A neural network model of speech acquisition and motor equivalent speech production

This article describes a neural network model that addresses the acquisition of speaking skills by infants and subsequent motor equivalent production of speech sounds. The model learns two mappings during a babbling phase. A phonetic-to-orosensory mapping specifies a vocal tract target for each speech sound; these targets take the form of convex regions in orosensory coordinates defining the shape of the vocal tract. The babbling process wherein these convex region targets are formed explains how an infant can learn phoneme-specific and language-specific limits on acceptable variability of articulator movements. The model also learns an orosensory-to-articulatory mapping wherein cells coding desired movement directions in orosensory space learn articulator movements that achieve these orosensory movement directions. The resulting mapping provides a natural explanation for the formation of coordinative structures. This mapping also makes efficient use of redundancy in the articulator system, thereby providing the model with motor equivalent capabilities. Simulations verify the model's ability to compensate for constraints or perturbations applied to the articulators automatically and without new learning and to explain contextual variability seen in human speech production.Supported in part by AFOSR F49620-92-J-0499     

A neurocomputational model for optimal temporal processing

Humans can estimate the duration of intervals of time, and psychophysical experiments show that these estimations are subject to timing errors. According to standard theories of timing, these errors increase linearly with the interval to be estimated (Weber's law), and both at longer and shorter intervals, deviations from linearity are reported. This is not easily reconciled with the accumulation of neuronal noise, which would only lead to an increase with the square root of the interval. Here, we offer a neuronal model which explains the form of the error function as a result of a constrained optimization process. The model consists of a number of synfire chains with different transmission times, which project onto a set of readout neurons. We show that an increase in the transmission time corresponds to a superlinear increase of the timing errors. Under the assumption of a fixed chain length, the experimentally observed error function emerges from optimal selection of chains for each given interval. Furthermore, we show how this optimal selection could be implemented by competitive spike-timing dependent plasticity in the connections from the chains to the readout network, and discuss implications of our model on selective temporal learning and possible neural architectures of interval timing.     

Laryngeal motor control in frogs: Role of vagal and laryngeal feedback

Using decerebrate frogs (Rana catesbeiana), we investigated the role of vagal and laryngeal sensory feedback in controlling motor activation of the larynx. Vagal and laryngeal nerve afferents were activated by electrical stimulation of the intact vagal and laryngeal nerves. Pulmonary afferents were activated by lung inflation. Reflex responses were recorded by measuring efferent activity in the laryngeal branch of the vagus (Xℓ) and changes in glottal aperture. Two glottic closure reflexes were identified, one evoked by lung inflation or electrical stimulation of the main branch of the vagus (Xm), and the other by electrical stimulation of Xℓ. Lung inflation evoked a decrementing burst of Xℓ efferent activity and electrical stimulation of Xm resulted in a brief burst of Xℓ action potentials. Electrical stimulation of Xℓ evoked a triphasic mechanical response, an abrupt glottal constriction followed by glottal dilatation followed by a long-lasting glottal constriction. The first phase was inferred to be a direct (nonreflex) response to the stimulus, whereas the second and third represent reflex responses to the activation of laryngeal afferents. Intracellular recordings of membrane potential of vagal motoneurons of lung and nonlung types revealed EPSPs in both types of neurons evoked by stimulation of Xm or Xℓ, indicating activation of glottal dilator and constrictor motoneurons. In summary, we have identified two novel reflexes producing glottic closure, one stimulated by activation of pulmonary receptors and the other by laryngeal receptors. The former may be part of an inspiratory terminating reflex and the latter may represent an airway protective reflex. © 1997 John Wiley & Sons, Inc. J Neurobiol 33: 213–222, 1997     

Electromyographic biofeedback for tension control during fine motor skill acquisition

