Search

We characterize the attractor-based expressive power of several models of recurrent neural networks.The deterministic rational-weighted networks are Muller Turing equivalent.The deterministic real-weighted and evolving networks recognize the class of B C ( ź 2 0 ) neural ω languages.The nondeterministic rational and real networks recognize the class of Σ 1 1 neural ω-languages. We provide a characterization of the expressive powers of several models of deterministic and nondeterministic first-order recurrent neural networks according to their attractor dynamics. The expressive power of neural nets is expressed as the topological complexity of their underlying neural ω-languages, and refers to the ability of the networks to perform more or less complicated classification tasks via the manifestation of specific attractor dynamics. In this context, we prove that most neural models under consideration are strictly more powerful than Muller Turing machines. These results provide new insights into the computational capabilities of recurrent neural networks.

International audience; We consider analog recurrent neural networks working on in nite input streams, provide a complete topological characterization of their expressive power, and compare it to the expressive power of classical in nite word reading abstract machines. More precisely, we consider analog recurrent neural networks as language recognizers over the Cantor space, and prove that the classes of !-languages recognized by deterministic and non-deterministic analog networks correspond precisely to the respective classes of 02 -sets and 11 -sets of the Cantor space. Furthermore, we show that the result can be generalized to more expressive analog networks equipped with any kind of Borel accepting condition. Therefore, in the deterministic case, the expressive power of analog neural nets turns out to be comparable to the expressive power of any kind of Buchi abstract machine, whereas in the non-deterministic case, analog recurrent networks turn out to be strictly more expressive than any other kind of Buchi or Muller abstract machine, including the main cases of classical automata, 1-counter automata, k-counter automata, pushdown automata, and Turing machines.

International audience; Neural development and differentiation are characterized by an overproduction of cells and a transient exuberant number of connections followed by cell death and selective synaptic pruning. We simulated large spiking neural networks (10,000 units at its maximum size) with and without an ontogenetic process corresponding to a brief initial phase of apoptosis driven by an excessive firing rate mimicking cell death due to glutamatergic neurotoxicity and glutamate-triggered apoptosis. This phase was followed by the onset of spike timing dependent synaptic plasticity (STDP), driven by spatiotemporal patterns of stimulation. Despite the reduction in cell counts the apoptosis tended to increase the excitatory/inhibitory ratio because the inhibitory cells were affected at first. Recurrent spatiotemporal firing patterns emerged in both developmental condition but they differed in dynamics. They were less numerous but repeated more often after apoptosis. The results suggest that initial cell death may be necessary for the emergence of stable cell assemblies, able to sustain and process temporal information, from the initially randomly connected networks.

Precise spatiotemporal sequences of neuronal discharges (i.e., intervals between epochs repeating more often than expected by chance), have been observed in a large set of experimental electrophysiological recordings. Sensitivity to temporal information, by itself, does not demonstrate that dynamics embedded in spike trains can be transmitted through a neural network. This study analyzes how synaptic transmission through three archetypical types of neurons (regular-spiking, thalamo-cortical and resonator), simulated by a simple spiking model, can affect the transmission of precise timings generated by a nonlinear deterministic system (i.e., the Zaslavskii mapping in the present study). The results show that cells with subthreshold oscillations (resonators) are very sensitive to stochastic inputs, and are not a good candidate for transmitting temporally coded information. Thalamo-cortical neurons may transmit very well temporal patterns in the absence of background activity, but jitter accumulates along the synaptic chain. Conversely, we observed that cortical regular-spiking neurons can propagate filtered temporal information in a reliable way through the network, and with high temporal accuracy. We discuss the results in the general framework of neural dynamics and brain theories.

B-MYB is an oncoprotein highly expressed and frequently amplified in human neoplasia. B-MYB is more expressed in neuroblastoma patients with adverse prognostic indicators, corroborating the hypothesis that it plays an important role in this pediatric malignancy. While attempting targeting strategies for therapeutic purposes, we found that the B-MYB protein was difficult to downregulate in neuroblastoma cells using siRNA approaches. This lead us to discover that the B-MYB protein half-life is increased in neuroblastoma compared to other normal or transformed human cell lines. The B-MYB protein is quickly destroyed and apoptosis is induced in Ewing sarcoma cells exposed to UV irradiation. In contrast, neuroblastoma cells are resistant to UV-induced apoptosis and B-MYB protein levels do not change in UV-treated cells. In further experiments, we show that the B-MYB protein extracted from neuroblastoma cells is hypophosphorylated. It was previously shown that B-MYB phosphorylation activates its transcriptional activity but also promotes its destruction. Overexpression of a non-phosphorylatable B-MYB mutant protects cells from UV-induced apoptosis, suggesting that its reduced phosphorylation, rather than causing its inactivation, facilitates B-MYB pro-survival activity. Thus, expression of stable, hypophosphorylated B-MYB in neuroblastoma may promote cell survival and induce aggressive tumour growth.

