate of a neuron, (2) demonstrate why sharpening tuning curves by either thresholding or the action of recurrent connectivity is generally a bad idea, (3) show how a single cortical expansion is sufficient to instantiate a redundant population code that can propagate across multiple cortical layers with minimal information loss, and (4) show that optimal recurrent connectivity strongly depends on the covariance structure of the inputs to the network.     

Dynamics and computation of continuous attractors

Continuous attractor is a promising model for describing the encoding of continuous stimuli in neural systems. In a continuous attractor, the stationary states of the neural system form a continuous parameter space, on which the system is neutrally stable. This property enables the neutral system to track time-varying stimuli smoothly, but it also degrades the accuracy of information retrieval, since these stationary states are easily disturbed by external noise. In this work, based on a simple model, we systematically investigate the dynamics and the computational properties of continuous attractors. In order to analyze the dynamics of a large-size network, which is otherwise extremely complicated, we develop a strategy to reduce its dimensionality by utilizing the fact that a continuous attractor can eliminate the noise components perpendicular to the attractor space very quickly. We therefore project the network dynamics onto the tangent of the attractor space and simplify it successfully as a one-dimensional Ornstein-Uhlenbeck process. Based on this simplified model, we investigate (1) the decoding error of a continuous attractor under the driving of external noisy inputs, (2) the tracking speed of a continuous attractor when external stimulus experiences abrupt changes, (3) the neural correlation structure associated with the specific dynamics of a continuous attractor, and (4) the consequence of asymmetric neural correlation on statistical population decoding. The potential implications of these results on our understanding of neural information processing are also discussed.     

Dynamic analyses of information encoding in neural ensembles

Neural spike train decoding algorithms and techniques to compute Shannon mutual information are important methods for analyzing how neural systems represent biological signals. Decoding algorithms are also one of several strategies being used to design controls for brain-machine interfaces. Developing optimal strategies to design decoding algorithms and compute mutual information are therefore important problems in computational neuroscience. We present a general recursive filter decoding algorithm based on a point process model of individual neuron spiking activity and a linear stochastic state-space model of the biological signal. We derive from the algorithm new instantaneous estimates of the entropy, entropy rate, and the mutual information between the signal and the ensemble spiking activity. We assess the accuracy of the algorithm by computing, along with the decoding error, the true coverage probability of the approximate 0.95 confidence regions for the individual signal estimates. We illustrate the new algorithm by reanalyzing the position and ensemble neural spiking activity of CA1 hippocampal neurons from two rats foraging in an open circular environment. We compare the performance of this algorithm with a linear filter constructed by the widely used reverse correlation method. The median decoding error for Animal 1 (2) during 10 minutes of open foraging was 5.9 (5.5) cm, the median entropy was 6.9 (7.0) bits, the median information was 9.4 (9.4) bits, and the true coverage probability for 0.95 confidence regions was 0.67 (0.75) using 34 (32) neurons. These findings improve significantly on our previous results and suggest an integrated approach to dynamically reading neural codes, measuring their properties, and quantifying the accuracy with which encoded information is extracted.     

Model-based decoding, information estimation, and change-point detection techniques for multineuron spike trains

One of the central problems in systems neuroscience is to understand how neural spike trains convey sensory information. Decoding methods, which provide an explicit means for reading out the information contained in neural spike responses, offer a powerful set of tools for studying the neural coding problem. Here we develop several decoding methods based on point-process neural encoding models, or forward models that predict spike responses to stimuli. These models have concave log-likelihood functions, which allow efficient maximum-likelihood model fitting and stimulus decoding. We present several applications of the encoding model framework to the problem of decoding stimulus information from population spike responses: (1) a tractable algorithm for computing the maximum a posteriori (MAP) estimate of the stimulus, the most probable stimulus to have generated an observed single- or multiple-neuron spike train response, given some prior distribution over the stimulus; (2) a gaussian approximation to the posterior stimulus distribution that can be used to quantify the fidelity with which various stimulus features are encoded; (3) an efficient method for estimating the mutual information between the stimulus and the spike trains emitted by a neural population; and (4) a framework for the detection of change-point times (the time at which the stimulus undergoes a change in mean or variance) by marginalizing over the posterior stimulus distribution. We provide several examples illustrating the performance of these estimators with simulated and real neural data.     