The presence of residual muscular tension has been implicated as a detrimental influence on the performance and learning of motor skills. A method for reducing muscular tension has been provided by the advent of biofeedback training. This study investigated the effects of tension-control training by electromyographic (EMG) biofeedback on learning and performance of the pursuit-rotor backing task. Thirty young adult males were pretested for pursuit-rotor (PR) tracking skill, ranked by performance scores, and divided into identical triplicates to form two experimental groups and a control group. After a total of 3 hours of EMG biofeedback training for the experimental groups, all subjects were reevaluated on the PR test. One experimental group received biofeedback during the posttests. Analysis of variance of pretest-posttest difference means andt tests of scores representing performance and tension indicated that the EMG biofeedback training (1) significantly reduced tension induced by the novel motor skill and (2) significantly improved performance of the motor skill. Transfer of tension-control training was shown to facilitate learning and performance more than direct EMG biofeedback during performance. Residual tension reduction during learning was particularly facilitated by EMG biofeedback training, a profound implication for the management of stress in a variety of situations.This investigation formed part of a Ph.D. dissertation research (1976) conducted by the author under the guidance of Dr. Donald E. Campbell, Department of Physical Education, and Dr. Carol A. Saslow, Department of Psychology, at Oregon State University.     

A neurocomputational model of stimulus-specific adaptation to oddball and Markov sequences

Stimulus-specific adaptation (SSA) occurs when the spike rate of a neuron decreases with repetitions of the same stimulus, but recovers when a different stimulus is presented. It has been suggested that SSA in single auditory neurons may provide information to change detection mechanisms evident at other scales (e.g., mismatch negativity in the event related potential), and participate in the control of attention and the formation of auditory streams. This article presents a spiking-neuron model that accounts for SSA in terms of the convergence of depressing synapses that convey feature-specific inputs. The model is anatomically plausible, comprising just a few homogeneously connected populations, and does not require organised feature maps. The model is calibrated to match the SSA measured in the cortex of the awake rat, as reported in one study. The effect of frequency separation, deviant probability, repetition rate and duration upon SSA are investigated. With the same parameter set, the model generates responses consistent with a wide range of published data obtained in other auditory regions using other stimulus configurations, such as block, sequential and random stimuli. A new stimulus paradigm is introduced, which generalises the oddball concept to Markov chains, allowing the experimenter to vary the tone probabilities and the rate of switching independently. The model predicts greater SSA for higher rates of switching. Finally, the issue of whether rarity or novelty elicits SSA is addressed by comparing the responses of the model to deviants in the context of a sequence of a single standard or many standards. The results support the view that synaptic adaptation alone can explain almost all aspects of SSA reported to date, including its purported novelty component, and that non-trivial networks of depressing synapses can intensify this novelty response.     

A control model of human tongue movements in speech

Tongue movements during speech production have been investigated by means of a simple yet realistic biomechanical model, based on a finite elements modeling of soft tissues, in the framework of the equilibrium point hypothesis (-model) of motor control. In particular, the model has been applied to the estimation of the “central” control commands issued to the muscles, for a data set of mid-sagittal digitized tracings of vocal tract shape, r ecorded by means of low-intensity X-ray cineradiographies during speech. In spite of the highly non-linear mapping between the shape of the oral cavity and its acoustic consequences, the organization of control commands preserves the peculiar spatial organization of vowel phonemes in acoustic space. A factor analysis of control commands, which have been decomposed into independent or “orthogonal” muscle groups, has shown that, in spite of the great mobility of the tongue and the highly complex arrangement of tongue muscles, its movements can be explained in terms of the activation of a small number of independent muscle groups, each corresponding to an elementary or “primitive” movement. These results are consistent with the hypothesis that the tongue is controlled by a small number of independent “articulators”, for which a precise biomechanical substrate is provided. The influence of the effect of jaw and hyoid movements on tongue equilibrium has also bee n evaluated, suggesting that the bony structures cannot be considered as a moving frame of reference, but, indeed, there may be a substantial interaction between them and the tongue, that may only be accounted for by a “global” model. The reported results also define a simple control model for the tongue and, in analogy with similar modelling studies, they suggest that, because of the peculiar geometrical arrangement of tongue muscles, the central nervous system (CNS) may not need a de tailed representation of tongue mechanics but rather may make use of a relatively small number of muscle synergies, that are invariant over the whole space of tongue configurations. Received: 27 August 1996 / Accepted in revised form: 25 February 1997     

