We present a novel benchmarking methodology for Large Language Models (LLMs) to evaluate their susceptibility to hallucinations, thereby determining their reliability for real-world applications involving greater responsibilities. This method, called Deception-Based Benchmarking, involves testing the model with a task that requires composing a short paragraph. Initially, the model performs under standard conditions. Then, it is required to begin with a misleading sentence. Based on these outputs, the model is assessed on three criteria: accuracy, susceptibility, and consistency. This approach can be integrated with existing benchmarks or applied to new ones, thus facilitating a comprehensive evaluation of models across multiple dimensions. It also encompasses various forms of hallucination. We applied this methodology to several small opensource models using a modified version of MMLU, DB-MMLU1 . Our findings indicate that most current models are not specifically designed to self-correct when the random sampling process leads them to produce inaccuracies. However, certain models, such as Solar-10.7B-Instruct, exhibit a reduced vulnerability to hallucination, as reflected by their susceptibility and consistency scores. These metrics are distinct from traditional benchmark scores. Our results align with TruthfulQA, a widely used benchmark for hallucination. Looking forward, DB-benchmarking can be readily applied to other benchmarks to monitor the advancement of LLMs.

Schizophrenia is characterized by disturbances in both perception and expression that profoundly impair social functioning, yet these domains are typically studied in isolation. Predictive processing theories suggest that such symptoms may arise from abnormal updating of internal models, but the neural mechanisms linking perceptual and expressive dysfunction remain unclear. Using magnetoencephalography during multisensory perception and motor tasks, we tested whether beta-band activity (15-30 Hz), a neural signal implicated in predictive control, provides a shared substrate across domains. Compared to healthy individuals, patients with schizophrenia showed attenuated event-related modulation of beta activity, including weaker beta suppression during sensory processing and delayed or reduced post-movement beta rebound. These abnormalities were associated with a widened audiovisual temporal binding window, indicating atypical multisensory integration. Multivariate analyses further revealed that reduced beta modulation across sensory and motor systems covaried with impoverished semantic diversity, simplified syntactic structure in natural speech, and greater clinical symptom burden. Notably, beta abnormalities emerged as distributed latent patterns spanning sensory, motor, and frontotemporal regions. Together, these findings identify diminished event-related beta modulation as a common neural signature linking disrupted multisensory integration, action monitoring, and language organization in schizophrenia. We situate perceptual and expressive impairments within a unified framework of predictive dysfunction and advance a mechanistically grounded account that highlights beta dynamics as a promising target for future mechanistic and translational studies in psychosis. Significance Statement: Schizophrenia disrupts both how people perceive the world and how they express their thoughts, yet these symptoms are usually studied separately. Using magnetoencephalography during multisensory and motor tasks, we identify a shared neural mechanism linking these domains: reduced event-related modulation of beta-band (15–30 Hz) activity, a signal implicated in predictive control. Compared to healthy individuals, patients showed diminished beta changes during sensory processing and movement completion, suggesting reduced flexibility in updating internal models. Importantly, this neural pattern covaried with abnormal audiovisual binding, disorganized natural speech, and greater clinical severity. These findings reveal distributed beta modulation as a cross-domain neural marker of predictive dysfunction, offering a unified framework for understanding perceptual and expressive disturbances in schizophrenia.

Background: Thought disorder in schizophrenia reflects disrupted internal model maintenance in predictive processing. Beta oscillations have been proposed to index top-down predictive signalling, but prior resting-state studies have focused on spectral power, overlooking transient burst dynamics that enable the stabilization and reinstatement of internal representations, especially in the default mode network (DMN). Methods: Resting-state MEG was acquired from 25 patients (7 females) with schizophrenia and 25 matched controls (8 females) between May 2024 and October 2025. Beta bursts (15-30 Hz) were extracted from source-localized cortical time series, and burst rate, duration, power, and inter-regional synchrony were quantified. Associations with computationally-derived semantic and syntactic speech features were assessed using principal-component regression within patients. Results: Conventional beta power did not differ between groups after correction. In contrast, beta-burst dynamics showed significant group-by-anterior-posterior-position interactions for burst rate (p = 0.018), power (p = 0.0014) and duration (p < 0.001). Within the DMN, patients showed shorter inferior parietal burst duration (t(46.0) = -2.74, FDR-p = 0.0343) and increased burst synchrony within left prefrontal cortex (t(31.0) = 2.81, FDR-p = 0.0343) and between prefrontal cortices (t(29.8) = 2.73, FDR-p = 0.0423). Within patients, burst components predicted speech organization (R-squared = 0.47, F(3,19) = 5.66, p = 0.006), driven by the inferior parietal burst-duration component (β = -0.55, p = 0.0036), but did not predict overall clinical severity. Limitations: The modest patient sample, short resting-state recording, cross-sectional design, and lack of laminar specificity limit causal and mechanistic inference; burst-speech associations require replication. Conclusion: Schizophrenia was associated with dissociable anterior and posterior beta-burst abnormalities that tracked objective speech markers of thought disorder. These findings suggest impaired hierarchical coordination of predictive beta signalling as a candidate mechanism linking intrinsic network dysfunction to disorganized thought.

Wilder Penfield (1891–1976), founder of the Montreal Neurological Institute and pioneer of the Montreal Procedure, made landmark contributions to neurology and neurosurgery. However, his concurrent advancement of the ethical framework governing awake craniotomy is less examined. In an era when medical paternalism remained common, Penfield and his colleagues upheld patient-centered care, multidisciplinary collaboration, procedural standardization and rigorous boundaries between clinical practice and research. Through primary sources, including publications, case records, correspondences and biological accounts, this historical review traces how Penfield’s ethical commitments shaped the modern practice of awake craniotomy and established an enduring legacy relevant to contemporary neurosurgery.

Schizophrenia disrupts both how people perceive the world and how they express and act on their thoughts, yet these symptoms are usually studied separately. Using magnetoencephalography during multisensory and motor tasks, we tested whether reduced event-related beta-band (15 to 30 hertz) modulation, a signal implicated in predictive control, provides a candidate neural marker across these domains. Compared to healthy individuals, patients showed diminished beta changes during sensory processing and movement completion, suggesting reduced flexibility in updating internal models. This neural pattern covaried with audiovisual simultaneity-judgment behavior, disorganized natural speech, and greater clinical severity. These findings suggest that distributed beta modulation may serve as a candidate cross-domain marker associated with perceptual, motor, and language disturbances in schizophrenia, motivating future studies that directly test predictive updating mechanisms.