European Journal of Neuroscience最新文献

Neural signals such as EEG, ECoG, and intracortical recordings offer a valuable window into brain dynamics but remain difficult to decode due to high dimensionality, nonstationarity, and substantial interindividual variability. Traditional machine learning and deep learning models often show limited generalizability and insufficient interpretability in these settings. Foundation models (FMs)—large-scale architectures pretrained on diverse datasets—have recently emerged as a promising paradigm for building robust, transferable, and physiologically grounded neural representations. Among these modalities, EEG currently serves as the most practical and representative platform for FM development due to its large-scale open datasets, standardized protocols, and broad clinical applicability, while the same conceptual framework remains generalizable to other neural recording types. This review synthesizes emerging FM approaches for neural decoding and critically examines representative EEG-based architectures. We highlight three essential design principles: physiology-aware representation learning that captures oscillatory and dynamic structure, structure-aware architectures that incorporate spatial and anatomical priors, and interpretability mechanisms that ensure neuroscientific and clinical validity. Although models such as the Patched Brain Transformer, CBraMod, and BrainGPT demonstrate encouraging adaptability, many still inherit objectives from non-neural domains and underutilize spatial priors such as electrode topology or functional connectivity. While this review focuses on EEG as the most data-rich and scalable testbed, the same framework can extend to ECoG and intracortical recordings to support unified neural representations across spatial and temporal scales. Fully realizing the potential of neural FMs will require biologically informed objectives, structure-aware architectures, interpretable representations, and standardized data ecosystems.

Type 3 diabetes, a term that underscores the pathological connection between Type 2 diabetes mellitus (T2DM) and Alzheimer's disease (AD), emphasizes the role of cerebral insulin resistance and metabolic dysfunction in neuron degeneration. Cerebral insulin resistance impairs Akt signaling, leading to the suppression of GSK-3β activity, enhancement of tau hyperphosphorylation and neurofibrillary tangle formation. Concurrent impaired GLUT-3/4 translocation, mitochondrial function, and amyloid-β clearance exacerbate oxidative stress and plaque deposition. Collectively, these disruptions compromise synaptic plasticity and cognition, accelerating AD progression. Moreover, activation of AGE–RAGE–NF-κB signaling amplifies neuroinflammation, further aggravating tau and Aβ pathology. This interaction activates downstream MAPK, ERK1/2, and JNK/STAT pathways, which in turn stimulate transcription factors such as NF-κB, TGF-β, HIF-1α, and AP-1. The resulting cascade promotes oxidative stress, neuroinflammation, and PI3K/Akt/IRS-1 signaling impairment. Together, these interconnected pathways accelerate neuronal loss and cognitive decline. Emerging evidence indicates that natural bioactive compounds offer therapeutic benefits in AD by attenuating AGE–RAGE–mediated oxidative stress, neuroinflammation, and cerebral insulin resistance, thereby reducing amyloid-β accumulation and tau hyperphosphorylation. This review highlights the AGE–RAGE axis as a critical molecular mediator connecting T2DM and AD, orchestrating neuroinflammation, mitochondrial dysfunction, and tau hyperphosphorylation. Therapeutic strategies aimed at inhibiting AGE formation or blocking RAGE activation represent promising approaches to attenuate cognitive decline associated with Type 3 diabetes.

Motor adaptation is a form of motor learning that enables the updating of motor commands in response to sensory inputs, requiring computations at the cerebellar level that must be integrated into cerebral cortical networks for their implementation. We proposed that cerebellar-cortical integration, which underlies motor adaptation, is related to the modulation of frequency-specific oscillatory activity. We examined motor error and electrophysiological correlates (power spectrum and phase locking value analysis) measured during different sessions of transcranial alternating stimulation (tACS) delivered to the cerebellum at relevant frequencies (50 Hz, 20 Hz, or sham). We found that 50 Hz tACS, but not 20 Hz or sham stimulation, reduced movement error, especially in initial practice trials. Electroencephalography (EEG) analysis revealed modulation of spectral power and phase synchrony (wPLI) in frontal, parietal, and occipital regions, with specific patterns for both the frequency range and the task stage. Power and wPLI modulation under fifty Hz stimulation were associated with the magnitude of motor adaptation. Our findings suggest that frequency-specific neural oscillations play a crucial role in the effective integration between the cerebellum and cortical regions of the brain. Significance: Our data indicate that cerebellar tACS at approximately 50 Hz may serve as an effective neuromodulation strategy to enhance motor adaptation in humans, with specific neural correlates.

