The diversity across populations and the variability between individuals have long posed a significant challenge in cognitive science. Although large language models (LLMs) have made notable progress in aligning with human values, faithfully capturing the high degree of diversity and uncertainty in human judgment remains an unresolved challenge.This study investigates whether computational models, or `proxy agents," can not only emulate human decision patterns but also systematically modulate them. We propose a framework wherein we first fine-tune BERT-based proxy agents to replicate both aggregate and individual-level human judgments on a large-scale moral dilemma dataset. We then hypothesize that stimuli identified as maximally divisive for these individualized agents will similarly elicit high disagreement among human participants. Through a manipulating experiment, we validate this hypothesis, demonstrating that agent-selected stimuli can predictably induce targeted divergence in human moral choices. Our findings provide empirical evidence that AI agents can bias human perceptual variability by strategically filtering information. We further analyze this induced moral divergence using a Bayesian framework and concept decomposition to identify the distinct conceptual dimensions driving individual differences. This work quantifies the potential for AI-driven cognitive modulation and underscores the urgent need for ethical guidelines to prevent the misuse of such capabilities.

In-context learning (ICL) has proven to be adept at adapting large language models (LLMs) to downstream tasks without parameter updates, based on a few demonstration examples. Prior work has found that the ICL performance is susceptible to the selection of examples in prompt and made efforts to stabilize it. However, existing example selection studies ignore the ethical risks behind the examples selected, such as gender and race bias. In this work, we conduct extensive experiments and discover that (1) example selection with high accuracy does not mean low bias; (2) example selection for ICL may amplify the biases of LLMs; (3) example selection contributes to spurious correlations of LLMs. Based on the above observations, we propose the Remind with Bias-aware Embedding (ReBE), which removes the spurious correlations through contrastive learning and obtains bias-aware embedding for LLMs based on prompt tuning. Finally, we demonstrate that ReBE effectively mitigates biases of LLMs without significantly compromising accuracy and is highly compatible with existing example selection methods.

Higher-order brain connectivity (HOBC), which captures interactions among three or more brain regions, provides richer organizational information than traditional pairwise functional connectivity (FC). Recent studies have begun to infer latent HOBC from noninvasive imaging data, but they mainly focus on static analyses, limiting their applicability in dynamic prediction tasks. To address this gap, we propose DCHO, a unified approach for modeling and forecasting the temporal evolution of HOBC based on a decomposition–composition framework, which is applicable to both non-predictive tasks (state classification) and predictive tasks (brain dynamics forecasting). DCHO adopts a decomposition–composition strategy that reformulates the prediction task into two manageable subproblems: HOBC inference and latent trajectory prediction. In the inference stage, we propose a dual-view encoder to extract multiscale topological features and a latent combinatorial learner to capture high-level HOBC information. In the forecasting stage, we introduce a latent-space prediction loss to enhance the modeling of temporal trajectories. Extensive experiments on multiple neuroimaging datasets demonstrate that DCHO achieves superior performance in both non-predictive tasks (state classification) and predictive tasks (brain dynamics forecasting), significantly outperforming existing methods.

Yuxuan Li, Xinwei Guo, Jiashi Gao, Guanhua Chen, Xiangyu Zhao, Jiaxin Zhang, Quanying Liu, Haiyan Wu, Xin Yao, Xuetao Wei. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Meta-reinforcement learning requires utilizing prior task distribution information obtained during exploration to rapidly adapt to unknown tasks. The efficiency of an agent's exploration hinges on accurately identifying the current task. Recent Bayes-Adaptive Deep RL approaches often rely on reconstructing the environment's reward signal, which is challenging in sparse reward settings, leading to suboptimal exploitation. Inspired by bisimulation metrics, which robustly extracts behavioral similarity in continuous MDPs, we propose SimBelief—a novel meta-RL framework via measuring similarity of task belief in Bayes-Adaptive MDP (BAMDP). SimBelief effectively extracts common features of similar task distributions, enabling efficient task identification and exploration in sparse reward environments. We introduce latent task belief metric to learn the common structure of similar tasks and incorporate it into the real task belief. By learning the latent dynamics across task distributions, we connect shared latent task belief features with specific task features, facilitating rapid task identification and adaptation. Our method outperforms state-of-the-art baselines on sparse reward MuJoCo and panda-gym tasks.