A neurocomputational model for the processing of conflicting information in context-dependent decision tasks

Context-dependent computation is a relevant characteristic of neural systems, endowing them with the capacity of adaptively modifying behavioral responses and flexibly discriminating between relevant and irrelevant information in a stimulus. This ability is particularly highlighted in solving conflicting tasks. A long-standing problem in computational neuroscience, flexible routing of information, is also closely linked with the ability to perform context-dependent associations. Here we present an extension of a context-dependent associative memory model to achieve context-dependent decision-making in the presence of conflicting and noisy multi-attribute stimuli. In these models, the input vectors are multiplied by context vectors via the Kronecker tensor product. To outfit the model with a noisy dynamic, we embedded the context-dependent associative memory in a leaky competing accumulator model, and, finally, we proved the power of the model in the reproduction of a behavioral experiment with monkeys in a context-dependent conflicting decision-making task. At the end, we discuss the neural feasibility of the tensor product and made the suggestive observation that the capacities of tensor context models are surprisingly in alignment with the more recent experimental findings about functional flexibility at different levels of brain organization.

     

A robotic model to investigate human motor control

The role of the mechanical properties of the neuromuscular system in motor control has been investigated for a long time in both human and animal subjects, mainly through the application of mechanical perturbations to the limb during natural movements and the observation of its corrective responses. These methods have provided a wealth of insight into how the central nervous system controls the limb. They suffer, however, from the fact that it is almost impossible to separate the active and passive components of the measured arm stiffness and that the measurement may themselves alter the stiffness characteristic of the arm. As a complement to these analyses, the implementation of a given neuroscientific hypothesis on a real mechanical system could overcome these measurement artifact and provide a tool that is, under full control of the experimenter, able to replicate the relevant functional features of the human arm. In this article, we introduce the NEURARM platform, a robotic arm intended to test hypotheses on the human motor control system. As such, NEURARM satisfies two key requirements. First, its kinematic parameters and inertia are similar to that of the human arm. Second, NEURARM mimics the main physical features of the human actuation system, specifically, the use of tendons to transfer force, the presence of antagonistic muscle pairs, the passive elasticity of muscles in the absence of any neural feedback and the non-linear elastic behaviour. This article presents the design and characterization of the NEURARM actuation system. The resulting mechanical behaviour, which has been tested in joint and Cartesian space under static and dynamic conditions, proves that the NEURARM platform can be exploited as a robotic model of the human arm, and could thus represent a powerful tool for neuroscience investigations.     

Cerebral areas associated with motor control of speech in humans

Murphy, K., D. R. Corfield, A. Guz, G. R. Fink, R. J. S. Wise, J. Harrison, and L. Adams. Cerebral areas associated withmotor control of speech in humans. J. Appl.Physiol. 83(5): 1438-1447, 1997.We have definedareas in the brain activated during speaking, utilizing positronemission tomography. Six normal subjects continuously repeated thephrase "Buy Bobby a poppy" (requiring minimal languageprocessing) in four ways: A) spoken aloud, B) mouthed silently,C) without articulation, andD) thought silently. Statisticalcomparison of images from conditions Awith C andB withD highlighted areas associated witharticulation alone, because control of breathing for speech wascontrolled for; we found bilateral activations in sensorimotor cortexand cerebellum with right-sided activation in the thalamus/caudate nucleus. Contrasting images from conditionsA with B andC with D highlighted areas associated withthe control of breathing for speech, vocalization, and hearing, becausearticulation was controlled for; we found bilateral activations insensorimotor and motor cortex, close to but distinct from theactivations in the preceding contrast, together with activations inthalamus, cerebellum, and supplementary motor area. In neithersubtraction was there activation in Broca's area. These resultsemphasize the bilaterality of the cerebral control of "speaking"without language processing.

