Stimulus-response Data Research Articles

Characterizing preferential infiltration of loess using geostatistical electrical resistivity tomography

Preferential infiltration is prevalent in loess and is pivotal in disasters such as landslides. However, the inherent multi-scale heterogeneity of loess makes traditional in-situ monitoring techniques challenging for characterizing the preferential infiltration process. This study employed the Geostatistical Electrical Resistivity Tomography (GERT) to examine the spatial distribution of mass water content (ωs) in loess strata to understand the preferential infiltration processes in loess. This study improved stimulus-response data quality and used GERT to characterize the spatial-temporal distribution of electrical conductivity (σ) and estimate ωs through the σ-ωs relationship. This study indicates that the surface loess layer has higher ωs than deeper layers, rapidly declining as depth increases. Under preferential infiltration, early-stage water forms a spherical saturated zone, transforming into an ellipsoid as it descends. Using equivalent homogeneous parameters as prior information effectively improved σ estimation. This study found a 15.9% error in the total change in water content based on the GERT survey compared to the known amount of injected water. It explores the possible impacts of the uncertainty of the unknown relationship between σ and ωs, its spatial variability, and the accuracy of the survey on the error. The formation of an electric double-layer structure within the loess as the saturation increases, which reduces the sensitivity of GERT to changes in water content, is likely another cause. Finally, these areas are essential for advancing future studies on applying geophysical tools to accurately estimate water distributions in loess formations.

Robot-aided assessment and associated brain lesions of impaired ankle proprioception in chronic stroke

BackgroundImpaired ankle proprioception strongly predicts balance dysfunction in chronic stroke. However, only sparse data on ankle position sense and no systematic data on ankle motion sense dysfunction in stroke are available. Moreover, the lesion sites underlying impaired ankle proprioception have not been comprehensively delineated. Using robotic technology, this study quantified ankle proprioceptive deficits post-stroke and determined the associated brain lesions.MethodsTwelve adults with chronic stroke and 13 neurotypical adults participated. A robot passively plantarflexed a participant’s ankle to two distinct positions or at two distinct velocities. Participants subsequently indicated which of the two movements was further/faster. Based on the stimulus-response data, psychometric just-noticeable-difference (JND) thresholds and intervals of uncertainty (IU) were derived as measures on proprioceptive bias and precision. To determine group differences, Welch’s t-test and the Wilcoxon-Mann-Whitney test were performed for the JND threshold and IU, respectively. Voxel-based lesion subtraction analysis identified the brain lesions associated with observed proprioceptive deficits in adults with stroke.Results83% of adults with stroke exhibited abnormalities in either position or motion sense, or both. JND and IU measures were significantly elevated compared to the control group (Position sense: + 77% in JND, + 148% in IU; Motion sense: +153% in JND, + 78% in IU). Adults with stroke with both impaired ankle position and motion sense had lesions in the parietal, frontal, and temporoparietal regions.ConclusionsThis is the first study to document the magnitude and frequency of ankle position and motion sense impairment in adults with chronic stroke. Proprioceptive dysfunction was characterized by elevated JND thresholds and increased uncertainty in perceiving ankle position/motion. Furthermore, the associated cortical lesions for impairment in both proprioceptive senses were largely overlapping.

Frequency-Dependent Multistep Ferroelectric Polarization Switching Mechanism in P(VDF-TrFE)-Based Capacitors Induced by Polystyrene Electret-like Modulation.

In this work, we investigate multistep ferroelectric polarization switching dynamics of a series of poly(vinylidene fluoride-trifluoroethylene)/polystyrene, P(VDF-TrFE)/PS, as active layers in ferroelectric capacitors with variable P(VDF-TrFE)/PS thickness ratios and a wide range of driving voltage frequencies (1-1000 Hz). The PS electret-like modulation effects on the depolarized field fluctuation are proven to be responsible for this multistep ferroelectric polarization switching process. To be specific, the switching current density peak splits into two peaks in both positive and negative voltage ranges according to the stimulus-response (S-R) data from the metal-ferroelectric-electret-metal capacitor driven by a periodic triangular voltage wave. The double-peak current trough appears when the transitorily suppressed ferroelectric polarization switching occurs while the discharge and recharge of the PS electret by external voltage brings a specific dynamic change in the electric field across ferroelectric (EFE). We also propose a theoretical model to simulate the ferroelectric polarization switching process at a current trough zone. This phenomenon provides new concepts on the electret-modulated multistep ferroelectric switching dynamics, and such switching mechanisms are critical for realizing reliable nonvolatile memory applications in flexible electronics.

