Cognitive Neurodynamics最新文献

Sleep is an essential part of human life, and the quality of one’s sleep is also an important indicator of one’s health. Analyzing the Electroencephalogram (EEG) signals of a person during sleep makes it possible to understand the sleep status and give relevant rest or medical advice. In this paper, a decent amount of artificial data generated with a data augmentation method based on Discrete Cosine Transform from a small amount of real experimental data of a specific individual is introduced. A classification model with an accuracy of 92.85% has been obtained. By mixing the data augmentation with the public database and training with the EEGNet, we obtained a classification model with significantly higher accuracy for the specific individual. The experiments have demonstrated that we can circumvent the subject-independent problem in sleep EEG in this way and use only a small amount of labeled data to customize a dedicated classification model with high accuracy.

Temporal interference deep-brain magnetic stimulation (TI-DMS) induces rhythmic electric field (EF) in the hippocampus to normalize cognitive function. The rhythmic time series of the hippocampal EF is essential for the assessment of TI-DMS. However, the finite element method (FEM) takes several hours to obtain the time series of EF. In order to reduce the time cost, the temporal convolutional network (TCN) model is adopted to predict the time series of hippocampal EF induced by TI-DMS. It takes coil configuration and loaded current as input and predicts the time series of maximum and mean values of the left and right hippocampal EF. The prediction takes only a few seconds. The model parameter combination of kernel size and layers is selected optimally by cross-validation method. The experimental results for multiple subjects show that the R² of all the time series predicted by the model exceed 0.98. And the prediction accuracy is even higher as the input parameters approach the training set. These results demonstrate that the adopted model can quickly predict the time series of hippocampal EF induced by TI-DMS with relatively high accuracy, which is beneficial for future clinical applications.

Abstract

Inspired by the motion processing pathway, this paper proposes a bio-inspired feedforward spiking network model based on Hodgkin–Huxley neurons for human motion perception. The proposed network mimics the mechanisms of direction selectivity found in simple and complex cells of the primary visual cortex. Simple cells' receptive fields are modeled using Gabor energy filters, while complex cells' receptive fields are constructed by integrating the responses of simple cells in an energy model. To generate the motion map, the spiking output of the network integrates motion information encoded by the responses of complex cells with various preferred directions. Simulation results demonstrate that the spiking neuron-based network effectively replicates the directional selectivity operation of the visual cortex when presented with a sequence of time-varying images. We evaluate the proposed model against state-of-the-art spiking neuron-based motion detection models using publicly available datasets. The results highlight the model's capability to extract motion energy from diverse video sequences, akin to human visual motion perception models. Additionally, we showcase the application of the proposed model in motion segmentation tasks and compare its performance with state-of-the-art motion-based segmentation models using challenging video segmentation benchmarks. The results indicate competitive performance. The motion maps generated by the proposed model can be utilized for action recognition in input videos.

The role of network metrics in exploring brain networks of mental illness is crucial. This study focuses on quantifying a node controllability index (CA-scores) and developing a novel framework for studying the dysfunction of attention deficit hyperactivity disorder (ADHD) brains. By analyzing fMRI data from 143 healthy controls and 102 ADHD patients, the controllability metric reveals distinct differences in nodes (brain regions) and subsystems (functional modules). There are significantly atypical CA-scores in the Rolandic operculum, superior medial orbitofrontal cortex, insula, posterior cingulate gyrus, supramarginal gyrus, angular gyrus, precuneus, heschl gyrus, and superior temporal gyrus of ADHD patients. A comparison with measures of connection strength, eigenvector centrality, and topology entropy suggests that the controllability index may be more effective in identifying abnormal regions in ADHD brains. Furthermore, our controllability index could be extended to investigate functional networks associated with other psychiatric disorders.

The ability to coactivate (or “superpose”) multiple conceptual representations is a fundamental function that we constantly rely upon; this is crucial in complex cognitive tasks requiring multi-item working memory, such as mental arithmetic, abstract reasoning, and language comprehension. As such, an artificial system aspiring to implement any of these aspects of general intelligence should be able to support this operation. I argue here that standard, feed-forward deep neural networks (DNNs) are unable to implement this function, whereas an alternative, fully brain-constrained class of neural architectures spontaneously exhibits it. On the basis of novel simulations, this proof-of-concept article shows that deep, brain-like networks trained with biologically realistic Hebbian learning mechanisms display the spontaneous emergence of internal circuits (cell assemblies) having features that make them natural candidates for supporting superposition. Building on previous computational modelling results, I also argue that, and offer an explanation as to why, in contrast, modern DNNs trained with gradient descent are generally unable to co-activate their internal representations. While deep brain-constrained neural architectures spontaneously develop the ability to support superposition as a result of (1) neurophysiologically accurate learning and (2) cortically realistic between-area connections, backpropagation-trained DNNs appear to be unsuited to implement this basic cognitive operation, arguably necessary for abstract thinking and general intelligence. The implications of this observation are briefly discussed in the larger context of existing and future artificial intelligence systems and neuro-realistic computational models.