Optimal short-term population coding: when Fisher information fails

Efficient coding has been proposed as a first principle explaining neuronal response properties in the central nervous system. The shape of optimal codes, however, strongly depends on the natural limitations of the particular physical system. Here we investigate how optimal neuronal encoding strategies are influenced by the finite number of neurons N (place constraint), the limited decoding time window length T (time constraint), the maximum neuronal firing rate f(max) (power constraint), and the maximal average rate (f)(max) (energy constraint). While Fisher information provides a general lower bound for the mean squared error of unbiased signal reconstruction, its use to characterize the coding precision is limited. Analyzing simple examples, we illustrate some typical pitfalls and thereby show that Fisher information provides a valid measure for the precision of a code only if the dynamic range (f(min)T, f(max)T) is sufficiently large. In particular, we demonstrate that the optimal width of gaussian tuning curves depends on the available decoding time T. Within the broader class of unimodal tuning functions, it turns out that the shape of a Fisher-optimal coding scheme is not unique. We solve this ambiguity by taking the minimum mean square error into account, which leads to flat tuning curves. The tuning width, however, remains to be determined by energy constraints rather than by the principle of efficient coding.     

Population coding with motion energy filters: the impact of correlations

The codes obtained from the responses of large populations of neurons are known as population codes. Several studies have shown that the amount of information conveyed by such codes, and the format of this information, is highly dependent on the pattern of correlations. However, very little is known about the impact of response correlations (as found in actual cortical circuits) on neural coding. To address this problem, we investigated the properties of population codes obtained from motion energy filters, which provide one of the best models for motion selectivity in early visual areas. It is therefore likely that the correlations that arise among energy filters also arise among motion-selective neurons. We adopted an ideal observer approach to analyze filter responses to three sets of images: noisy sine gratings, random dots kinematograms, and images of natural scenes. We report that in our model, the structure of the population code varies with the type of image. We also show that for all sets of images, correlations convey a large fraction of the information: 40% to 90% of the total information. Moreover, ignoring those correlations when decoding leads to considerable information loss-from 50% to 93%, depending on the image type. Finally we show that it is important to consider a large population of motion energy filters in order to see the impact of correlations. Study of pairs of neurons, as is often done experimentally, can underestimate the effect of correlations.     

Optimal tuning widths in population coding of periodic variables

We study the relationship between the accuracy of a large neuronal population in encoding periodic sensory stimuli and the width of the tuning curves of individual neurons in the population. By using general simple models of population activity, we show that when considering one or two periodic stimulus features, a narrow tuning width provides better population encoding accuracy. When encoding more than two periodic stimulus features, the information conveyed by the population is instead maximal for finite values of the tuning width. These optimal values are only weakly dependent on model parameters and are similar to the width of tuning to orientation or motion direction of real visual cortical neurons. A very large tuning width leads to poor encoding accuracy, whatever the number of stimulus features encoded. Thus, optimal coding of periodic stimuli is different from that of nonperiodic stimuli, which, as shown in previous studies, would require infinitely large tuning widths when coding more than two stimulus features.     

Transmission of population-coded information

As neural activity is transmitted through the nervous system, neuronal noise degrades the encoded information and limits performance. It is therefore important to know how information loss can be prevented. We study this question in the context of neural population codes. Using Fisher information, we show how information loss in a layered network depends on the connectivity between the layers. We introduce an algorithm, reminiscent of the water filling algorithm for Shannon information that minimizes the loss. The optimal connection profile has a center-surround structure with a spatial extent closely matching the neurons' tuning curves. In addition, we show how the optimal connectivity depends on the correlation structure of the trial-to-trial variability in the neuronal responses. Our results explain how optimal communication of population codes requires the center-surround architectures found in the nervous system and provide explicit predictions on the connectivity parameters.     

Efficient Markov chain Monte Carlo methods for decoding neural spike trains

Stimulus reconstruction or decoding methods provide an important tool for understanding how sensory and motor information is represented in neural activity. We discuss Bayesian decoding methods based on an encoding generalized linear model (GLM) that accurately describes how stimuli are transformed into the spike trains of a group of neurons. The form of the GLM likelihood ensures that the posterior distribution over the stimuli that caused an observed set of spike trains is log concave so long as the prior is. This allows the maximum a posteriori (MAP) stimulus estimate to be obtained using efficient optimization algorithms. Unfortunately, the MAP estimate can have a relatively large average error when the posterior is highly nongaussian. Here we compare several Markov chain Monte Carlo (MCMC) algorithms that allow for the calculation of general Bayesian estimators involving posterior expectations (conditional on model parameters). An efficient version of the hybrid Monte Carlo (HMC) algorithm was significantly superior to other MCMC methods for gaussian priors. When the prior distribution has sharp edges and corners, on the other hand, the "hit-and-run" algorithm performed better than other MCMC methods. Using these algorithms, we show that for this latter class of priors, the posterior mean estimate can have a considerably lower average error than MAP, whereas for gaussian priors, the two estimators have roughly equal efficiency. We also address the application of MCMC methods for extracting nonmarginal properties of the posterior distribution. For example, by using MCMC to calculate the mutual information between the stimulus and response, we verify the validity of a computationally efficient Laplace approximation to this quantity for gaussian priors in a wide range of model parameters; this makes direct model-based computation of the mutual information tractable even in the case of large observed neural populations, where methods based on binning the spike train fail. Finally, we consider the effect of uncertainty in the GLM parameters on the posterior estimators.     