Understanding the role of pathways in a deep neural network

Deep neural networks have demonstrated superior performance in artificial intelligence applications, but the opaqueness of their inner working mechanism is one major drawback in their application. The prevailing unit-based interpretation is a statistical observation of stimulus–response data, which fails to show a detailed internal process of inherent mechanisms of neural networks. In this work, we analyze a convolutional neural network (CNN) trained in the classification task and present an algorithm to extract the diffusion pathways of individual pixels to identify the locations of pixels in an input image associated with object classes. The pathways allow us to test the causal components which are important for classification and the pathway-based representations are clearly distinguishable between categories. We find that the few largest pathways of an individual pixel from an image tend to cross the feature maps in each layer that is important for classification. And the large pathways of images of the same category are more consistent in their trends than those of different categories. We also apply the pathways to understanding adversarial attacks, object completion, and movement perception. Further, the total number of pathways on feature maps in all layers can clearly discriminate the original, deformed, and target samples.

LATENT VARIABLE MODELING FOR BIOLOGICALLY INFORMED PROGNOSIS IN STUDIES OF PHYSICAL RESILIENCE

Abstract Physical resilience – rebound in relevant functioning and biomarkers following a health stressor – is hypothesized to be rooted in the level of fitness of stress-response physiology defining one’s “physiological resilience capacity” (PRC). This physiology forms a dynamical system comprising specific modules (the individual stress response systems and their underlying components) and their dynamic interactions with each other via feedback and other protocols. Such a system can be modeled using differential equations whose parameters may then define the PRC. We do not yet know how to measure these parameters directly, however. Rather, they are conceptually defined “constructs” which must be inferred using indirect measures—ideally, stimulus response data probing multiple aspects of the relevant physiology. Latent variable models are ideally suited to this setting. Two challenges for their application in studies of resilience are presented: (1) Integrating specific scientific knowledge on the dynamical systems in modeling the co-distribution of the indirect measures. (2) Synergizing such models’ advantages for construct measurement with advantages of machine learning approaches to optimize accuracy for predicting resilient outcomes by inferred PRC. Challenges and their proposed solutions are illustrated using multifaceted stimulus response data being collected in an ongoing investigation on the physiological basis of resilience to clinical stressors in older adults, the Study of Physical Resilience in agING. The work aims to produce physiologically rooted measures providing effective risk predictors for older adults facing impending stressors as well as intervention candidates by which to promote PRC in the longer term.

Cathodal Genesis of Ipsilateral Hand (Crossover) Motor Evoked Potentials: Evidence from a Patient with Previous Stroke

ABSTRACT A case is described where baseline transcranial electrical motor evoked potentials (TcMEP) and somatosensory evoked potentials (SSEP) results were unilaterally absent in a patient with previous hemispheric stroke undergoing a right-sided carotid endarterectomy. SSEP data confirmed right cortical pathology and excluded a technical rationale for absent motor evoked responses. Attempts at generating left-hand (contralateral) TcMEP from right cortical anodal stimulation failed despite high stimulus intensities. However, TcMEP responses from anodal stimulation of the right cortex were recorded from the right-hand (ipsilateral) which were attributed to “crossover.” Ipsilateral TcMEP onset latencies derived from the stimulus-response data supports the idea that crossover is a product of cathodal stimulation initially acting on pericortical motor pathways.

On modeling of a recurrent neural network from neural spiking data.

