Refractoriness in the maintained discharge of the cat's retinal ganglion cell

A numerical method for computing interval distributions for an inhomogeneous Poisson point process modified by random dead times

The inhomogeneous Poisson point process is a common model for time series of discrete, stochastic events. When an event from a point process is detected, it may trigger a random dead time in the detector, during which subsequent events will fail to be detected. It can be difficult or impossible to obtain a closed-form expression for the distribution of intervals between detections, even when the rate function (often referred to as the intensity function) and the dead-time distribution are given. Here, a method is presented to numerically compute the interval distribution expected for any arbitrary inhomogeneous Poisson point process modified by dead times drawn from any arbitrary distribution. In neuroscience, such a point process is used to model trains of neuronal spikes triggered by the detection of excitatory events while the neuron is not refractory. The assumptions of the method are that the process is observed over a finite observation window and that the detector is not in a dead state at the start of the observation window. Simulations are used to verify the method for several example point processes. The method should be useful for modeling and understanding the relationships between the rate functions and interval distributions of the event and detection processes, and how these relationships depend on the dead-time distribution.

Modeling elucidates how refractory period can provide profound nonlinear gain control to graded potential neurons

Refractory period (RP) plays a central role in neural signaling. Because it limits an excitable membrane's recovery time from a previous excitation, it can restrict information transmission. Classically, RP means the recovery time from an action potential (spike), and its impact to encoding has been mostly studied in spiking neurons. However, many sensory neurons do not communicate with spikes but convey information by graded potential changes. In these systems, RP can arise as an intrinsic property of their quantal micro/nanodomain sampling events, as recently revealed for quantum bumps (single photon responses) in microvillar photoreceptors. Whilst RP is directly unobservable and hard to measure, masked by the graded macroscopic response that integrates numerous quantal events, modeling can uncover its role in encoding. Here, we investigate computationally how RP can affect encoding of graded neural responses. Simulations in a simple stochastic process model for a fly photoreceptor elucidate how RP can profoundly contribute to nonlinear gain control to achieve a large dynamic range.

Dynamic patterns in a two-dimensional neural field with refractoriness

The formation of dynamic patterns such as localized propagating waves is a fascinating self-organizing phenomenon that happens in a wide range of spatially extended systems including neural systems, in which they might play important functional roles. Here we derive a type of two-dimensional neural-field model with refractoriness to study the formation mechanism of localized waves. After comparing this model with existing neural-field models, we show that it is able to generate a variety of localized patterns, including stationary bumps, localized waves rotating along a circular path, and localized waves with longer-range propagation. We construct explicit bump solutions for the two-dimensional neural field and conduct a linear stability analysis on how a stationary bump transitions to a propagating wave under different spatial eigenmode perturbations. The neural-field model is then partially solved in a comoving frame to obtain localized wave solutions, whose spatial profiles are in good agreement with those obtained from simulations. We demonstrate that when there are multiple such propagating waves, they exhibit rich propagation dynamics, including propagation along periodically oscillating and irregular trajectories; these propagation dynamics are quantitatively characterized. In addition, we show that these waves can have repulsive or merging collisions, depending on their collision angles and the refractoriness parameter. Due to its analytical tractability, the two-dimensional neural-field model provides a modeling framework for studying localized propagating waves and their interactions.

Prediction of human's ability in sound localization based on the statistical properties of spike trains along the brainstem auditory pathway

The minimum audible angle test which is commonly used for evaluating human localization ability depends on interaural time delay, interaural level differences, and spectral information about the acoustic stimulus. These physical properties are estimated at different stages along the brainstem auditory pathway. The interaural time delay is ambiguous at certain frequencies, thus confusion arises as to the source of these frequencies. It is assumed that in a typical minimum audible angle experiment, the brain acts as an unbiased optimal estimator and thus the human performance can be obtained by deriving optimal lower bounds. Two types of lower bounds are tested: the Cramer-Rao and the Barankin. The Cramer-Rao bound only takes into account the approximation of the true direction of the stimulus; the Barankin bound considers other possible directions that arise from the ambiguous phase information. These lower bounds are derived at the output of the auditory nerve and of the superior olivary complex where binaural cues are estimated. An agreement between human experimental data was obtained only when the superior olivary complex was considered and the Barankin lower bound was used. This result suggests that sound localization is estimated by the auditory nuclei using ambiguous binaural information.