Threshold behaviour of the maximum likelihood method in population decoding

We study the performance of the maximum likelihood (ML) method in population decoding as a function of the population size. Assuming uncorrelated noise in neural responses, the ML performance, quantified by the expected square difference between the estimated and the actual quantity, follows closely the optimal Cramer-Rao bound, provided that the population size is sufficiently large. However, when the population size decreases below a certain threshold, the performance of the ML method undergoes a rapid deterioration, experiencing a large deviation from the optimal bound. We explain the cause of such threshold behaviour, and present a phenomenological approach for estimating the threshold population size, which is found to be linearly proportional to the inverse of the square of the system's signal-to-noise ratio. If the ML method is used by neural systems, we expect the number of neurons involved in population coding to be above this threshold.     

Implications of neuronal diversity on population coding

In many cortical and subcortical areas, neurons are known to modulate their average firing rate in response to certain external stimulus features. It is widely believed that information about the stimulus features is coded by a weighted average of the neural responses. Recent theoretical studies have shown that the information capacity of such a coding scheme is very limited in the presence of the experimentally observed pairwise correlations. However, central to the analysis of these studies was the assumption of a homogeneous population of neurons. Experimental findings show a considerable measure of heterogeneity in the response properties of different neurons. In this study, we investigate the effect of neuronal heterogeneity on the information capacity of a correlated population of neurons. We show that information capacity of a heterogeneous network is not limited by the correlated noise, but scales linearly with the number of cells in the population. This information cannot be extracted by the population vector readout, whose accuracy is greatly suppressed by the correlated noise. On the other hand, we show that an optimal linear readout that takes into account the neuronal heterogeneity can extract most of this information. We study analytically the nature of the dependence of the optimal linear readout weights on the neuronal diversity. We show that simple online learning can generate readout weights with the appropriate dependence on the neuronal diversity, thereby yielding efficient readout.     

Precision constrained stochastic resonance in a feedforward neural network

Stochastic resonance (SR) is a phenomenon in which the response of a nonlinear system to a subthreshold information-bearing signal is optimized by the presence of noise. By considering a nonlinear system (network of leaky integrate-and-fire (LIF) neurons) that captures the functional dynamics of neuronal firing, we demonstrate that sensory neurons could, in principle harness SR to optimize the detection and transmission of weak stimuli. We have previously characterized this effect by use of signal-to-noise ratio (SNR). Here in addition to SNR, we apply an entropy-based measure (Fisher information) and compare the two measures of quantifying SR. We also discuss the performance of these two SR measures in a full precision floating point model simulated in Java and in a precision limited integer model simulated on a field programmable gate array (FPGA). We report in this study that stochastic resonance which is mainly associated with floating point implementations is possible in both a single LIF neuron and a network of LIF neurons implemented on lower resolution integer based digital hardware. We also report that such a network can improve the SNR and Fisher information of the output over a single LIF neuron.     

Neural information processing with feedback modulations

Descending feedback connections, together with ascending feedforward ones, are the indispensable parts of the sensory pathways in the central nervous system. This study investigates the potential roles of feedback interactions in neural information processing. We consider a two-layer continuous attractor neural network (CANN), in which neurons in the first layer receive feedback inputs from those in the second one. By utilizing the intrinsic property of a CANN, we use a projection method to reduce the dimensionality of the network dynamics significantly. The simplified dynamics allows us to elucidate the effects of feedback modulation analytically. We find that positive feedback enhances the stability of the network state, leading to an improved population decoding performance, whereas negative feedback increases the mobility of the network state, inducing spontaneously moving bumps. For strong, negative feedback interaction, the network response to a moving stimulus can lead the actual stimulus position, achieving an anticipative behavior. The biological implications of these findings are discussed. The simulation results agree well with our theoretical analysis.