Representing "What" and "When" in Working Memory: A Computational Neuroscience Approach Using Laplace Neural Manifolds
Neurons in working memory must exhibit conjunctive receptive fields for stimulus identity ("What") and elapsed time ("When") in order to maintain a compositional representation of recent events. The dynamics of such a representation depend critically on the choice of temporal basis functions, with logarithmically-spaced basis functions providing a good match to empirical data.
Multi-modal Neuroimaging Integration via Cross-domain and Cross-modal Self-supervised Pre-training
The proposed Multi-modal Cross-domain Self-supervised Pre-training Model (MCSP) leverages self-supervised learning to synergize multi-modal information across spatial, temporal, and spectral domains, enabling effective fusion of fMRI and EEG data for improved analysis of brain disorders.
Decoding Past Visual Memories from Continuous Brain Activity: A Memory Disentangling Approach
The human brain can temporarily store and manipulate visual information, but how it encodes and retrieves recent visual memories during continuous visual processing remains unexplored. This study proposes a new task called "Memory Disentangling" to extract and decode past visual information from fMRI signals, mitigating the effects of proactive interference.
Mapping Causal Brain Interactions Using Virtual Perturbations of a Surrogate Neural Network
Neural Perturbational Inference (NPI) is a data-driven framework that non-invasively maps the directional, strength, and excitatory/inhibitory properties of effective connectivity across the entire human brain.
Efficient Learning of Sequences in Structured Recurrent Networks: A Biologically Plausible Model for Cortical Sequence Learning
A biologically plausible model for learning complex, non-Markovian sequences in recurrent networks of structured neurons, using a fully local, always-on plasticity rule that enables efficient and robust sequence learning.
자연스러운 편광각 자극에 대한 중추복합체 뉴런의 효율적인 부호화
중추복합체 뉴런의 활동은 현재 자극뿐만 아니라 이전 활동 이력에 의해서도 결정되며, 이는 빠른 방향 변화에 대한 반응을 촉진하고 직선 비행 중 에너지 소비를 줄일 수 있다.
Spiking History Shapes Dynamic Encoding of Polarization Angles in Central Complex Neurons of Bumblebees
Spiking history of polarization-sensitive neurons in the central complex of bumblebees facilitates faster responses to dynamic changes in polarization angles and reduces overall energy consumption during straight flight.
NeuroPath: A Deep Learning Model for Uncovering the Coupling Between Structural and Functional Brain Connectivity
NeuroPath is a novel deep learning model that integrates the coupling between structural and functional brain connectivity to enhance the prediction of cognitive states and uncover new insights into the complex relationship between brain structure and function.
Decoding Human Interaction Behaviors Using a Fuzzy-based Approach with Functional Near-Infrared Spectroscopy
A novel Fuzzy Attention Layer integrated within a Transformer Encoder model can effectively capture interpretable patterns of neural activity from fNIRS data to decode human social interaction behaviors, such as handholding.
Leveraging Differentiable Brain Simulation for Accurate Multi-scale Modeling of Brain Structure and Function
A novel workflow for integrating differentiable brain simulation, gradient-based optimization, and biologically-constrained neural network models to enable accurate multi-scale modeling of brain structure and function.
© 2024 by Linnk AI