Decoding visual stimuli from neural recordings is a critical challenge in the development of brain-computer interfaces (BCIs). Although recent EEG-based decoding approaches have made progress in tasks such as visual classification, retrieval, and reconstruction, they remain constrained by unstable representation learning and a lack of interpretability. This gap highlights the need for more efficient representation learning and the integration of effective language interaction to enhance both understanding and practical usability in visual decoding tasks. To address this limitation, we introduce RealMind, a novel EEG-based framework designed to handle a diverse range of downstream tasks. Specifically, RealMind leverages both semantic and geometric consistency learning to enhance feature representation and improve alignment across tasks. Notably, beyond excelling in traditional tasks, our framework marks the first attempt at visual captioning from EEG data through vision-language model (VLM). It achieves a Top-1 decoding accuracy of 27.58% in a 200-class zero-shot retrieval task and a BLEU-1 score of 26.59% in a 200-class zero-shot captioning task. Overall, RealMind provides a comprehensive multitask EEG decoding framework, establishing a foundational approach for EEG-based visual decoding in real-world applications.

Human decision-making in cognitive tasks and daily life exhibits considerable variability, shaped by factors such as task difficulty, individual preferences, and personal experiences. Understanding this variability across individuals is essential for uncovering the perceptual and decision-making mechanisms that humans rely on when faced with uncertainty and ambiguity. We propose a systematic Boundary Alignment Manipulation (BAM) framework for studying human perceptual variability through image generation. BAM combines perceptual boundary sampling in ANNs and human behavioral experiments to systematically investigate this phenomenon. Our perceptual boundary sampling algorithm generates stimuli along ANN perceptual boundaries that intrinsically induce significant perceptual variability. The efficacy of these stimuli is empirically validated through large-scale behavioral experiments involving 246 participants across 116,715 trials, culminating in the variMNIST dataset containing 19,943 systematically annotated images. Through personalized model alignment and adversarial generation, we establish a reliable method for simultaneously predicting and manipulating the divergent perceptual decisions of pairs of participants. This work bridges the gap between computational models and human individual difference research, providing new tools for personalized perception analysis. Code and data for this work are publicly available.

A fundamental challenge in affective cognitive science is to develop models that accurately capture the relationship between external emotional stimuli and human internal experiences. While ANNs have demonstrated remarkable accuracy in facial expression recognition, their ability to model inter-individual differences in human perception remains underexplored. This study investigates the phenomenon of high perceptual variability—where individuals exhibit significant differences in emotion categorization even when viewing the same stimulus. Inspired by the similarity between ANNs and human perception, we hypothesize that facial expression samples that are ambiguous for ANN classifiers also elicit divergent perceptual judgments among human observers. To examine this hypothesis, we introduce a novel perceptual boundary sampling method to generate facial expression stimuli that lie along ANN decision boundaries. These ambiguous samples form the basis of the varEmotion dataset, constructed through large-scale human behavioral experiments. Our analysis reveals that these ANN-confusing stimuli also provoke heightened perceptual uncertainty in human participants, highlighting shared computational principles in emotion perception. Finally, by fine-tuning ANN representations using behavioral data, we achieve alignment between ANN predictions and both group-level and individual-level human perceptual patterns. Our findings establish a systematic link between ANN decision boundaries and human perceptual variability, offering new insights into personalized modeling of emotional interpretation.

Understanding how the human brain encodes and processes external visual stimuli has been a fundamental challenge in neuroscience. With advancements in artificial intelligence, sophisticated visual decoding architectures have achieved remarkable success in fMRI research, enabling more precise and fine-grained spatial concept localization. This has provided new tools for exploring the spatial representation of concepts in the brain. However, despite the millisecond-scale temporal resolution of EEG, which offers unparalleled advantages in tracking the dynamic evolution of cognitive processes, the temporal dynamics of neural representations based on EEG remain underexplored. This is primarily due to EEG’s inherently low signal-to-noise ratio and its complex spatiotemporal coupling characteristics. To bridge this research gap, we propose a novel approach that integrates advanced neural decoding algorithms to systematically investigate how low-dimensional object properties are temporally encoded in EEG signals. We are the first to attempt to identify the specificity and prototypical temporal characteristics of concepts within temporal distributions. Our framework not only enhances the interpretability of neural representations but also provides new insights into visual decoding in brain-computer interfaces (BCI).