We present a theoretical and computational work, aiming at the estimation of firing rate based excitatory and inhibitory neural network from realistic stimulus-response data. The stimulus and response recordings are taken from a previous study which performs a measurement on the H1 neurons of the order Diptera flies. The parameter estimation is performed by maximum likelihood method. As the stimulus-response data is a single recording of 20 minutes, it is segmented and individual segments are superimposed on each other to increase the statistical content of information. The true values of the model parameters are unknown as we are not using synthetic data. Because of this fact, two sample Kolmogorov-Smirnov test is applied to compare the interspiking intervals of the recorded and model responses. Estimation and analysis results are presented in tabular and graphical forms. In addition, a comparison with previous research employing a modified Fitzhugh-Nagumo model is made.

Parametric Model Identification of Axisymmetric MEMS Resonators

This paper proposes a technique for developing parametric models of modally degenerate resonators from stimulus-response data. The technique complements traditional empirical frequency response estimates that are commonly used for testing MEMS resonators, however, the parametric models have distinct advantages when the modal frequency differences are close enough to frustrate estimates of quality factors, natural frequencies and mode orientations. The proposed technique also completely rejects parasitic coupling between the stimulus and pick-off electrodes. It is shown how a modification of the algorithm using demodulated measurement signals greatly reduces storage and computational requirements. [2020-0281]

Extracting the ON and OFF contributions to the full-field photopic flash electroretinogram using summed growth curves

Under cone-mediated (photopic) conditions, an “instantaneous” flash of light, including both stimulus onset and offset, will simultaneously activate both “ON” and “OFF” bipolar cells, which either depolarize (ON) or hyperpolarize (OFF) in response and, respectively, produce positive-going and negative-going deflections in the electroretinogram (ERG). The stimulus-response (SR) relationship of the photopic ON response demonstrates logistic growth, like that manifested in the rod-mediated (scotopic) b-wave, which is driven by a single class of depolarizing bipolar cell. However, the photopic b-wave SR function is importantly shaped by OFF responses, leading to a “photopic hill.” Furthermore, both on and off stimuli elicit activity in both ON and OFF bipolar cells. This has made it difficult to produce meaningful parameters for ready interpretation of the photopic b-wave SR relationship. Therefore, we evaluated whether the sum of sigmoidal SR functions, as descriptors of the depolarizing and hyperpolarizing components of the photopic flash ERG, could be used to elucidate and quantitate the mechanisms that produce the photopic hill. We used a novel fitting routine to optimize a sum of simple sigmoidal curves to SR data in five groups of subjects: Healthy adult, 10-week-old infant, congenital stationary night blindness (CSNB), X-linked juvenile retinoschisis (XJR), and preterm-born, both without and with a history of retinopathy of prematurity (ROP). Differences in ON and OFF amplitude, sensitivity, and implicit time among the groups were then compared using parameters extracted from these fits. We found that our modeling procedure enabled plausible derivations of ON and OFF pathway contributions to the ERG, and that the parameters produced appeared to have physiological relevance. In adult subjects, the ON and OFF amplitudes were similar in magnitude with respectively longer and shorter implicit times. Infant, CSNB, and XJR subjects showed significant ON pathway deficits. History of preterm-birth, without or with a diagnosis of ROP, did not much affect cone responses.

Fitting a recurrent dynamical neural network to neural spiking data: tackling the sigmoidal gain function issues

This is a continuation of a recent study (Doruk RO, Zhang K. Fitting of dynamic recurrent neural network models to sensory stimulus-response data. J Biol Phys 2018; 44: 449-469), where a continuous time dynamical recurrent neural network is fitted to neural spiking data. In this research, we address the issues arising from the inclusion of sigmoidal gain function parameters to the estimation algorithm. The neural spiking data will be obtained from the same model as that of Doruk and Zhang, but we propose a different model for identification. This will also be a continuous time recurrent neural network, but with generic sigmoidal gains. The simulation framework and estimation algorithms are kept similar to that of Doruk and Zhang so that we can have a solid base to compare the results. We evaluate the estimation performance in two different ways. First, we compare the firing rate responses of the original and the estimated model. We find that responses of both models to the same stimuli are similar. Secondly, we evaluate variations of the standard deviations of the estimates against a number of samples and stimulus parameters. They show a similar pattern to that of Doruk and Zhang. We thus conclude that our model serves as a reasonable alternative provided that firing rate is the response of interest (to any stimulus).