Waveform transitions have high correlation to spike wave discharges and polyspike wave discharges in seizure dynamics. This research adopts nonlinear dynamics to study the waveform transitions in a cerebral thalamo-coritcal neural network subjected to a square sensory control via discretization and mappings. The continuous non-smooth network outputs are discretized to establish implicit mapping chains or loops for stable and unstable waveform solutions. Bifurcation trees of period-1 to period-2 waveforms as well as independent bifurcation tree of period-3 to period-6 waveforms are obtained theoretically. The independent bifurcation tree should be taken much care during the control since it coexists with global stable waveforms but contains more spikes. Stability and bifurcations of the nonlinear waveform transitions are predicted by eigenvalue analysis of the discretized model. The transient process from unstable waveform to stable waveform is illustrated. The spike adding and period-doubling phenomenon are presented for illustration of the network response after control. The dominant frequency components and the detailed quantity levels of the corresponding amplitudes are exhibited in the harmonic spectrums which can be implemented to controller design for reduction and elimination of the absence seizures. This research presents new perspectives for the waveform transitions and provides theories and data for seizure prediction and regulation.

The functional neurons are basic building blocks of the nervous system and are responsible for transmitting information between different parts of the body. However, it is less known about the interaction between the neuron and the field. In this work, we propose a novel functional neuron by introducing a flux-controlled memristor into the FitzHugh-Nagumo neuron model, and the field effect is estimated by the memristor. We investigate the dynamics and energy characteristics of the neuron, and the stochastic resonance is also considered by applying the additive Gaussian noise. The intrinsic energy of the neuron is enlarged after introducing the memristor. Moreover, the energy of the periodic oscillation is larger than that of the adjacent chaotic oscillation with the changing of memristor-related parameters, and same results is obtained by varying stimuli-related parameters. In addition, the energy is proved to be another effective method to estimate stochastic resonance and inverse stochastic resonance. Furthermore, the analog implementation is achieved for the physical realization of the neuron. These results shed lights on the understanding of the firing mechanism for neurons detecting electromagnetic field.

Abstract

As the basis of musical emotions, dynamic tension experience is felt by listeners as music unfolds over time. The effects of musical harmonic and melodic structures on tension have been widely investigated, however, the potential roles of metrical structures in tension perception remain largely unexplored. This experiment examined how different metrical structures affect tension experience and explored the underlying neural activities. The electroencephalogram (EEG) was recorded and subjective tension was rated simultaneously while participants listened to music meter sequences. On large time scale of whole meter sequences, it was found that different overall tension and low-frequency (1 ~ 4 Hz) steady-state evoked potentials were elicited by metrical structures with different periods of strong beats, and the higher overall tension was associated with metrical structure with the shorter intervals between strong beats. On small time scale of measures, dynamic tension fluctuations within measures was found to be associated with the periodic modulations of high-frequency (10 ~ 25 Hz) neural activities. The comparisons between the same beats within measures and across different meters both on small and large time scales verified the contextual effects of meter on tension induced by beats. Our findings suggest that the overall tension is determined by temporal intervals between strong beats, and the dynamic tension experience may arise from cognitive processing of hierarchical temporal expectation and attention, which are discussed under the theoretical frameworks of metrical hierarchy, musical expectation and dynamic attention.

The mesial temporal lobe epilepsy (MTLE) seizures are believed to originate from medial temporal structures, including the amygdala, hippocampus, and temporal cortex. Thus, the seizures onset zones (SOZs) of MTLE locate in these regions. However, whether the neural features of SOZs are specific to different medial temporal structures are still unclear and need more investigation. To address this question, the present study tracked the features of two different high frequency oscillations (HFOs) in the SOZs of these regions during MTLE seizures from 10 drug-resistant MTLE patients, who received the stereo electroencephalography (SEEG) electrodes implantation surgery in the medial temporal structures. Remarkable difference of HFOs features, including the proportions of HFOs contacts, percentages of HFOs contacts with significant coupling and firing rates of HFOs, could be observed in the SOZs among three medial temporal structures during seizures. Specifically, we found that the amygdala might contribute to the generation of MTLE seizures, while the hippocampus plays a critical role for the propagation of MTLE seizures. In addition, the HFOs firing rates in SOZ regions were significantly larger than those in NonSOZ regions, suggesting the potential biomarkers of HFOs for MTLE seizure. Moreover, there existed higher percentages of SOZs contacts in the HFOs contacts than in all SEEG contacts, especially those with significant coupling to slow oscillations, implying that specific HFOs features would help identify the SOZ regions. Taken together, our results displayed the features of HFOs in different medial temporal structures during MTLE seizures, and could deepen our understanding concerning the neural mechanism of MTLE.

In this paper, the exponential synchronization of quaternion-valued memristor-based Cohen–Grossberg neural networks with time-varying delays is discussed. By using the differential inclusion theory and the set-valued map theory, the discontinuous quaternion-valued memristor-based Cohen–Grossberg neural networks are transformed into an uncertain system with interval parameters. A novel controller is designed to achieve the control goal. With some inequality techniques, several criteria of exponential synchronization for quaternion-valued memristor-based Cohen–Grossberg neural networks are given. Different from the existing results using decomposition techniques, a direct analytical approach is used to study the synchronization problem by introducing an improved one-norm method. Moreover, the activation function is less restricted and the Lyapunov analysis process is simpler. Finally, a numerical simulation is given to prove the validity of the main results.