How does a neuron vary its mean output firing rate if the input changes from random to coherent activity? What are the critical parameters of the neuronal dynamics and input statistics? To answer these questions, we investigate the coincidence detection properties of an integrate-and-fire neuron. We derive an expression indicating how coincidence detection depends on neuronal parameters. Specifically, (i) we show how coincidence detection depends on the shape of the postsynaptic response function, the number of synapses, and the input statistics, and (ii) we demonstrate that there is an optimal threshold. Our considerations can be used to predict from neuronal parameters whether and to what extent a neuron can act as a coincidence detector and thus can convert a temporal code into a rate code. Physik-Department der TU Munchen (T35), D-85747 Garching bei Munchen, Germany y Swiss Federal Institute of Technology, Center of Neuromimetic Systems, EPFL-DI, CH-1015 Lausanne, Switz...

Introduction Small networks that generate rhythmic motor patterns in invertebrates have been the subject of intense experimental and theoretical analyses aimed at articulating some of the basic principles that govern the dynamics of neural circuits (Getting, 1989). Although there are significant differences between small invertebrate networks and large vertebrate brain structures, they share a large number of organizational and functional mechanisms. Because invertebrate pattern generators contain relatively small numbers of large, easily studied neurons, they offer the potential for uncovering some of the basic computational and functional characteristics of neural circuits in the brain. Central pattern generating networks are groups of neurons that produce rhythmic motor patterns even in the absence of timing signals from sensory or central inputs (Marder and Calabrese, 1996). These circuits are involved in behaviors such as walking, swimming, flying, breathing, and chewing.

Some of the cochlear nuclei in the auditory pathway are specialized for the sound localization. They compute the interaural time difference. The difference in sound timing is transduced by the dedicated neuronal circuit into a labeled line difference. The detector neurons along the delay line fire only when synaptic inputs reflecting signals from both ears arrive within a short time window. It was therefore called coincidence detection. We show, (1) what are the limits of coincidence detection in the leaky integrator model, which is a linear system, (2) how should the ideal coincidence detector based on the Hodkin -- Huxley equations from real neurons look like, (3) what are the properties and physical limits in the real coincidence detection system. The conclusion is that the neuron with the Hodgkin -- Huxley dynamics has a fixed precision for the coincidence detection. The limits of the sound localization precision are set by the frequency of the sound and, therefore, by the vector strength of spike trains generated in the neuronal circuit in response to the sound. 2000 Elsevier Science Ireland Ltd. All rights reserved.

Precise temporal coding is a hallmark of the auditory system. Selective pressures to improve accuracy or encode more rapid changes have produced a suite of convergent physiological and morphological features that contribute to temporal coding. Comparative studies of temporal coding also point to shared computational strategies, and suggest how selection acts to improve coding. Both the avian cochlear nucleus angularis and the mammalian cochlear nuclei have heterogeneous cell populations, and similar responses to sound. These shared characteristics may represent convergent responses to similar selective pressures to encode features of airborne sound.

Introduction In birds and mammals, precisely timed spikes encode the timing of acoustic stimuli, and interaural acoustic disparities propagate to binaural processing centers such as the avian nucleus laminaris (NL) and the mammalian medial superior olive (MSO; Young & Rubel, 1983; Carr & Konishi, 1990; Smith et al., 1993). The projections from the cochlear nucleus magnocellularis (NM) to NL and from mammalian spherical bushy cells to MSO resemble the Jeffress model for encoding interaural time differences (Jeffress, 1948). The Jeffress model has two fundamental elements: delay lines and coincidence detectors. A Jeffress circuit is an array of coincidence detectors, every element of which has a different relative delay between its ipsilateral and contralateral excitatory inputs. Thus the interaural time difference (ITD) is encoded into the position (a place code) of the coincidence detector whose delay lines best nullify the acoustic ITD. The neurons of NL and MSO phase-lock to both mo

Unreliability is a ubiquitous feature of synaptic transmission in the brain. The information conveyed in the discharges of an ensemble of cells (e.g., in the spike count or in the timing of synchronous events) may not be faithfully transmitted to the postsynaptic cell because a large fraction of the spikes fail to elicit a synaptic response. In addition, short-term depression increases the failure rate with the presynaptic activity. We use a simple neuron model with stochastic depressing synapses to understand the transformations undergone by the spatiotemporal patterns of incoming spikes as these are first converted into synaptic current and afterward into the cell response. We analyze the mean and SD of the current produced by different stimuli with spatiotemporal correlations. We find that the mean, which carries information only about the spike count, rapidly saturates as the input rate increases. In contrast, the current deviation carries information about the correlations. If the afferent action potentials are uncorrelated, it saturates monotonically, whereas if they are correlated it increases, reaches a maximum, and then decreases to the value produced by the uncorrelated stimulus. This means that, at high input rates, depression erases from the synaptic current any trace of the spatiotemporal structure of the input. The non-monotonic behavior of the deviation can be inherited by the response rate provided that the mean current saturates below the current threshold setting the cell in the fluctuation-driven regimen. Afferent correlations therefore enable the modulation of the response beyond the saturation of the mean current. Key words: synaptic integration; fluctuation-driven regimen; presynaptic spike correlations; synaptic short-term depression; vesicle depletion; neural coding

In this paper, we examine unsupervised learning for sequence of sub-millisecond temporal coded information in a network of neurons, which are assumed to have high temporal resolution. The learning scheme is based on a spatially and temporally local one, i.e., unsupervised Hebbian learning. The input sequence is temporal information that needs an accuracy on the order of sub-milliseconds. Through the learning, segregation of the synaptic connections occurs to form systematic structures in the network. Namely, the network develops in a self-organizing manner. The trained network works like an "associative memory" of the learned sequence, namely, the network responds when a newly input sequence is similar to the learned sequence. Consequently, the assembly of neurons is able to learn and distinguish an input sequence that carries information on the order of sub-milliseconds, although the spike emission intervals of the neurons are on the order of milliseconds. Unsupervised network learn...

Introduction The relevance of precise temporal spike timing in the cortex -- or neural systems in general -- is a fundamental, yet unsolved question [Abeles, 1994; Bialek et al., 1991; Hopfield, 1995; O'Keefe and Recce, 1993; Mainen and Sejnowski, 1995; Shadlen and Newsome, 1994; Softky, 1995; Rieke et al., 1996]; see also Chapters 1 and 4 of this book. There are, however, a few specialized subsystems for which the relevance of temporal information has been clearly shown. Prominent examples are the electro-sensory system of electric fish and the auditory system of barn owls [Carr and Konishi, 1990; Carr, 1993; Heiligenberg, 1991; Konishi, 1986; Konishi, 1993]. Here we use the latter as an example to study the following question: How can the timing of spikes be learned during early development? Specifically, how can processes of pulse generation and signal transmission be fine-tuned so as to achieve the required temporal precision? The problem of spike-based learning arises wh

www.elsevier.com/locate/jphysparis Coding interaural time differences at low best frequencies in the barn owl