Chris Rorden

Researchers in artificial intelligence increasingly intervene on language models to study how their functions are organized, silencing attention components in ways reminiscent of the brain lesions long used to map language in post-stroke aphasia. A common move is to ablate an attention head and read the resulting drop in a behavior as evidence that the head implements it. How much that establishes remains unclear, because a head whose removal disrupts a behavior is necessary for it but need not be the site where it is computed. We examined this question for picture naming, taking the loss of naming (anomia) that defines aphasia as the behavior of interest. Across six vision-language models spanning three language backbones, we ablated each attention head in turn during single-word picture naming and measured accuracy before and after. Localization varied widely. In LLaVA-1.6-Vicuna-13B, a single early head was necessary: removing it alone reduced naming accuracy from 0.99 to 0.006. It was not sufficient, because retaining it while ablating the other 1,599 heads also produced 0% accuracy. Two Mistral-backbone models had no critical head. Early-layer dependence appeared in every model but varied in strength, whereas dependence on any single head ranged from dominant to absent. Within Qwen2.5-VL, a dominant head appeared in the 7B but not the 3B model, indicating that this concentration emerged with scale. Ablating the head whose removal disrupts naming does not establish that the head computes the behavior, or that naming localizes to a head at all.

Interactive visualization is essential for collaborative analysis of complex neuroimaging data. Modern workflows span multiple computing environments including notebooks, integrated development environments, and standalone applications. All these workflows rely on visualization for quality control, exploration, and interpretation. However, visualization tools are typically siloed, forcing users to switch between environments and disrupting interactive analysis. Here, we address this challenge by integrating a shared visualization module directly into these diverse environments, providing a consistent visualization experience across formats, representations, and usage contexts. We developed a generalizable ecosystem of open-source tools built around the NiiVue rendering engine, supporting four primary use cases. First, we provide integrated file viewer capability for VS Code and JupyterLab. Second, we showcase three different architectures for standalone desktop applications. Third, we integrate NiiVue for Python notebooks, Julia, and R. Finally, we integrate NiiVue into traditional web applications. This unified ecosystem enables immediate, in-place inspection of neuroimaging data across the majority of modern computing platforms, eliminating the need to export data to external viewers. The tools support interactive inspection, dashboard creation, and the development of custom analysis applications within users’ existing workflows. The components are validated by automated continuous-integration testing, including unit tests and browser visualization tests against stored reference targets. By embedding visualization directly into commonly used computing environments, this ecosystem lowers the barrier to interactive data exploration and streamlines neuroimaging workflows. Visualization becomes an integral component of automated, reproducible, and shareable analysis pipelines, supporting more efficient collaboration and interpretation of complex neuroimaging data. By sharing a common rendering core, file format support, and interaction model, users can transfer visualization skills seamlessly across environments.

Lesion‐symptom mapping results can vary substantially as a function of specific analysis parameters, but the extent to which individual methodological choices interact to modulate the sensitivity and specificity of results is not clear. Here, we employed a large‐scale simulation approach to inform practical recommendations for lesion‐symptom mapping studies. Routine clinical imaging from 959 stroke survivors (mean age = 72.5, 49.3% female) was used to conduct 384,780 lesion‐symptom mapping analyses based on simulated behavioural data. Each simulated analysis used different combinations of plausible sample inclusion criteria, analysis parameters (e.g., correction factors), analysis types (e.g., univariate vs. multivariate), and underlying target correlates. Simulated analysis accuracy (percent coverage of target correlates, Dice similarity coefficient, and false positives) was compared across designs. Overall, analysis accuracy varied widely and was substantially modulated by the specific design used. Analyses that maximised lesion coverage by including large and diverse samples reliably outperformed analyses using more restricted samples. Analyses using direct total lesion volume controls outperformed analyses using other (or no) volume corrections across all accuracy measures. False discovery rate corrections yielded the best performance in terms of target coverage, while permutation corrections yielded the best Dice coefficients. While multivariate approaches were more accurate than univariate analyses in terms of Dice coefficient, univariate analyses generated higher target hit rates and percent target coverage. These results identify specific analysis designs suitable for studies aiming to maximise their sensitivity and/or specificity to underlying critical correlates, while highlighting the inferential strengths and weaknesses of these complementary approaches.