Adaptive Stimulus Design for Dynamic Recurrent Neural Network Models.

We present an adaptive stimulus design method for efficiently estimating the parameters of a dynamic recurrent network model with interacting excitatory and inhibitory neuronal populations. Although stimuli that are optimized for model parameter estimation should, in theory, have advantages over nonadaptive random stimuli, in practice it remains unclear in what way and to what extent an optimal design of time-varying stimuli may actually improve parameter estimation for this common type of recurrent network models. Here we specified the time course of each stimulus by a Fourier series whose amplitudes and phases were determined by maximizing a utility function based on the Fisher information matrix. To facilitate the optimization process, we have derived differential equations that govern the time evolution of the gradients of the utility function with respect to the stimulus parameters. The network parameters were estimated by maximum likelihood from the spike train data generated by an inhomogeneous Poisson process from the continuous network state. The adaptive design process was repeated in a closed loop, alternating between optimal stimulus design and parameter estimation from the updated stimulus-response data. Our results confirmed that, compared with random stimuli, optimally designed stimuli elicited responses with significantly better likelihood values for parameter estimation. Furthermore, all individual parameters, including the time constants and the connection weights, were recovered more accurately by the optimal design method. We also examined how the errors of different parameter estimates were correlated, and proposed heuristic formulas to account for the correlation patterns by an approximate parameter-confounding theory. Our results suggest that although adaptive optimal stimulus design incurs considerable computational cost even for the simplest excitatory-inhibitory recurrent network model, it may potentially help save time in experiments by reducing the number of stimuli needed for network parameter estimation.

Adaptive Cancellation of Parasitic Coupling

This paper reports a signal processing technique that adaptively identifies a finite-impulse-response (FIR) model of the parasitic coupling between the input ports and the output ports of a microelectromechanical resonator. The identified model is used as a feedforward filter to reduce the severity of the parasitic coupling by subtracting the filtered input port signal from the resonator’s output port signal. The compensated signals reveal the resonator’s motional response, which was previously obscured by the coupling, so that modal frequency and quality factor measurements can be performed. The experimental results also show that the adaptive filter tracks changes in the parasitic coupling during turn-on and warm-up periods. More than 30 dB of suppression of the parasitic coupling is achieved over a broad frequency band. The adaptive FIR filter is implemented on the signal processing equipment that is used to gather resonator stimulus-response data so no modifications to the resonator or its buffer electronics are required. [2018-0035]

Fitting of dynamic recurrent neural network models to sensory stimulus-response data.

We present a theoretical study aiming at model fitting for sensory neurons. Conventional neural network training approaches are not applicable to this problem due to lack of continuous data. Although the stimulus can be considered as a smooth time-dependent variable, the associated response will be a set of neural spike timings (roughly the instants of successive action potential peaks) that have no amplitude information. A recurrent neural network model can be fitted to such a stimulus-response data pair by using the maximum likelihood estimation method where the likelihood function is derived from Poisson statistics of neural spiking. The universal approximation feature of the recurrent dynamical neuron network models allows us to describe excitatory-inhibitory characteristics of an actual sensory neural network with any desired number of neurons. The stimulus data are generated by a phased cosine Fourier series having a fixed amplitude and frequency but a randomly shot phase. Various values of amplitude, stimulus component size, and sample size are applied in order to examine the effect of the stimulus to the identification process. Results are presented in tabular and graphical forms at the end of this text. In addition, to demonstrate the success of this research, a study involving the same model, nominal parameters and stimulus structure, and another study that works on different models are compared to that of this research.