Text-to-Image (T2I) diffusion models exhibit concerning tendencies to generate harmful imagery that perpetuates social biases and stereotypes, posing significant ethical risks in real-world applications. While existing mitigation approaches predominantly employ black-box methodologies through dataset augmentation or constrained fine-tuning, they face critical limitations, including high data acquisition costs and potential exacerbation of stereotypes during model retraining. Inspired by neuroscience principles where neurological dysfunction often stems from aberrant neural activation patterns, we propose a novel framework, StereoClinic, targeting the root cause of stereotype generation through direct neural intervention. Our solution introduces two synergistic components: Diffusion Deep Taylor Decomposition (DDTD) for precisely localizing stereotype-related neurons via Layer-wise Relevance Propagation (LRP) attribution analysis, and Stereotype Neuron Suppression (SNS) implementing targeted activation damping to neutralize bias propagation. Through extensive empirical evaluations across multiple bias dimensions, we demonstrate that our method achieves significant stereotype mitigation without compromising image quality or requiring additional training data. This neuro-inspired approach establishes a new paradigm for model interpretability and ethical alignment in generative AI systems.

Brain atlases are essential for reducing the dimensionality of neuroimaging data and enabling interpretable analysis. However, most existing atlases are predefined, group-level templates with limited flexibility and resolution. We present Deep Cluster Atlas (DCA), a graph-guided deep embedding clustering framework for generating individualized, voxel-wise brain parcellations. DCA combines a pretrained voxel-level fMRI autoencoder with spatially regularized deep clustering to produce functionally coherent and spatially contiguous regions. Our method supports flexible control over resolution and anatomical scope, and generalizes to arbitrary brain structures. We further introduce a standardized benchmarking platform for atlas evaluation, using multiple large-scale fMRI datasets. Across multiple datasets and scales, DCA outperforms state-of-the-art atlases, improving functional homogeneity by 98.8% and silhouette coefficient by 29%, and achieves superior performance in downstream tasks. Furthermore, atlases demonstrate heterogeneous performance across various tasks and an atlas derived from a fine-tuned model yields superior results for its specific application. Codes are available at https://github.com/ncclab-sustech/DCA.

Task-specific pre-training is essential when task representations diverge from generic pre-training features. Existing task-general pre-training EEG models struggle with complex tasks like emotion recognition due to mismatches between task-specific features and broad pre-training approaches. This work aims to develop a task-specific multi-dataset joint pre-training framework for cross-dataset emotion recognition, tackling problems of large inter-dataset distribution shifts, inconsistent emotion category definitions, and substantial inter-subject variability. We introduce a cross-dataset covariance alignment loss to align second-order statistical properties across datasets, enabling robust generalization without the need for extensive labels or per-subject calibration. To capture the long-term dependency and complex dynamics of EEG, we propose a hybrid encoder combining a Mamba-like linear attention channel encoder and a spatiotemporal dynamics model. Our method outperforms state-of-the-art large-scale EEG models by an average of 4.57% in AUROC for few-shot emotion recognition and 11.92% in accuracy for zero-shot generalization to a new dataset. Performance scales with the increase of datasets used in pre-training. Multi-dataset joint pre-training achieves a performance gain of 8.55\% over single-dataset training. This work provides a scalable framework for task-specific pre-training and highlights its benefit in generalizable affective computing. Our code is available at https://github.com/ncclab-sustech/mdJPT_nips2025.

Human sensorimotor control is remarkably fast and accurate at the system level despite severe speed-accuracy trade-offs at the component level. The discrepancy between the contrasting speed-accuracy trade-offs at these two levels is a paradox. Meanwhile, speed accuracy trade-offs, heterogeneity, and layered architectures are ubiquitous in nerves, skeletons, and muscles, but they have only been studied in isolation using domain-specific models. In this article, we develop a mechanistic model for how component speed-accuracy trade-offs constrain sensorimotor control that is consistent with Fitts' law for reaching. The model suggests that diversity among components deconstrains the limitations of individual components in sensorimotor control. Such diversity-enabled sweet spots (DESSs) are ubiquitous in nature, explaining why large heterogeneities exist in the components of biological systems and how natural selection routinely evolves systems with fast and accurate responses using imperfect components.