     

Early language acquisition: cracking the speech code

Infants learn language with remarkable speed, but how they do it remains a mystery. New data show that infants use computational strategies to detect the statistical and prosodic patterns in language input, and that this leads to the discovery of phonemes and words. Social interaction with another human being affects speech learning in a way that resembles communicative learning in songbirds. The brain's commitment to the statistical and prosodic patterns that are experienced early in life might help to explain the long-standing puzzle of why infants are better language learners than adults. Successful learning by infants, as well as constraints on that learning, are changing theories of language acquisition.     

A physical model for motor proteins

A general stochastic theory is outlined for chemical to mechanical energy transduction by motor enzymes. In addition to ATP hydrolysis and fiber binding phenomena, thermal noise effects are taken into account. A minimal, 4-state model is identified that gives the hydrolysis rate as well as mechanical quantities such as sliding velocity and generated force, as functions of ATP concentration and the number of motors. It explains in a unified way many results of recent in vitro assays, both in myosins/actin and kinesins or dyneins/microtubule systems.     

A computer model of intermediate cerebellum dynamic operations in motor control

An intermediate cerebellum theoretical model for processing central programming discharges and muscle force signals is described which can perform a correct motor task under different peripheral perturbations (loads). An indispensable condition is that the simulated interpositus nucleus cells controlling a given effector (muscle) are inhibited by impulses coming from that effector (negative feedback from muscle force detectors). The hypothesis is proposed that the intermediate cerebellum can act via the rubrospinal tract as an interface between programming and executing motor structures.     

A statistical model providing comprehensive predictions for the mRNA differential display

MOTIVATION: Differential display (DD) or arbitrarily primed fingerprinting serves to identify differentially expressed genes, but these techniques cannot determine how many of the theoretically available genes have been uncovered. Previous mathematical models are unsatisfying as they are not suitable to analyze experimental data. RESULTS: In the present study, we provide a statistical model based on the redundancy of cDNA fragments amplified during DD experiments. This model is applicable to any DD and predicts (1) the total number of genes expressed in a sample cell type or tissue, (2) the number of differentially expressed genes, (3) the coverage obtained with any given number of primer combinations. In a DD experiment comparing two developmental stages of the post natal rat inner ear, we estimated the total number of differentially expressed genes accessible by DD to be 445, and the number of primer combinations required to uncover 90% of these to be 127. AVAILABILITY: The algorithms were implemented in Matlab (The Mathworks, Inc., Natick, MA) environment and are available at www.physiologie.uni-freiburg.de/download.html CONTACT: ellen.reisinger@physiologie.uni-freiburg.de.     

Chains of opportunity: A Markov model for acquisition of reusable resources

Summary A vacancy chain is a unique type of resource acquisition process composed of an interconnected series of events in which the gaining of a particular resource unit by one individual depends directly on prior acquisition events by other individuals. Taken from the sociological literature, vacancy chains may also describe the distribution of many types of animal resources such as burrows, dwellings and shelters. Using data on hermit crabs, we present a Markov model simulating a vacancy chain process, and test the model against field data. Our results show that a simple Markov model adequately describes shell acquisition in hermit crabs, and that models combining shell size and crude estimates of quality fit the data extremely well. We illustrate in detail how to generate vacancy chain models from ecological data, how to determine the number and size of organisms gaining new resource units from resource introductions of specific sizes, and how to statistically evaluate the accuracy of Markov models. Not recognizing the presence of a vacancy chain system may lead to serious errors in estimating resource dynamics and therefore in demographic and competition models based on these dynamics. Finally, we suggest some ways in which vacancy chain models can aid studies of competition, population dynamics, life histories, and conservation in species using this type of resource acquisition process.