Spatiotemporal pattern formation in two-dimensional neural circuits: roles of refractoriness and noise

Refractoriness is one of the most fundamental states of neural firing activity, in which neurons that have just fired are unable to produce another spike, regardless of the strength of afferent stimuli. Another essential and unavoidable feature of neural systems is the existence of noise. To study the role of these essential factors in spatiotemporal pattern formation in neural systems, a spatially expended neural network model is constructed, with the dynamics of its individual neurons capturing the three most essential states of the neural firing behavior: firing, refractory and resting, and the network topology consistent with the widely observed center-surround coupling manner in the real brain. By changing the refractory period with and without noise in a systematic way in the network, it is shown numerically and analytically that without refractoriness, or when the refractory period is smaller than a certain value, the collective activity pattern of the system consists of localized, oscillating patterns. However, when the refractory period is greater than a certain value, crescent-shaped, localized propagating patterns emerge in the presence of noise. It is further illustrated that the formation of the dynamical spiking patterns is due to a symmetry breaking mechanism, refractoriness-induced symmetry breaking; that is generated by the interplay of noise and refractoriness in the network model. This refractoriness-induced symmetry breaking provides a novel perspective on the emergence of localized, spiking wave patterns or spike timing sequences as ubiquitously observed in real neural systems; it therefore suggests that refractoriness may benefit neural systems in their temporal information processing, rather than limiting the performance of neurons, as has been conventionally thought. Our results also highlight the importance of considering noise in studying spatially extended neural systems, where it may facilitate the formation of spatiotemporal order.

Optimal sequential detection of stimuli from multiunit recordings taken in densely populated brain regions

We address the problem of detecting the presence of a recurring stimulus by monitoring the voltage on a multiunit electrode located in a brain region densely populated by stimulus reactive neurons. Published experimental results suggest that under these conditions, when a stimulus is present, the measurements are gaussian with typical second-order statistics. In this letter we systematically derive a generic, optimal detector for the presence of a stimulus in these conditions and describe its implementation. The optimality of the proposed detector is in the sense that it maximizes the life span (or time to injury) of the subject. In addition, we construct a model for the acquired multiunit signal drawing on basic assumptions regarding the nature of a single neuron, which explains the second-order statistics of the raw electrode voltage measurements that are high-pass-filtered above 300 Hz. The operation of the optimal detector and that of a simpler suboptimal detection scheme is demonstrated by simulations and on real electrophysiological data.

The temporal properties of the response of macaque ganglion cells and central mechanisms of flicker detection

This analysis assesses sensitivity of primate ganglion cells to sinusoidal modulation as a function of temporal frequency, based on the structure of their impulse trains; sensitivity to luminance and chromatic modulation was compared to human psychophysical sensitivity to similar stimuli. Each stimulus cycle was Fourier analyzed, and response amplitudes subjected to neurometric analysis; this assumes a detector with duration inversely proportional to frequency, that is, the stimulus epoch analyzed is a single cycle rather than a fixed duration, and provides an upper bound for a detection by an observer who bases judgments on a single cell. Signal-to-noise ratio for a given Fourier amplitude rapidly decreased with temporal frequency. This is a consequence of the statistics of impulse trains making up the response; at higher temporal frequencies, there are fewer impulses per cycle. Performance of this "single-cell" observer was then compared with that of modeled central detection mechanisms of fixed duration. For chromatic modulation, a filter/detector with a time constant of approximately 40 ms operating upon the parvocellular (PC) pathway provided a match to psychophysical results, whereas for luminance modulation, a filter/detection mechanism operating upon the magnocellular (MC) pathway with a time constant of approximately 5-10 ms provided a suitable match. The effects of summation and nonlinear interactions between cell inputs to detection are also considered in terms of enhanced sensitivity and "sharpness" of thresholds, that is, the steepness of the neurometric function. For both luminance (MC cells) and chromatic modulation (PC cells), restricted convergence (<20 cells) appears adequate to provide sharp thresholds and sensitivity comparable to psychophysical performance.

High Response Reliability of Neurons in Primary Visual Cortex (V1) of Alert, Trained Monkeys

The reliability of neuronal responses determines the resources needed to represent the external world and constrains the nature of the neural code. Studies of anesthetized animals have indicated that neuronal responses become progressively more variable as information travels from the retina to the cortex. These results have been interpreted to indicate that perception must be based on pooling across relatively large numbers of cells. However, we find that in alert monkeys, responses in primary visual cortex (V1) are as reliable as the inputs from the retina and the thalamus. Moreover, when the effects of fixational eye movements were minimized, response variability (variance/mean - Fano factor, FF) in all V1 layers was low. When presenting optimal stimuli, the median FF was 0.3. High variability, FF approximately 1, was found only near threshold. Our results suggest that in natural vision, suprathreshold perception can be based on small numbers of optimally stimulated cells.