Electroencephalogram (EEG)-based emotion decoding can objectively quantify people's emotional state and has broad application prospects in human-computer interaction and early detection of emotional disorders. Recently emerging deep learning architectures have significantly improved the performance of EEG emotion decoding. However, existing methods still fall short of fully capturing the complex spatiotemporal dynamics of neural signals, which are crucial for representing emotion processing. This study proposes a Dynamic-Attention-based EEG State Transition (DAEST) modeling method to characterize EEG spatiotemporal dynamics. The model extracts spatiotemporal components of EEG that represent multiple parallel neural processes and estimates dynamic attention weights on these components to capture transitions in brain states. The model is optimized within a contrastive learning framework for cross-subject emotion recognition. The proposed method achieved state-of-the-art performance on three publicly available datasets: FACED, SEED, and SEED-V. It achieved$81.7\pm 4.3\%$accuracy in the binary classification of positive and negative emotions and$67.9\pm 7.3\%$in nine-class discrete emotion classification on the FACED dataset,$88.1\pm 3.6\%$in the three-class classification of positive, negative, and neutral emotions on the SEED dataset, and$73.6\pm 12.7\%$in five-class discrete emotion classification on the SEED-V dataset. The learned EEG spatiotemporal patterns and dynamic transition properties offer valuable insights into neural dynamics underlying emotion processing.

How to decode human vision through neural signals has attracted a long-standing interest in neuroscience and machine learning. Modern contrastive learning and generative models improved the performance of visual decoding and reconstruction based on functional Magnetic Resonance Imaging (fMRI). However, the high cost and low temporal resolution of fMRI limit their applications in brain-computer interfaces (BCIs), prompting a high need for visual decoding based on electroencephalography (EEG). In this study, we present an end-to-end EEG-based visual reconstruction zero-shot framework, consisting of a tailored brain encoder, called the Adaptive Thinking Mapper (ATM), which projects neural signals from different sources into the shared subspace as the clip embedding, and a two-stage multi-pipe EEG-to-image generation strategy. In stage one, EEG is embedded to align the high-level clip embedding, and then the prior diffusion model refines EEG embedding into image priors. A blurry image also decoded from EEG for maintaining the low-level feature. In stage two, we input both the high-level clip embedding, the blurry image and caption from EEG latent to a pre-trained diffusion model. Furthermore, we analyzed the impacts of different time windows and brain regions on decoding and reconstruction. The versatility of our framework is demonstrated in the magnetoencephalogram (MEG) data modality. The experimental results indicate that our EEG-based visual zero-shot framework achieves SOTA performance in classification, retrieval and reconstruction, highlighting the portability, low cost, and high temporal resolution of EEG, enabling a wide range of BCI applications. Our code is available at https://github.com/ncclab-sustech/EEG_Image_decode.

Post-hoc explanation techniques on graph neural networks (GNNs) provide economical solutions for opening the black-box graph models without model retraining. Many GNN explanation variants have achieved state-of-the-art explaining results on a diverse set of benchmarks, while they rarely provide theoretical analysis for their inherent properties and explanatory capability. In this work, we propose $\underline{\text{S}}$tructure-$\underline{\text{A}}$ware Shapley-based $\underline{\text{M}}$ultipiece $\underline{\text{E}}$xplanation (SAME) method to address the structure-aware feature interactions challenges for GNNs explanation. Specifically, SAME leverages an expansion-based Monte Carlo tree search to explore the multi-grained structure-aware connected substructure. Afterward, the explanation results are encouraged to be informative of the graph properties by optimizing the combination of distinct single substructures. With the consideration of fair feature interactions in the process of investigating multiple connected important substructures, the explanation provided by SAME has the potential to be as explainable as the theoretically optimal explanation obtained by the Shapley value within polynomial time. Extensive experiments on real-world and synthetic benchmarks show that SAME improves the previous state-of-the-art fidelity performance by 12.9\% on BBBP, 7.01\% on MUTAG, 42.3\% on Graph-SST2, 38.9\% on Graph-SST5, 11.3\% on BA-2Motifs and 18.2\% on BA-Shapes under the same testing condition. Code is available at https://github.com/same2023neurips/same.