We investigated the role of the two calcium-binding proteins parvalbumin (PV) and calbindin D-28k (CB) in the locomotor activity and motor coordination using null-mutant mice for PV (PV−/−), CB (CB−/−) or both proteins (PV−/−CB−/−). These proteins are expressed in distinct, mainly non-overlapping populations of neurons of the central and peripheral nervous system and PV additionally in fast-twitch muscles. In a test measuring repeated locomotor activity during 18–20 days, the analysis revealed a slightly increased activity in mice lacking either protein, while the lack of both decreased the number of beams crossed during active periods. An increase in the characteristic speed during the first 8 days could be attributed to PVdeficiency, while the elimination of CB in CB−/− and double-KO mice decreased the percentage of fast movements at all time points. In the latter, additionally a reduction of the fastest speed was observed. The alterations in locomotor activity (fast movements, fastest speed) strongly correlate with the impairment in locomotor coordination in mice deficient for CB evidenced in the runway assay and the rotarod assay. The graded locomotor phenotype (CB > PV) is qualitatively correlated with alterations in Purkinje cell firing reported previously in these mice. The presence or absence of either protein did not affect the spontaneous locomotor activity when animals were placed in a novel environment and tested only once for 30 min. In summary, the lack of these calcium-binding proteins yields characteristic, yet distinct phenotypes with respect to locomotor activity and coordination. © 2007 Elsevier B.V. All rights reserved.

International audience; The present study demonstrates the application of the Unsupervised Spike Sorting algorithm (USS) to separation of multi-unit recordings and investigation of neuronal activity patterns in the subthalamic nucleus (STN). This nucleus is the main target for deep brain stimulation (DBS) in Parkinsonian patients. The USS comprises a fast unsupervised learning procedure and allows sorting of multiple single units, if any, out of a bioelectric signal. The algorithm was tested on a simulated signal with different levels of noise and with application of Time and Spatial Adaptation (TSA) algorithm for denoising. The results of the test showed a good quality of spike separation and allow its application to investigation of neuronal activity patterns in a medical application. One hundred twenty-four single channel multi-unit records from STN of 6 Parkinsonian patients were separated with USS into 492 single unit trains. Auto- and crosscorrellograms for each unit were analyzed in order to reveal oscillatory, bursting and synchronized activity patterns. We analyzed separately two brain hemispheres. For each hemisphere the percentage of units of each activity pattern were calculated. The results were compared for the first and the second operated hemispheres of each patient and in total.

International audience; Evoked potentials were recorded from the auditory cortex of both freely moving and anesthetized rats when deviant sounds were presented in a homogenous series of standard sounds (oddball condition). A component of the evoked response to deviant sounds, the mismatch negativity (MMN), may underlie the ability to discriminate acoustic differences, a fundamental aspect of auditory perception. Whereas most MMN studies in animals have been done using simple sounds, this study involved a more complex set of sounds (synthesized vowels). The freely moving rats had previously undergone behavioral training in which they learned to respond differentially to these sounds. Although we found little evidence in this preparation for the typical, epidurally recorded, MMN response, a significant difference between deviant and standard evoked potentials was noted for the freely moving animals in the 100-200 ms range following stimulus onset. No such difference was found in the anesthetized animals.

The effects of iontophoretic application of dopamine and selective D1 or D2 dopamine receptor agonists and antagonists on impulse activity of neurons of the deep layers of the sensorimotor cortex of cat were investigated during performance of a conditioned paw movement task. The application of dopamine, Quinpirole (selective D2 receptor agonist) or SKF 38393 (selective D1 receptor agonist) increased both background (P

A critical feature of brain theories is whether neurons convey a noisy rate code or a precise temporal code. One of most valuable ways to test these theories consists in collecting the electrophysiological activity of cell assembles under several experimental conditions. The sequences of cell discharges—the spike trains—form time series whose dynamics is strongly related to the information processing carried out in the brain areas under study. Our purpose is to provide a user-friendly framework of a ‘Virtual Laboratory’ where computational neuroscience analyses and display of results can be distributed over a computer network, like Internet.