Fractal features of dark, maintained, and driven neural discharges in the cat visual system

We employ a number of statistical measures to characterize neural discharge activity in cat retinal ganglion cells (RGCs) and in their target lateral geniculate nucleus (LGN) neurons under various stimulus conditions, and we develop a new measure to examine correlations in fractal activity between spike-train pairs. In the absence of stimulation (i.e., in the dark), RGC and LGN discharges exhibit similar properties. The presentation of a constant, uniform luminance to the eye reduces the fractal fluctuations in the RGC maintained discharge but enhances them in the target LGN discharge, so that neural activities in the pair cease to be mirror images of each other. A drifting-grating stimulus yields RGC and LGN driven spike trains similar in character to those observed in the maintained discharge, with two notable distinctions: action potentials are reorganized along the time axis so that they occur only during certain phases of the stimulus waveform, and fractal activity is suppressed. Under both uniform-luminance and drifting-grating stimulus conditions (but not in the dark), the discharges of pairs of LGN cells are highly correlated over long time scales; in contrast discharges of RGCs are nearly uncorrelated with each other. This indicates that action-potential activity at the LGN is subject to a common fractal modulation to which the RGCs are not subjected.

Low response variability in simultaneously recorded retinal, thalamic, and cortical neurons

The response of a cortical cell to a repeated stimulus can be highly variable from one trial to the next. Much lower variability has been reported of retinal cells. We recorded visual responses simultaneously from three successive stages of the cat visual system: retinal ganglion cells (RGCs), thalamic (LGN) relay cells, and simple cells in layer 4 of primary visual cortex. Spike count variability was lower than that of a Poisson process at all three stages but increased at each stage. Absolute and relative refractory periods largely accounted for the reliability at all three stages. Our results show that cortical responses can be more reliable than previously thought. The differences in reliability in retina, LGN, and cortex can be explained by (1) decreasing firing rates and (2) decreasing absolute and relative refractory periods.

Fractal character of the neural spike train in the visual system of the cat

We used a variety of statistical measures to identify the point process that describes the maintained discharge of retinal ganglion cells (RGC's) and neurons in the lateral geniculate nucleus (LGN) of the cat. These measures are based on both interevent intervals and event counts and include the interevent-interval histogram, rescaled range analysis, the event-number histogram, the Fano factor, Allan factor, and the periodogram. In addition, we applied these measures to surrogate versions of the data, generated by random shuffling of the order of interevent intervals. The continuing statistics reveal 1/f-type fluctuations in the data (long-duration power-law correlation), which are not present in the shuffled data. Estimates of the fractal exponents measured for RGC- and their target LGN-spike trains are similar in value, indicating that the fractal behavior either is transmitted form one cell to the other or has a common origin. The gamma-r renewal process model, often used in the analysis of visual-neuron interevent intervals, describes certain short-term features of the RGC and LGN data reasonably well but fails to account for the long-duration correlation. We present a new model for visual-system nerve-spike firings: a gamma-r renewal process whose mean is modulated by fractal binomial noise. This fractal, doubly stochastic point process characterizes the statistical behavior of both RGC and LGN data sets remarkably well.

A nonstationary Poisson point process describes the sequence of action potentials over long time scales in lateral-superior-olive auditory neurons

The behavior of lateral-superior-olive (LSO) auditory neurons over large time scales was investigated. Of particular interest was the determination as to whether LSO neurons exhibit the same type of fractal behavior as that observed in primary VIII-nerve auditory neurons. It has been suggested that this fractal behavior, apparent on long time scales, may play a role in optimally coding natural sounds. We found that a nonfractal model, the nonstationary dead-time-modified Poisson point process (DTMP), describes the LSO firing patterns well for time scales greater than a few tens of milliseconds, a region where the specific details of refractoriness are unimportant. The rate is given by the sum of two decaying exponential functions. The process is completely specified by the initial values and time constants of the two exponentials and by the dead-time relation. Specific measures of the firing patterns investigated were the interspike-interval (ISI) histogram, the Fano-factor time curve (FFC), and the serial count correlation coefficient (SCC) with the number of action potentials in successive counting times serving as the random variable. For all the data sets we examined, the latter portion of the recording was well approximated by a single exponential rate function since the initial exponential portion rapidly decreases to a negligible value. Analytical expressions available for the statistics of a DTMP with a single exponential rate function can therefore be used for this portion of the data. Good agreement was obtained among the analytical results, the computer simulation, and the experimental data on time scales where the details of refractoriness are insignificant.(ABSTRACT TRUNCATED AT 250 WORDS)