Electrical neuromodulation as a palliative treatment has been increasingly used in the control of epilepsy. However, current neuromodulations commonly implement predetermined actuation strategies and lack the capability of self-adaptively adjusting stimulation inputs. In this work, rooted in optimal control theory, we propose a Koopman-MPC framework for real-time closed-loop electrical neuromodulation in epilepsy, which integrates i) a deep Koopman operator based dynamical model to predict the temporal evolution of epileptic electroencephalogram (EEG) with an approximate finite-dimensional linear dynamics and ii) a model predictive control (MPC) module to design optimal seizure suppression strategies. The Koopman operator based linear dynamical model is embedded in the latent state space of the autoencoder neural network, in which we can approximate and update the Koopman operator online. The linear dynamical property of the Koopman operator ensures the convexity of the optimization problem for subsequent MPC control. The proposed deep Koopman operator model shows greater predictive capability than the baseline models (e.g., vector autoregressive model, kernel based method and recurrent neural network (RNN)) in both synthetic and real epileptic EEG data. Moreover, compared with the RNN-MPC framework, our Koopman-MPC framework can suppress seizure dynamics with better computational efficiency in both the Jansen-Rit model and the Epileptor model. Koopman-MPC framework opens a new window for model-based closed-loop neuromodulation and sheds light on nonlinear neurodynamics and feedback control policies.

Nonunion is one of the challenges faced by orthopedics clinics for the technical difficulties and high costs in photographing interosseous capillaries. Segmenting vessels and filling capillaries are critical in understanding the obstacles encountered in capillary growth. However, existing datasets for blood vessel segmentation mainly focus on the large blood vessels of the body, and the lack of labeled capillary image datasets greatly limits the methodological development and applications of vessel segmentation and capillary filling. Here, we present a benchmark dataset, named IFCIS-155, consisting of 155 2D capillary images with segmentation boundaries and vessel fillings annotated by biomedical experts, and 19 large-scale, high-resolution 3D capillary images. To obtain better images of interosseous capillaries, we leverage state-of-the-art immunofluorescence imaging techniques to highlight the rich vascular morphology of interosseous capillaries. We conduct comprehensive experiments to verify the effectiveness of the dataset and the benchmarking deep learning models (e.g. UNet/UNet++ and the modified UNet/UNet++). Our work offers a benchmark dataset for training deep learning models for capillary image segmentation and provides a potential tool for future capillary research. The IFCIS-155 dataset and code are all publicly available at https://github.com/ncclabsustech/IFCIS-55.

The oxytocin effects on large-scale brain networks such as Default Mode Network (DMN) and Frontoparietal Network (FPN) have been largely studied using fMRI data. However, these studies are mainly based on the statistical correlation or Bayesian causality inference, lacking interpretability at the physical and neuroscience level. Here, we propose a physics-based framework of the Kuramoto model to investigate oxytocin effects on the phase dynamic neural coupling in DMN and FPN. Testing on fMRI data of 59 participants administrated with either oxytocin or placebo, we demonstrate that oxytocin changes the topology of brain communities in DMN and FPN, leading to higher synchronization in the FPN and lower synchronization in the DMN, as well as a higher variance of the coupling strength within the DMN and more flexible coupling patterns at group level. These results together indicate that oxytocin may increase the ability to overcome the corresponding internal oscillation dispersion and support the flexibility in neural synchrony in various social contexts, providing new evidence for explaining the oxytocin modulated social behaviors. Our proposed Kuramoto model-based framework can be a potential tool in network neuroscience and offers physical and neural insights into phase dynamics of the brain.

