Witali L. Dunin-Barkowski

Endogenous excitatory drive to the respiratory system in rapid eye movement sleep in cats

1 A putative endogenous excitatory drive to the respiratory system in rapid eye movement (REM) sleep may explain many characteristics of breathing in that state, e.g. its irregularity and variable ventilatory responses to chemical stimuli. This drive is hypothetical, and determinations of its existence and character are complicated by control of the respiratory system by the oscillator and its feedback mechanisms. In the present study, endogenous drive was studied during apnoea caused by mechanical hyperventilation. We reasoned that if there was a REM‐dependent drive to the respiratory system, then respiratory activity should emerge out of the background apnoea as a manifestation of the drive. 2 Diaphragmatic muscle or medullary respiratory neuronal activity was studied in five intact, unanaesthetized adult cats who were either mechanically hyperventilated or breathed spontaneously in more than 100 REM sleep periods. 3 Diaphragmatic activity emerged out of a background apnoea caused by mechanical hyperventilation an average of 34 s after the onset of REM sleep. Emergent activity occurred in 60 % of 10 s epochs in REM sleep and the amount of activity per unit time averaged approximately 40 % of eupnoeic activity. The activity occurred in episodes and was poorly related to pontogeniculo‐occipital waves. At low CO2 levels, this activity was non‐rhythmic. At higher CO2 levels (less than 0.5 % below eupnoeic end‐tidal percentage CO2 levels in non‐REM (NREM) sleep), activity became rhythmic. 4 Medullary respiratory neurons were recorded in one of the five animals. Nineteen of twenty‐seven medullary respiratory neurons were excited in REM sleep during apnoea. Excited neurons included inspiratory, expiratory and phase‐spanning neurons. Excitation began about 43 s after the onset of REM sleep. Activity increased from an average of 6 impulses s−1 in NREM sleep to 15.5 impulses s−1 in REM sleep. Neuronal activity was non‐rhythmic at low CO2 levels and became rhythmic when levels were less than 0.5 % below eupnoeic end‐tidal levels in NREM sleep. The level of CO2 at which rhythmic neuronal activity developed corresponded to eupnoeic end‐tidal CO2 levels in REM sleep. 5 These results demonstrate an endogenous excitatory drive to the respiratory system in REM sleep and account for rapid and irregular breathing and the lower set‐point to CO2 in that state.

Respiratory pattern generator model using Ca++-induced Ca++ release in neurons shows both pacemaker and reciprocal network properties

Abstract.There are two contradictory explanations for central respiratory rhythmogenesis. One suggests that respiratory rhythm emerges from interaction between inspiratory and expiratory neural semicenters that inhibit each other and thereby provide reciprocal rhythmic activity (Brown 1914). The other uses bursting pacemaker activity of individual neurons to produce the rhythm (Feldman and Cleland 1982). Hybrid models have been developed to reconcile these two seemingly conflicting mechanisms (Smith et al. 2000; Rybak et al. 2001). Here we report computer simulations that demonstrate a unified mechanism of the two types of oscillator. In the model, we use the interaction of Ca++-dependent K+ channels (Mifflin et al. 1985) with Ca++-induced Ca++ release from intracellular stores (McPherson and Campbell 1993), which was recently revealed in neurons (Hernandez-Cruz et al. 1997; Mitra and Slaughter 2002a,b; Scornik et al. 2001). Our computations demonstrate that uncoupled neurons with these intracellular mechanisms show conditional pacemaker properties (Butera et al. 1999) when exposed to steady excitatory inputs. Adding weak inhibitory synapses (based on increased K+ conductivity) between two model neural pools surprisingly synchronizes the activity of both neural pools. As inhibitory synaptic connections between the two pools increase from zero to higher values, the model produces first dissociated pacemaker activity of individual neurons, then periodic synchronous bursts of all neurons (inspiratory and expiratory), and finally reciprocal rhythmic activity of the neural pools.

An approximate backpropagation learning rule for memristor based neural networks using synaptic plasticity

We describe an approximation to backpropagation algorithm for training deep neural networks, which is designed to work with synapses implemented with memristors. The key idea is to represent the values of both the input signal and the backpropagated delta value with a series of pulses that trigger multiple positive or negative updates of the synaptic weight, and to use the min operation instead of the product of the two signals. In computational simulations, we show that the proposed approximation to backpropagation is well converged and may be suitable for memristor implementations of multilayer neural networks.

Models of Innate Neural Attractors and Their Applications for Neural Information Processing.

In this work we reveal and explore a new class of attractor neural networks, based on inborn connections provided by model molecular markers, the molecular marker based attractor neural networks (MMBANN). Each set of markers has a metric, which is used to make connections between neurons containing the markers. We have explored conditions for the existence of attractor states, critical relations between their parameters and the spectrum of single neuron models, which can implement the MMBANN. Besides, we describe functional models (perceptron and SOM), which obtain significant advantages over the traditional implementation of these models, while using MMBANN. In particular, a perceptron, based on MMBANN, gets specificity gain in orders of error probabilities values, MMBANN SOM obtains real neurophysiological meaning, the number of possible grandma cells increases 1000-fold with MMBANN. MMBANN have sets of attractor states, which can serve as finite grids for representation of variables in computations. These grids may show dimensions of d = 0, 1, 2,…. We work with static and dynamic attractor neural networks of the dimensions d = 0 and 1. We also argue that the number of dimensions which can be represented by attractors of activities of neural networks with the number of elements N = 104 does not exceed 8.