The existence of precise temporal relations in sequences of spike intervals, referred to as “spatiotemporal patterns”, is suggested by brain theories that emphasize the role of temporal coding. A pattern grouping algorithm was designed to identify and to evaluate the statistical significance of such patterns, particularly for data generated according to stationary Poisson processes. The experimental time series, however, can be characterized by considerable deviations from independent stationary Poisson processes. This article describes a filtering method that de-convolute time series according to their correlation functions and makes possible an application of the pattern grouping algorithm for such data too.

When porcine endothelial cells were exposed to hypertonicity, both the level of ATA2 (amino acid transporter 2) mRNA and activity of amino acid transport System A increased transiently, peaking after about 6 and 9 h, respectively. Cycloheximide, like actinomycin D, prevented both responses, showing that an earlier step also involves protein synthesis. Withdrawal of hypertonicity after 6 h increased the rate of down regulation. These findings confirm that ATA2 is a major isoform of System A and show that changes in the expression of ATA2 mRNA precede both the induction and subsequent down regulation of transport activity.

The existence of precise temporal relations in sequences of spike intervals, referred to as ‘spatiotemporal patterns’, is suggested by brain theories that emphasize the role of temporal coding. Specific analytical methods able to assess the significance of such patterned activity are extremely important to establish its function for information processing in the brain. This study proposes a new method called ‘pattern grouping algorithm’ (PGA), designed to identify and evaluate the statistical significance of patterns which differ from each other by a defined and small jitter in spike timing of the order of few ms. The algorithm performs a pre-selection of template patterns with a fast computational approach, optimizes the jitter for each spike in the template and evaluates the statistical significance of the pattern group using three complementary statistical approaches. Simulated data sets characterized by various types of known non stationarities are used for validation of PGA and for comparison of its performance to other methods. Applications of PGA to experimental data sets of simultaneously recorded spike trains are described in a companion paper (Tetko IV, Villa AEP. A pattern grouping algorithm for analysis of spatiotemporal patterns in neuronal spike trains. 2. Application to simultaneous single unit recordings. J Neurosci Methods 2000; accompanying article).

This study demonstrates the practical application of the pattern grouping algorithm (PGA), presented in the companion paper (Tetko IV, Villa AEP. A pattern grouping algorithm for analysis of spatiotemporal patterns in neuronal spike trains. 1. Detection of repeated patterns. J. Neurosci. Methods 2000; accompanying article), to data sets including up to 30 simultaneously recorded spike trains. The analysis of a large network of simulated neurons shows that the incidence of patterns cannot be simply related to an increase in firing rates obtained after Hebbian learning. Patterns that disappeared and reappeared in the thalamus of anesthetized rats when the cerebral cortex was reversibly inactivated suggest that widespread cell assemblies contribute to the generation and propagation of precisely timed activity. In an another experiment multiple spike trains were recorded from the temporal cortex of freely moving rats performing a complex two-choice discrimination task. The presence or absence of particular patterns in the period preceding the cue was associated with changes in reaction time. In conclusion, neuronal network interactions may generate spatiotemporal firing patterns detectable by PGA. We provide evidence of such patterned activity associated with specific animal's behavior, thus suggesting the existence of complex temporal coding schemes in the higher nervous centers of the brain.

Neuronal activity was established in the auditory pathways in relation to behavioural response and cognitive information processing during a sensory-motor acoustic learning. Rats were trained in three consecutive phases. The first phase was an association between an auditory stimulus and a food reward; the second phase a simple discrimination between two sounds of different frequency components, and the third phase a more complex discrimination involving both spectral and spatial sound dimensions. Auditory stimuli were bursts of complex sounds lasting 500 ms. Neuronal activity related to the behaviourally relevant stimuli was established in 20 “learning” rats undergoing this protocol, which were progressively sacrificed at the beginning, middle and end of each phase. For comparison, activity was also established in four “control” rats exposed to the same stimuli delivered pseudo-randomly, thus carrying no behavioural meaning. Neuronal activity was assessed immunocytochemically using the functional marker Fos. To establish a baseline, two rats were unexposed to controlled acoustic stimulation (“unstimulated” rats). In the superior olivary complex (SOC), inferior colliculus (IC) and medial geniculate body (MGB), the number of Fos-like immunopositive cells was comparable in “learning” and “control” animals, but higher than in the “unstimulated” rats. In the auditory cortex (AC), most prominently in the secondary area Te2, the number of Fos-like positive cells differed between “learning” and “control” rats, suggesting that the auditory cortical areas may be involved in the encoding of the behavioural significance of the acoustic stimuli.