Objective: Diagnosis of the severity of heroin addiction with electroencephalography (EEG) signals is a challenging problem. It has been shown that brain microstates are associated with brain status and healthy condition. However, there is no study on how heroin addiction affects brain microstates. Approach: We propose a hybrid classifier based on the microstate features, extracting from resting state EEGs, to objectively and effectively identify abstinent heroin-addicted individuals (AHAIs) and healthy controls (HCs). In addition to the commonly used features such as duration, occurrence, and transition, we calculated three new features. Main Results: The results showed that the support vector machine (SVM), which allows classification of the AHAIs and HCs with a 73% accuracy rate, was an optimal classifier. Moreover, the weight setting-based genetic algorithm (GA) further improved the accuracy rate to 81%. The hybrid classification not only provides direct evidence showing the differences in EEG microstate features between AHAIs and HCs, but also offers a method to distinguish the heroin brain states of people addicted to heroin and healthy individuals and demonstrates that microstate features could serve as potential bio-markers for identifying AHAIs. Significance: our methods and the selected features may provide electrophysiological insights for the assessment of the heroin withdrawal treatment effects.

The recorded electroencephalography (EEG) signals are usually contaminated by many artifacts. In recent years, deep learning models have been used for denoising of electroencephalography (EEG) data and provided comparable performance with that of traditional techniques. However, the performance of the existing networks in electromyograph (EMG) artifact removal was limited and suffered from the over-fitting problem. Here we introduce a novel convolutional neural network (CNN) with gradually ascending feature dimensions and downsampling in time series for removing muscle artifacts in EEG data. Compared with other types of convolutional networks, this model largely eliminates the over-fitting and significantly outperforms four benchmark networks in EEGdenoiseNet. Our study suggested that the deep network architecture might help avoid overfitting and better remove EMG artifacts in EEG.

EEG source localization is an important technical issue in EEG analysis. Despite many numerical methods existed for EEG source localization, they all rely on strong priors and the deep sources are intractable. Here we propose a deep learning framework using spatial basis function decomposition for EEG source localization. This framework combines the edge sparsity prior and Gaussian source basis, called Edge Sparse Basis Network (ESBN). The performance of ESBN is validated by both synthetic data and real EEG data during motor tasks. The results suggest that the supervised ESBN outperforms the traditional numerical methods in synthetic data and the unsupervised fine-tuning provides more focal and accurate localizations in real data. Our proposed deep learning framework can be extended to account for other source priors, and the real-time property of ESBN can facilitate the applications of EEG in brain-computer interfaces and clinics.

It has been reported that chronic heroin intake induces changes in central nervous system of human brain; however, few studies investigate the carry-over adverse effects on brain after heroin withdrawal. In this work we examined the alpha rhythms of resting-state Electroencephalogram (EEG) signals to measure the neuroelectrical differences between the heroin addicts after heroin withdrawal and normal control. Eyes-closed resting EEG signals from 20 heroin addicts with the abstinence length ranging from 4-16 months and 20 normal controls were recorded using 64 electrodes. Comparing the nonlinear characteristics of EEG signals, such as the correlation dimension, Kolmogorov entropy and Lempel-Ziv complexity, we found that the EEG signals from heroin addicts were significantly more irregular than those from normal controls. Furthermore, the topography of the each nonlinear feature was examined, and the abnormal changes were widely spread over the brain. These findings suggest that nonlinear methods may contribute to gain new insights into brain dysfunction in heroin addicts even after heroin abstinence.

With the booms of mobile communication, especially mobile smart phone, technologies to identify individuals for mobile security calls for some more strict requirements in user-friendly, real-time and ubiquitous aspects. In addition to traditional approaches (for example, password check), some advanced biometric methodologies have been applied in practice, such as fingerprint and iris based solutions, however, these solutions generally lack a true ubiquitous nature for mobile security. In this paper, we present a real time EEG based individual identification interface to support ubiquitous applications. The EEG signals are collected through a mono-polar single channel in real time via a mobile EEG device. An experiment involving about 20 subjects has been conducted to evaluate the interface. The experiment comprises three types of tests: accuracy test, time dimension test and capacity dimension test. The results of these experiments demonstrate that our approach is highly suitable to the demands of mobile security in ubiquitous environment. In addition, we integrate this interface into scenarios of ubiquitous application - Online Predictive Tools for Intervention in Mental Illness (OPTIMI).