Cognitive Neurodynamics最新文献

In this work, two capacitors connected by a thermistor are used to explore the electrical property of double-layer membrane in a neuron, which the membrane property is sensitive to changes of temperature and two capacitive variables are used to measure the potentials of inner and outer membrane. The circuit characteristics and energy definition for the neural circuit and its equivalent neuron model in oscillator form are clarified from physical aspect. Considering the shape deformation of cell membrane under external physical stimuli and energy injection, intrinsic parameters of the neuron can be controlled with adaptive growth under energy flow, an adaptive control law is proposed to regulate the firing modes accompanying with energy shift. In presence of noisy excitation, coherence resonance can be induced and confirmed by taming the noise intensity carefully. The distributions of CV (coefficient variability) and average energy value < H > vs. noise intensity provide a feasible way to predict the coherence resonance and even stochastic resonance in the neural activities. Adaptive parameter observers are designed to identify the unknown parameters in this neuron model. The research findings of this study lay a foundation for the design of temperature-adaptive biomimetic neuromorphic devices and the research on multi-functional perception neural networks with temperature sensitivity.

Electroencephalogram (EEG) signals are a popular tool to analyze sleep patterns. Cyclic alternating patterns (CAP) can be observed in EEG signals during unconscious periods of sleep. Detailed study of CAP can help in early diagnosis of many sleep disorders. Firstly, the CAP cycles need to be segregated into their constituents, phase A and phase B periods. In this work, we develop an accurate and easy-to-implement system to distinguish between the two CAP phases. The EEG signals are denoised and divided into smaller segments for an easier processing. These segments are decomposed into different frequency sub-bands using zero-phase filtering. Thereafter, statistical features are extracted from the sub-band components, and significant features are selected using the Kruskal-Wallis test. We consider four different algorithms for classification, namely, k-nearest neighbour (kNN), support vector machine (SVM), bagged tree (BT) and neural network (NN). The classification results are compiled for the datasets that include healthy subjects and those suffering from insomnia. The BT classifier produces the best results for the combined balanced dataset, with 83.29% accuracy and 83.58% F-1 score. The proposed method is more accurate and efficient than the existing schemes and can be considered for widespread deployments in real-world scenarios.

Although deep brain stimulation (DBS) is effective in treating Parkinson's disease (PD) related to bursting, the underlying mechanisms remain unclear. In the present paper, the dynamical and synaptic mechanisms are studied in a basal ganglia-thalamus model. Firstly, slow and large oscillations of synaptic gating variables/currents are identified as the cause of the irregular and non-synchronous bursting for PD, indicating that interruption of these slow modulations may be a feasible measure to treat PD. Secondly, strong DBS with high frequency applied to subthalamic nucleus (STN) can induce fast synchronous spiking in both STN and external globus pallidus (GPe), then interrupt the slow gating variables, thereby eliminating the irregular bursting. Meanwhile, the gating variables of the excitatory and inhibitory synapses respectively from STN and GPe to the internal globus pallidus (GPi) become fast. Finally, competition between these two opposite synapses can induce two manners to eliminate the bursting of GPi and restore the normal state, appearing in vast majority of parameter space composed of multiple synaptic conductances. One is the synchronous silence of GPi, and the other the synchronous regular fast spiking, which occurs for large conductance of the inhibitory and excitatory synapse, respectively. Both result in regular spiking of thalamus, via interrupting slow gating variables of synapse projected to thalamus. In addition, as the two conductances approach each other, the synaptic current to GPi oscillates around zero slowly, resulting in irregular firings of GPi and thalamus for PD in a narrow parameter space. Furthermore, the bursting observed in PD before DBS and three types of electrical activities of GPi during DBS are explained, using a saddle-node bifurcation of limit cycles and oscillation patterns of synaptic current. The distinction from the post inhibitory rebound bursting reported in previous studies is discussed. The results present the mechanisms for DBS to treat PD via eliminating bursting in wide parameter region.

Cognitive load refers to the mental effort required to process information and perform tasks, significantly influencing learning and performance outcomes. This paper presents a novel approach for cognitive load classification using a hybrid model that integrates Long Short-Term Memory (LSTM) networks with the Block Attention Module (BAM). Leveraging functional Near-Infrared Spectroscopy (fNIRS), we investigate the relationship between cognitive load and brain activity in a controlled experimental setting. Our methodology encompasses data collection from 50 participants engaged in various problem-solving tasks, with cognitive load categorized as high, medium, or low. The acquired fNIRS data underwent a rigorous preprocessing pipeline, including normalization and wavelet transform for feature extraction, enabling a comprehensive analysis of hemodynamic responses. The proposed model employs BAM to enhance feature representation by refining the importance of spatial and channel dimensions, thus improving the LSTM's ability to capture temporal dependencies in the data. The experimental results demonstrate significant performance improvements in cognitive load classification, showcasing the efficacy of the integrated LSTM-BAM architecture. This work not only contributes to the understanding of cognitive load dynamics but also highlights the potential of fNIRS as a non-invasive tool for real-time monitoring of cognitive performance, paving the way for advancements in instructional design and cognitive research.