Discriminative Learning of Receptive Fields from Responses to Non-Gaussian Stimulus Ensembles

Analysis of sensory neurons' processing characteristics requires simultaneous measurement of presented stimuli and concurrent spike responses. The functional transformation from high-dimensional stimulus space to the binary space of spike and non-spike responses is commonly described with linear-nonlinear models, whose linear filter component describes the neuron's receptive field. From a machine learning perspective, this corresponds to the binary classification problem of discriminating spike-eliciting from non-spike-eliciting stimulus examples. The classification-based receptive field (CbRF) estimation method proposed here adapts a linear large-margin classifier to optimally predict experimental stimulus-response data and subsequently interprets learned classifier weights as the neuron's receptive field filter. Computational learning theory provides a theoretical framework for learning from data and guarantees optimality in the sense that the risk of erroneously assigning a spike-eliciting stimulus example to the non-spike class (and vice versa) is minimized. Efficacy of the CbRF method is validated with simulations and for auditory spectro-temporal receptive field (STRF) estimation from experimental recordings in the auditory midbrain of Mongolian gerbils. Acoustic stimulation is performed with frequency-modulated tone complexes that mimic properties of natural stimuli, specifically non-Gaussian amplitude distribution and higher-order correlations. Results demonstrate that the proposed approach successfully identifies correct underlying STRFs, even in cases where second-order methods based on the spike-triggered average (STA) do not. Applied to small data samples, the method is shown to converge on smaller amounts of experimental recordings and with lower estimation variance than the generalized linear model and recent information theoretic methods. Thus, CbRF estimation may prove useful for investigation of neuronal processes in response to natural stimuli and in settings where rapid adaptation is induced by experimental design.

System Identification of Electro-Hydraulic Actuator System using ANFIS Approach

Precise control of an electro-hydraulic actuator (EHA) system has been an interesting subject due to its nonlinearities and uncertainties characteristics. Suitable controller can be designed when the precise model of the system is available. Adaptive Neuro-Fuzzy Inference System (ANFIS) modeling technique has proven can model various nonlinear systems at high accuracy. The objective of this paper is to obtain an ANFIS model from EHA system stimulus-response data with less complicated model structure and fewer system parameters. The validation of ANFIS model is done using various data sets which contain different operating region and limited data set, where data set is reduced to small operating region. Results show that ANFIS model can estimate the response nonlinear EHA system with more than 97% high best-fitting accuracy, with simple structure, under different operating region condition.

Sparse regularized regression identifies behaviorally-relevant stimulus features from psychophysical data

As a prerequisite to quantitative psychophysical models of sensory processing it is necessary to learn to what extent decisions in behavioral tasks depend on specific stimulus features, the perceptual cues. Based on relative linear combination weights, this study demonstrates how stimulus-response data can be analyzed in this regard relying on an L(1)-regularized multiple logistic regression, a modern statistical procedure developed in machine learning. This method prevents complex models from over-fitting to noisy data. In addition, it enforces "sparse" solutions, a computational approximation to the postulate that a good model should contain the minimal set of predictors necessary to explain the data. In simulations, behavioral data from a classical auditory tone-in-noise detection task were generated. The proposed method is shown to precisely identify observer cues from a large set of covarying, interdependent stimulus features--a setting where standard correlational and regression methods fail. The proposed method succeeds for a wide range of signal-to-noise ratios and for deterministic as well as probabilistic observers. Furthermore, the detailed decision rules of the simulated observers were reconstructed from the estimated linear model weights allowing predictions of responses on the basis of individual stimuli.

Construction of kinetic models for metabolic reaction networks: Lessons learned in analysing short-term stimulus response data

Construction of dynamic models of large-scale metabolic networks is one of the central issues in the engineering of living cells. However, construction of such models is often hampered by a number of challenges, for example, data availability, compartmentalization and parameter identification coupled with design of in vivo perturbations. As a solution to the latter, short-term perturbation experiments are proposed and are proven to be a useful experimental method to obtain insights into the in vivo kinetic properties of the metabolic pathways. The aim of this work is to construct a kinetic model using the available experimental data obtained by short-term perturbation experiments, where the steady state of a glucose-limited anaerobic chemostat culture of Saccharomyces cerevisiae was perturbed. In constructing the model, we first determined the steady-state flux distribution using the data before the glucose pulse and the known stoichiometry. For the rate expressions, we used approximative linlog kinetics, which allows the enzyme–metabolite kinetic interactions to be represented by an elasticity matrix. We performed a priori model reduction based on timescale analysis and parameter identifiability analysis allowing the information content of the experimental data to be assessed. The final values of the elasticities are estimated by fitting the model to the available short-term kinetic response data. The final model consists of 16 metabolites and 14 reactions. With 25 parameters, the model adequately describes the short-term response of the cells to the glucose perturbation, pointing to the fact that the assumed kinetic interactions in the model are sufficient to account for the observed response.