We aimed to test a neurocognitive model positing that auditory hallucinations arise from a hyperactive left temporoparietal cortex and a hypoactive left dorsolateral prefrontal cortex. Specifically, 104 healthy individuals completed a signal detection task that induces auditory false perceptions while receiving transcranial direct current stimulation (tDCS). All participants were tested twice, once with real and once with sham tDCS. However, in half of the participants, the electrode montage mimicked the hypertemporal/hypofrontal model (anode over temporoparietal cortex and cathode over dorsolateral prefrontal cortex). In the other half, electrodes were reversed as in treatments of auditory verbal hallucinations. We hypothesized that these montages would lead to increased and decreased rates of auditory false perceptions, respectively. Moreover, we expected that auditory false perceptions would be triggered more often when participants expected certain words (top-down effect) and when these words were masked by human noise (bottom-up effect). The results revealed robust top-down and bottom-up interactions, but they were neither modulated by electrode montage nor by whether real or sham tDCS was administered. The results thus support the notion that auditory verbal hallucinations arise from an interaction of bottom-up and top-down processes but do not support the hypertemporal/hypofrontal model. Possible explanations for the lack of tDCS effects are discussed, including the idea that the tDCS electrodes might not have stimulated the targeted areas.

Continuous interactions between respiration and cortical rhythms influence how individuals adaptively perceive their environment. However, the intricate relationship among respiration, cortical rhythms, and perceptual function remains elusive. Breathing is a flexible process that can be partially voluntarily controlled. This study leveraged this unique characteristic to probe how brain-respiration interactions coordinate visual perception. Participants performed normal- or slow-paced breathing while discriminating fearful and neutral expressions presented during the midpoint of ongoing inspiration or expiration. The behavioral results showed that compared to normal-paced breathing, perceptual sensitivity was decreased during expiration but increased during inspiration during slow-paced breathing. The corroborating cortical dynamics, as detected by magnetoencephalography, identified perceptual sensitivity-related neural correlates that mirrored the behavioral results. These correlates emerged in the prestimulus period and were predominantly manifested within the alpha/beta frequency range. Subsequent analyses suggest that this observed pace-dependent effect is mediated across a hierarchy of time-frequency scales. First, compared to normal-paced breathing, slow-paced breathing reduces prestimulus theta phase coherence due to diminished respiration-brain phase synchronization. Consequently, the theta phase becomes weakly entrained, thereby preserving resources for long-range communication with response-related networks during slow-paced expiration rather than inspiration, as evidenced by enhanced phase-power coupling. In turn, this cortical adjustment differentially modulates response-related power and ultimately produces decreased or increased perceptual sensitivity during slow-paced expiration or inspiration, respectively. Our overall results elucidate how voluntarily controlled breathing pace may represent a top-down mechanism that influences visual perception.

Sleep dysfunction (SD) is common in Alzheimer's disease (AD) and associated with cognitive impairment. The hypothalamus regulates circadian rhythms and exhibits AD pathology; we investigated the abnormal alterations in the structure and function of the hypothalamus in AD patients with sleep dysfunction (ADSD). Twelve ADSD patients, 19 AD patients without sleep dysfunction (ADNSD), and 13 age- and sex-matched healthy controls (HCs) underwent neuropsychological assessments, including the Montreal Cognitive Assessment and Mini-Mental State Examination, and sleep quality was evaluated using the Pittsburgh Sleep Quality Index (PSQI). Hypothalamic volume and its subregional volumes were derived from T1-weighted magnetic resonance imaging (MRI), whereas resting-state functional MRI was used to assess functional connectivity (FC) with the hypothalamus. We found reduced volumes in the bilateral anterior inferior and left anterior superior hypothalamic subregions in both AD groups compared to HCs. Compared to HCs, the ADNSD group exhibited enhanced hypothalamic FC with the left middle frontal gyrus (MFG) and right inferior parietal lobule (IPL) and reduced FC with right MFG, whereas the ADSD group exhibited enhanced FC with the left MFG and right superior frontal gyrus and reduced FC with right IPL. Furthermore, the ADSD group exhibited reduced FC with the right IPL and enhanced FC with right MFG relative to the ADNSD group. Crucially, within the combined AD cohort, FC with the right IPL was negatively correlated with PSQI, whereas FC with the right MFG was positively correlated. These findings indicate the hypothalamus is a critical target for interventions to improve sleep quality and cognition in AD.