Tonic and phasic drive to medullary respiratory neurons during periodic breathing

It is unknown how central neural activity produces the repetitive termination and restart of periodic breathing (PB). We hypothesized that inspiratory and expiratory neural activities would be greatest during the waxing phase and least during the waning phase. We analyzed diaphragmatic and medullary respiratory neural activities during PB in intact unanesthetized adult cats. Diaphragmatic activity was increased and phasic during the waxing phase and was decreased and tonic during the waning phase. Activity of expiratory (n=21) and inspiratory (n=40) neurons was generally increased and phasic during the waxing phase and was decreased and more tonic during the waning phase. During apneas associated with PB, diaphragmatic activity was silent and most, but not all, inspiratory cells were inactive whereas most expiratory cells decreased activity but remained tonically active. We suggest that reduced strength of reciprocal inhibition, secondary to reduced respiratory drive, allows for simultaneous tonic activity of inspiratory and expiratory neurons of the central pattern generator, ultimately resulting in central apnea.

Phase-based storage of information in the cerebellum

Abstract Recently the importance of both increasing and decreasing synaptic strength of parallel fibers to the cerebellar Purkinje cells has been stressed (Kenyon et al., J. Comput. Neusci. 5 (1998) 17-25, 71-82). This idea resolves some of the paradoxes connected to the old problem of cerebellar learning. A phase-based system for information storage can be proposed to model cerebellar information storage. We describe here the behavior of such a system, when input signals are piece-wise constant, and discuss more general cases qualitatively. The theory gives ideas for experiments, which may yield new light on cerebellar functions and modus operandi. The general formulation of cerebellar functions is also presented in the context of the discussion of its information storage properties.

Stability Properties of Cerebellar Neural Networks: The Purkinje Cell – Climbing Fiber Dynamic Module

In the last few decades it has been proven, that the cerebellum takes part in learning the bulk of motor control. The mechanisms which provide such properties are still largely unknown, but an involvement of parallel fibers and climbing fibers in this process, as have been proposed decades ago in cerebellar learning theories, is now clear. Among difficulties of the learning theories is an evident necessity for spontaneous activity of the cerebellar climbing fibers [5]. Recently, the group of M. Mauk proposed an elegant explanation of this inconsistency [11, 12]. We present here a stochastic model of a cerebellar module, based on this new approach. Theoretical treatment yields some consequences for experimental verification. Besides an explanation of real cerebellar functions, the analyzed control system presents a new paradigm for neural network memorizing systems.

L-plotting-A method for visual analysis of physiological experimental and modeling multi-component data

A method for visualization of dynamic multidimensional data-L-plotting, similar to recurrence plotting, is described. For multi-neuronal brainstem recordings the method demonstrates that the neural respiratory pattern generator (RPG) switches between the two phases: inspiratory and expiratory. The method helps to mark phase switching moments and to characterize the pattern of the RPG restart after temporary cessation of rhythmicity. Comparison of L-plots for experimental data and network simulations helps verification of computational models.

Analysis of output of all Purkinje cells controlled by one climbing fiber cell

Abstract It was hypothesized earlier that synaptic plasticity of numerous granule cells–Purkinje cell connections might equalize input to climbing fiber cell using the closed loop, climbing fiber cell–Purkinje cells–climbing fiber cell, and bi-directional synaptic plasticity of granule cells–Purkinje cell synapses. This equalization is demonstrated in our computational modeling. The system stability depends on metabotropic (non-electric) action of granule cells synapses. After equalization, Purkinje cell output presents a “restored replica” of extra-cerebellar synaptic input to the climbing fiber cell. Inter-spike intervals of the latter are virtually chaotic and their statistical pattern resembles natural climbing fiber activity. Modes of action of sets of Purkinje cells, controlled with the same climbing fiber cell and of ensembles of climbing fiber cells are treated analytically and qualitatively. Experimental consequences of the described cerebellar modus operandi are discussed.

Precise rhythmicity in activity of neocortical, thalamic and brain stem neurons in behaving cats and rabbits

Rhythmic discharges of neurons are believed to be involved in information processing in both sensory and motor systems. However their fine structure and functional role need further elucidation. We employed a pattern-based approach to search for episodes of precisely rhythmic activity of single neurons recorded in different brain structures in behaving cats and rabbits. We defined discharge patterns using an algorithmic description, which is different from the previously suggested template methods. We detected episodes of precisely rhythmic discharges, specifically, triads of constant (precision +/-2.5%) inter-spike intervals in the 10-70 ms range. In 54% (67/125) of neurons tested, these patterns could not be explained by random occurrences or by steady or slowly changing input. Rhythmic patterns occurred at a wide range of inter-spike intervals, and were imbedded in non-rhythmic activity. In many neurons, timing of these precisely rhythmic patterns was related to different locomotion tasks or to respiration.