Stefan D. Wilke

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.

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.

Neural spike statistics modify the impact of background noise

Abstract Neural populations in the neocortex typically encode multiple stimulus features, e.g., position, brightness, contrast, and orientation of a visual stimulus in the case of cells in area 17. Here, we perform a Fisher information analysis of the encoding accuracy of a neural population which is sensitive to D stimulus features. The neurons are assumed to exhibit a non-vanishing level of baseline activity. It is shown that the encoding accuracy decreases drastically with D if the spike count variance depends on the mean spike count, as is the case for Poissonian spike statistics. The need to reduce the susceptibility to background noise thus poses severe restrictions on the neural firing statistics or the number of encoded stimulus features. The results hold for uncorrelated as well as for correlated activity in the neural population.

On the functional role of noise correlations in the nervous system

Abstract Neurons are known to respond to external stimuli in a noisy manner. This response variability is often correlated across the neural population. However, the functional role of these noise correlations is still unclear. In this paper, Fisher information is used to analyze the effects of noise correlations on the accuracy with which a neural population encodes a stimulus. 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 can considerably improve the average encoding quality. This means that diversity in the correlations, as found empirically, could be exploited for improving stimulus representation.

Stimulus representation in rat primary visual cortex: multi-electrode recordings with micro-machined silicon probes and estimation theory

The study of neural population codes relies on massively parallel recordings in combination with theoretically motivated analysis tools. We applied two multi-site recording techniques to

Neural Coding of Dynamic Stimuli

An estimation-theoretical approach is used to characterize the accuracy by which a pair of neurons encodes a time-varying stimulus. Whether high firing rates yield a representational advantage depends on the type of noise involved in spike generation: High rates are favorable for Poissonian noise, but fail to yield improvements in the case of multiplicative Gaussian noise. Moreover, we find that the performance of single-phase stimulus filters increases monotonically with decreasing temporal width, while biphasic filters have a unique optimal time scale that roughly corresponds to the stimulus autocorrelation time. This study demonstrates that estimation-theoretic methods, which have previously been used mainly for static stimuli, can also yield quantitative results on neural codes for dynamic stimuli.

Temporally Faithful Representation of Salient Stimulus Movement Patterns in the Early Visual System

In human visual search, the time required to detect an object that starts to move is approximately independent of the number of distractors in the visual field (pop-out effect)1. Motion onset is therefore considered as being a “basic feature”. Several theoretical models of the visual system attempt to explain the interplay of such basic features, attentional selection, and higher-level processing2,3,4,5. Local contrast of basic features is thought to be calculated in parallel across the visual field in a first step. On the basis of this feature contrast computation, salient positions are determined. The most salient one is then selected by some form of winner-take-all mechanism, the output of which is used to direct attention to this potentially most interesting part of the visual scene.

Processing of movement information in the early stages of the visual system

We present a model for the processing of different movement patterns in the early visual system. The model consists of two processing stages: retina and optic tectum. The temporal structure of movement patterns is translated into a temporal structure of neural responses. This is accomplished by applying gain control in the retina, depressive synaptic transmission and adaptation in tectum to obtain a model that recovers during low activity in motion pauses. The model is especially sensitive to beginning movement after such pauses, enabling fast prey capture reactions in tongue-projecting salamanders.