This study investigates the emerging possibilities of combining unsupervised and supervised learning in neural network ensembles. Such strategy is used to get an efficient partition of a noisy input data set in order to focus the training of neural networks on the most complex and informative domains of the data set and accelerate the learning phase. The proposed algorithm provides a good prediction accuracy using fewer cases from non-informative domains according to a correlative measure of dependency between cases of the training set. This measure takes into account internal relationships amid analyzed data and can be used to cluster neighbor cases in a multidimensional space and to filter out the outliers. The possible relation of the proposed algorithm to brain processing occurring in the thalamo-cortical pathway is discussed.

This study compares the temporal pattern of discharges of extracellularly recorded substantia nigra pars reticulata (SNr) single units in two experimental conditions: Equithesin- and ketamine-induced anesthesia. The analysis of the statistical properties of the spike trains recorded in the Equithesin group of animals showed that this experimental condition could be considered as a control condition with respect to previous data reported in the literature. We investigated the glutamatergic modulation of SNr activity at spike train level in a steady-state condition by using the anesthetic agent ketamine, which is a noncompetitive antagonist of the N -methyl- d -aspartate (NMDA) glutamatergic receptors. The most relevant effect of ketamine at single unit level was to induce burst discharges, with an intraburst frequency rate near 50 Hz, specifically in units characterized by an initial long refractoriness in the Equithesin condition. The other classes of single units tended to discharge at a higher rate without any significant change in their temporal pattern of firing. Simultaneous recording of the spike trains of 108 SNr pairs (46 and 62 during Equithesin and ketamine condition, respectively) were equally distributed between pairs of units simultaneously recorded from the same electrode and from distinct electrodes at a distance up to 400 μ m in the same hemisphere. Ketamine induced a significant increase in the number of pairs with synchronous firing (from 4 to 49%), which was strongly, but not exclusively, associated with an increased tendency to fire in bursts. Neighboring cells tended to fire with a similar pattern in either condition of recording, whereas synchronous firing between distant cells was observed only during ketamine condition.

Previous electrophysiological studies have shown that spontaneously active mesencephalic 5-hydroxytryptaminergic neurons of anaesthetized or freely moving animals fire solitary spikes in a slow, regular pattern. In the present study, using extracellular single unit recordings from dorsal and median raphe neurons of the anaesthetized rat, an additional electrophysiological property of a sub-population of presumed 5-hydroxytryptaminergic neurons was observed. These neurons, during their otherwise regular firing pattern, repeatedly fired two (or occasionally three or even four) spikes where only one was expected. Spikes in this burst-like repetitive firing mode (spikes in doublets or triplets) occurred in a short time interval (range: 2.4–11.5 ms), and with a diminishing spike amplitude. Cross-correlation analysis of spikes in doublets revealed a very high interdependency between them. The proportion of spikes in doublets to solitary spikes showed great variation between different neurons, ranging from 5 to 95% of the total spikes displayed. However, for each neuron the proportion of spikes in doublets to solitary spikes, and the time interval between the spikes in doublets, remained constant during control recordings. All these features are characteristic of single neurons firing in a repetitive firing pattern rather than simultaneous recordings of two separate 5-hydroxytryptaminergic neurons. Repetitive firing neurons were recorded with a similar frequency in both chloral hydrate and Saffan anaesthetized rats, and were detected using both glass and metal electrodes. Furthermore, neurons with a repetitive firing pattern were inhibited by intravenous administration of a selective 5-hydroxytryptamine1A receptor agonist and a 5-hydroxytryptamine reuptake inhibitor, thus displaying responses typical of 5-hydroxytryptaminergic neurons. Repetitive firing neurons occurred in both the dorsal and median raphe nuclei, although they were much more frequent in the dorsal raphe nucleus (91 of 332 neurons). The occurrence of repetitive firing neurons in the midbrain raphe nuclei is a newly described phenomenon which may indicate unique properties of a sub-population of 5-hydroxytryptaminergic neurons. In functional terms, it could modify both the axonal and dendritic 5-hydroxytryptamine release, and provide an additional option for neuronal information signalling.