Driver fatigue is a major contributor to traffic accidents, leading to increased fatality rates and severe damage compared to incidents involving alert drivers. Electroencephalography (EEG) has emerged as a widely used method for detecting driver fatigue due to its ability to capture brain activity patterns. This survey provides a thorough analysis of devices that detect driver fatigue using EEG, analyzing existing methodologies, challenges, and future research directions. This study was carried out according to PRISMA criteria. Relevant studies were retrieved from SpringerLink, Web of Science, IEEE Xplore, Scopus, and ScienceDirect, covering research published until February 16, 2025. After 267 publications were identified, 87 scientific papers were fully analyzed based on their relevance and contribution to the identification of driver fatigue using EEG. The review explores the article selection process, followed by an in-depth discussion of driver fatigue detection systems across various domains. Applications of Machine Learning (ML) in EEG-based fatigue evaluation are carefully reviewed, covering data collection, preliminary processing, feature extraction, categorization techniques, and performance assessment. Additionally, a comparative evaluation of cutting-edge research provides a comprehensive visualization of current research trends. This survey highlights the advantages, limitations, and future prospects of EEG-based driver fatigue detection, offering valuable insights for improving road safety. The findings contribute to the development of more reliable and real-time fatigue detection systems by addressing existing challenges and recommending potential solutions.

Accurately distinguishing stages of Alzheimer's disease (AD) is crucial for diagnosis and treatment. In this paper, we introduce a stacking classifier method that combines six single classifiers into a stacking classifier. Using brain network models and network metrics, we employ t-tests to identify abnormal brain regions, from which we construct a subnetwork and extract its features to form the training dataset. Our method is then applied to the ADNI (Alzheimer's Disease Neuroimaging Initiative) datasets, categorizing the stages into four categories: Alzheimer's disease, mild cognitive impairment (MCI), mixed Alzheimer's mild cognitive impairment (ADMCI), and healthy controls (HCs). We investigate four classification groups: AD-HCs, AD-MCI, HCs-ADMCI, and HCs-MCI. Finally, we compare the classification accuracy between a single classifier and our stacking classifier, demonstrating superior accuracy with our stacking classifier from a subnetwork-based viewpoint.

Decoding motor imagery electroencephalogram (MI-EEG) signals is fundamental to the development of brain-computer interface (BCI) systems. However, robust decoding remains a challenge due to the inherent complexity and variability of MI-EEG signals. This study proposes the Temporal Convolutional Attention Network (TCANet), a novel end-to-end model that hierarchically captures spatiotemporal dependencies by progressively integrating local, fused, and global features. Specifically, TCANet employs a multi-scale convolutional module to extract local spatiotemporal representations across multiple temporal resolutions. A temporal convolutional module then fuses and compresses these multi-scale features while modeling both short- and long-term dependencies. Subsequently, a stacked multi-head self-attention mechanism refines the global representations, followed by a fully connected layer that performs MI-EEG classification. The proposed model was systematically evaluated on the BCI IV-2a and IV-2b datasets under both subject-dependent and subject-independent settings. In subject-dependent classification, TCANet achieved accuracies of 83.06% and 88.52% on BCI IV-2a and IV-2b respectively, with corresponding Kappa values of 0.7742 and 0.7703, outperforming multiple representative baselines. In the more challenging subject-independent setting, TCANet achieved competitive performance on IV-2a and demonstrated potential for improvement on IV-2b. The code is available at https://github.com/snailpt/TCANet.

In recent years, emotion recognition, particularly EEG-based emotion recognition, has found widespread application across various domains. Enhancing EEG data processing and emotion recognition models remains a key research focus in this field. This paper presents an emotion recognition framework combining the CUSUM algorithm-based adaptive window selection technique with the convolutional attention-enhanced Kolmogorov-Arnold Networks (CA-KAN). The improved CUSUM algorithm effectively extracts the most emotion-relevant segments from raw EEG data. Furthermore, by enhancing the KAN network, the CA-KAN model achieves both high accuracy and efficiency in emotion recognition. The proposed framework achieved peak classification accuracies of 94.63% and 94.73% on the SEED and SEED-IV datasets, respectively. Additionally, the framework offers a lightweight advantage, demonstrating significant potential for real-world applications, including medical emotion monitoring and driver emotion detection.

This paper investigates the global exponential stability and periodicity of the Cohen-Grossberg neural network model with generalized piecewise constant delay. By applying Schaefer's fixed-point theorem, a sufficient condition for the existence of periodic solutions in the model is established. Additionally, by constructing appropriate differential inequalities with generalized piecewise constant delay, sufficient conditions for the global exponential stability of the model are obtained. Finally, computer simulations are conducted to illustrate a globally exponentially stable periodic Cohen-Grossberg neural network model, thereby confirming the feasibility and effectiveness of the proposed results.

The visual system has the ability to learn the statistical regularities (temporal and/or spatial) that characterize the visual scene automatically and implicitly. This ability is referred to as the visual statistical learning (VSL). The VSL could group several objects that have fixed statistical properties into a chunk. This complex process relies on the collaborative involvement of multiple brain regions that work together to learn the chunk. Although behavioral experiments have explored cognitive functions of the VSL, its computational mechanisms remain poorly understood. To address this issue, this study proposes a coupled shape-position recurrent neural network model based on the anatomical structure of the visual system to explain how chunk information is learned and represented in neural networks. The model comprises three core modules: the position network, which encodes object position information; the shape network, which encodes object shape information; and the decision network, which integrates the neuronal activity in the position and shape networks to make decisions. The model successfully simulates the results of a classic spatial VSL experiment. The distribution of neural firing rates in the decision network shows a significant difference between chunk and non-chunk conditions. Specifically, these neurons in the chunk condition exhibit stronger firing rates than those in the non-chunk condition. Furthermore, after the model learns a scene containing both chunk and non-chunk stimuli, neurons in the position network selectively encode far and near stimuli, respectively. In contrast, neurons in the shape network distinguish between chunk and non-chunk. The chunk encoding neurons selectively respond to specific chunks. These results indicate that the proposed model is able to learn spatial regularities of the stimuli to discriminate chunks from non-chunks, and neurons in the shape network selectively respond to chuck and non-chunk information. These findings offer important theoretical insights into the representation mechanisms of chunk information in neural networks and propose a new framework for modeling spatial VSL.