Rate fluctuations and fractional power-law noise recorded from cells in the lower auditory pathway of the cat

The noise properties of the sequence of action potentials recorded from adult-cat auditory nerve fibers and lateral superior olivary units have been investigated under various stimulus conditions. Large fluctuations exhibited by the spike rate, and spike clusters evident in the pulse-number distribution, both indicate an unusual underlying sequence of neural events. We present results demonstrating that (i) the firing rate calculated with different averaging times can exhibit self-similar behavior; (ii) the pulse-number distribution remains irregular even for large numbers of samples; (iii) the spike-number variance-to-mean ratio increases with the counting time T in fractional power-law fashion for sufficiently large T; and (iv) the exponent in the power law generally depends on the stimulus level. The results obtained in our laboratories support the notion that all auditory-nerve and LSO units exhibit fractal neural firing patterns, as indicated earlier by Teich (IEEE Trans. Biomed. Eng. 36, 150-160, 1989).

The effect of a random initial value in neural first-passage-time models

The effect of a random initial value is examined in several stochastic integrate-and-fire neural models with a constant threshold and a constant input. The three models considered are approximations of Stein's model, namely: (1) a leaky integrator with deterministic trajectories, (2) a Wiener process with drift, and (3) an Ornstein-Uhlenbeck process. For model 1, different distributions for the initial value lead to commonly observed interspike interval distributions. For model 2, a discrete and a uniform distribution for the initial value are examined along with some parameter estimation procedures. For model 3, with a truncated normal distribution for the initial value, the coefficient of variation is shown to be greater than 1, and as the threshold becomes large the first-passage-time distribution approaches an exponential distribution. The relationships among the models and between them and previous models are also discussed, along with the robustness of the model assumptions and methods of their verification. The effects of a random initial value are found to be most pronounced at high firing rates.

Multiplication noise in the human visual system at threshold: 1. Quantum fluctuations and minimum detectable energy

We have carried out a series of frequency-of-seeing experiments similar to those performed by Hecht, Shlaer, and Pirenne [J. Gen. Physiol. 25, 819-840 (1942)], using an Ar+ laser operated at 514.5 nm as the source of light. In certain blocks of trials, our subjects were encouraged to report as seen those trials in which the stimulus might have been present. It was determined that sensitivity and reliability were traded against each other over a broad range: for our subjects, the detection of 147 photons at the cornea with 60% frequency of seeing entailed, on the average, a 1% false-positive rate (FPR), whereas the detection of 34 photons at the cornea with 60% frequency of seeing was accompanied by a 33% FPR. A new neural-counting model has been developed in the framework of signal-detection theory. It combines Poisson stimulus fluctuations with additive and multiplicative neural noise, both of which are known to be present in the visual system at threshold. The resulting probability-of-detection curves, derived from the Neyman Type-A counting distribution, are in good accord with our experimental frequency-of-seeing data for sensible values of the model parameters. We deduce that, on the average, our four subjects are able to detect a single photon at the retina with 60% frequency of seeing, at the expense of a 55% FPR. In Part 2 of this set of papers [P.R. Prucnal and M.C. Teich, Biol. Cybern. 43, 87-96 (1982)], we use the normalizing transform, together with probit analysis, to provide improved estimates of threshold parameters, whereas in Part 3 [M.C. Teich, P.R. Prucnal, G. Vannucci, M.E. Breton, and W.J. McGill, submitted to Biol. Cybern.], we consider the effects of non-Poisson quantum fluctuations.

Firing rate of a retinal neuron are not predictable from interspike interval statistics

The intervals between successive action potentials (impulses, or "spikes") produced the maintained firing of a neuron (ISIs) are often treated as if they were independent on each other; that is, an impulse train is considered as a stationary renewal process. If this is so, the variability of the mean rate of firing impulses in a sequence of temporal windows should be predictable from the distribution of ISIs. This was found not to be the case for the maintained firing of retinal ganglion cells in goldfish. Although some evident nonstationarity sometimes resulted in greater variability of the observed rate distributions than those predicted (for relatively long temporal windows), as a general rule the observed rate distributions were considerable less dispersed than would be predicted by sampling of the ISI distributions. This was taken as evidence of long-term serial dependency between successive ISIs; however, two standard test for dependency (autocorrelations and serial correlograms failed to to reveal structure of sufficiently long duration to account for the effect noted.