Visual processing of sinusoidally modulated gratings was studied in a group of patients (n = 11) with Alzheimer's disease (AD) and an elderly normal control group (n = 9). Spatial square wave gratings (1.47 c/d) were reversed at a temporal frequency of 4 or 8 Hz. EEG recordings at rest and during visual stimulation were obtained from 20 channels using the 10/20 international system. The power spectrum of the 2nd and 4th harmonic of the stimulation frequency was calculated by Fast Fourier Transform (FFT) at a resolution of 0.25 Hz. Association of activity between occipital, temporal, parietal and central regions was measured by intra- and inter-hemispheric coherence and phase at harmonics of the stimulation frequency. A significant difference (P

The response properties to clicks, noise and tone bursts of 2152 single units located in the ventral division of the medial geniculate body were analysed as a function of their anatomical position. A particular spatial distribution of these properties was observed in the pars lateralis (LV) and ovoidea (OV). The distribution of different response characteristics changed along the rostro-caudal axis. Units located posteriorly were in majority either insensitive to simple acoustical stimuli or responded exclusively to pure tones, presenting generally a broad tuning and a loose tonotopic arrangement. Inhibitory response patterns were about as frequent as excitatory ones, response latencies were long on the average and widely distributed. Only a few units showed time-locking of their discharges in response to repetitive clicks. Most units had non-monotonic intensity functions. Going anteriorly, the distribution of response properties progressively changed: the number of units sensitive to various simple acoustical stimuli (pure tones and broad band stimuli together) increased, the tonotopic arrangement was more precise and more units were sharply tuned. Response patterns were in majority of the excitatory type, and latencies were shorter on the average and less dispersed. More units were precisely time-locked to repetitive clicks. The proportion of units with monotonie intensity functions increased. The origin of thalamo-cortical projections was studied with focal injections of wheat-germ agglutinin labeled with horseradish peroxidase in functionally defined loci of the various auditory cortical fields. An evolution of the density of labeled cells in LV and OV was observed along the same rostro-caudal axis for which a gradient of functional properties is described above. Thalamo-cortical projections to the primary auditory area and the anterior auditory field originated predominantly from the anterior half of LV, whereas the posterior auditory field received inputs from a wider rostro-caudal extend of LV including its posterior half.

The discharge properties of 735 single units located in the pars magnocellularis (M) of the medial division of the medial geniculate body (MGB) were studied in 23 nitrous oxide anesthetized cats in response to simple acoustic stimuli (clicks, noise and tone bursts). A systematic decrease of single unit characteristic frequencies (CF) was observed along electrode track portions crossing M from dorso-medial to ventro-lateral. These data indicate that M is tonotopically organized with an arrangement of low CF units latero-ventrally and high CF units dorso-medially. This preferential arrangement of single units as a function of their CF was consistent with the location and orientation of clusters of labeled cells in M resulting from wheat-germ agglutinin labeled with horseradish peroxidase (WGA-HRP) injections in CF defined loci in the anterior (AAF) or primary (AI) auditory cortical fields. The quality of the tonotopic arrangement was low caudally and increased in the rostral direction, indicating that this tonotopicity concerns mainly the anterior half of M. Response latencies to clicks, noise and tone bursts were on average longer in the posterior part of M than in its anterior part. Time-locking of discharges in response to repetitive acoustic pulses was more frequent anteriorly than posteriorly and the upper limiting rate of locking was on average higher rostrally (up to 200-300 Hz). In contrast, other response properties such as responsiveness to the various combinations of simple acoustic stimuli, response patterns and tuning were more randomly distributed in M, showing the whole range of response properties seen in the MGB. Data derived from several injections of WGA-HRP performed in distinct auditory cortical fields in several animals indicated that M projects to the tonotopic cortical fields (AAF, AI and PAF) as well as to the non-tonotopically organized secondary auditory cortex (AII). The contribution of M to the total thalamic input reaching each field of the auditory cortex was quantitatively more important for AAF (30%) and PAF (20%) than for AI and AII (about 10% each).