An Anthropomimetic Approach to High Performance Traction Control

Abstract The ability to learn and adapt to changing environmental conditions, as well as develop perceptive models based on stimulus-response data, provides expert human drivers with significant advantages. When it comes to bandwidth, accuracy, and repeatability, automatic control systems have clear advantages over humans; however, most high performance control systems lack many of the unique abilities of a human expert. This paper documents our first step toward the development of a novel automatic traction control algorithm using an anthropomimetic approach. The primary objective of this approach was to synthesize a high performance longitudinal traction control system by incorporating desirable human behavior distilled from human-in-the-loop (HIL) testing on a 6-DOF driving simulator. The proposed control algorithm was developed in a general framework, and applied to the specific task of longitudinal traction control. Simulation results confirm that the proposed anthropomimetic traction control algorithm provides improved performance relative to a well-tuned conventional PID-based traction control algorithm. Results are also compared with the HIL response data from a behavioral study.

Predictive feedback in human simulated pendulum balancing

In studies of human balance, it is common to fit stimulus-response data by tuning the time-delay and gain parameters of a simple delayed feedback model. Many interpret this fitted model, a simple delayed feedback model, as evidence that predictive processes are not required to explain existing data on standing balance. However, two questions lead us to doubt this approach. First, does fitting a delayed feedback model lead to reliable estimates of the time-delay? Second, can a non-predictive controller provide an explanation compatible with the independently estimated time delay? For methodological and experimental clarity, we study human balancing of a simulated inverted pendulum via joystick and screen. A two-step approach to data analysis is used: firstly a non-parametric model--the closed-loop impulse response--is estimated from the experimental data; second, a parametric model is fitted to the non-parametric impulse-response by adjusting time-delay and controller parameters. To support the second step, a new explicit formula relating controller parameters to closed-loop impulse response is derived. Two classes of controller are investigated within a common state-space context: non-predictive and predictive. It is found that the time-delay estimate arising from the second step is strongly dependent on which controller class is assumed; in particular, the non-predictive control assumption leads to time-delay estimates that are smaller than those arising from the predictive assumption. Moreover, the time-delays estimated using the non-predictive control assumption are not consistent with a lower-bound on the time-delay of the non-parametric model whereas the corresponding predictive result is consistent. Thus while the goodness of fit only marginally favoured predictive over non-predictive control, if we add the additional constraint that the model must reproduce the non-parametric time delay, then the non-predictive control model fails. We conclude (1) the time-delay should be estimated independently of fitting a low order parametric model, (2) that balance of the simulated inverted pendulum could not be explained by the non-predictive control model and (3) that predictive control provided a better explanation than non-predictive control.

Automated-parameterization of the motor evoked potential and cortical silent period induced by transcranial magnetic stimulation

To standardize the characterization of motor evoked potential (MEP) and cortical silent period (CSP) recordings elicited with transcranial magnetic stimulation (TMS). A computer-based, automated-parameterization program (APP) was developed and tested which provides a comprehensive set of electromyography (EMG) magnitude and temporal measures. The APP was tested using MEP, CSP, and isolated CSP (iCSP) TMS stimulus-response data from a healthy adult population (N=13). The APP had the highest internal reliability (Cronbach's alpha=.98) for CSP offset time compared with two prominent automated methods. The immediate post-CSP EMG recovery level was 49% higher than the pre-TMS EMG level. MEP size (peak amplitude, mean amplitude, peak-to-peak amplitude, and area) correlated higher with effective E-field (E(eff)) than other intensity measures (r approximately 0.5 vs. r approximately 0.3) suggesting that E(eff) is better suited for standardizing MEP stimulus-response relationships. The APP successfully characterized individual and mean epochs containing MEP, CSP, and iCSP responses. The APP provided common signal and temporal measures consistent with previous studies and novel additional parameters. With the use of the APP modeling method and the E(eff), a standard approach for the analysis and reporting of MEP-CSP complex and iCSP measurements is achievable.