Large language models (LLMs) have achieved remarkable capabilities, yet methods to verify which model components are truly necessary for language function remain limited. Current interpretability approaches rely on internal metrics and lack external validation. Here we present the Brain-LLM Unified Model (BLUM), a framework that leverages lesion-symptom mapping—the gold standard for establishing causal brain-behavior relationships for over a century—as an external reference structure for evaluating LLM perturbation effects. Using data from individuals with chronic post-stroke aphasia (N = 410), we trained symptom-to-lesion models that predict brain damage location from behavioral error profiles, applied systematic perturbations to transformer layers, administered identical clinical assessments to perturbed LLMs and human patients, and projected LLM error profiles into human lesion space. LLM error profiles were sufficiently similar to human error profiles that predicted lesions corresponded to actual lesions in error-matched humans above chance in 67% of picture naming conditions and 68.3% of sentence completion conditions, with semantic-dominant errors mapping onto ventral-stream lesion patterns and phonemic-dominant errors onto dorsal-stream patterns. These findings open a new methodological avenue for LLM interpretability in which clinical neuroscience provides external validation, establishing human lesion-symptom mapping as a reference framework for evaluating artificial language systems and motivating direct investigation of whether behavioral alignment reflects shared computational principles.

Background and Objectives: Advanced brain age relative to chronological age may predict cognitive decline in healthy aging, but prospective evidence remains limited. We tested whether the brain age gap (BAG) predicts cognition and 5-year cognitive decline in the AGES-Reykjavik Study. Methods: Participants with structural MRI and neuropsychological data were included cross-sectionally (n=4,606); mean age 76.4y; 58% female) and longitudinally (n=2,561; mean follow-up 5.2y). BAG was estimated from T1-weighted MRI. Mixed-effects models examined BAG effects on Memory (M), processing speed (PS), and executive function (EF), adjusting for age, sex, and education. Results: Higher BAG was associated with worse cognition across all domains (b=-0.037 to -0.047, all p<0.001). BAG significantly predicted greater decline in M (b=-0.007, p=0.046); subgroup analyses revealed steeper decline in M and EF with advanced brain age. Discussion: BAG predicts longitudinal cognitive decline beyond chronological age, supporting its utility as a biomarker in preventative care of healthy aging.

Language depends on distributed brain networks whose neurochemical organisation remains largely unknown. Here, we combine large-scale functional MRI, positron-emission tomography-derived receptor and transporter maps, and structural connectomics to characterise the neurotransmitter architecture of the human language network and its disruption in post-stroke aphasia. We show that language circuits are organised by connection class: cortico-cortical pathways are dominated by serotonergic and glutamatergic signalling, thalamo-cortical projections are predominantly cholinergic, and anterior temporal connections show GABAergic circuit enrichment. In 239 stroke patients, aphasia was associated with distinct neurochemical disruption profiles, with lobar lesions showing postsynaptic serotonergic injury and subcortical aphasia showing preferential presynaptic cholinergic or serotonergic disruption. These findings map the neurochemical architecture of the language network and establish a neurotransmitter circuit-level framework for post-stroke aphasia. The post-hoc analysis of four pharmacological trials revealed an association between treatment-to-neurotransmitter matching and aphasia improvement, suggesting that personalised pharmacological strategies may enhance behavioural intervention responsiveness.

Effective rehabilitation tools are essential for improving language outcomes in chronic aphasia. Speech entrainment is a behavioral treatment that has shown promise in enhancing speech output in nonfluent aphasia, potentially by acting as an external mechanism to synchronize anterior and posterior language regions in the left hemisphere. Transcranial alternating current stimulation has been hypothesized to enhance functional connectivity between brain regions by amplifying endogenous oscillations. This proof-of-concept study explored whether high-definition tACS (HD-tACS) could improve speech fluency in nonfluent aphasia when paired with speech entrainment. In a double-blind, pseudorandomized study, 1 mA of HD-tACS at 7 Hz was applied to anterior and posterior left-hemisphere regions of individuals with nonfluent aphasia (N = 13). Stimulation was applied under three conditions: in-phase, anti-phase, and sham, and paired speech entrainment. Three outcome measures were examined: (1) number of words produced; (2) number of errors, and (3) ‘entrainment’ to the speech entrainment model. Group-level analyses for two of the three outcome measures reveal statistically significant differences between the experimental conditions. In-phase alternating current stimulation yielded more words and better entrainment to the audiovisual model than the sham condition. This study provides promising evidence that HD-tACS could improve speech production in individuals with nonfluent aphasia. These results contribute to growing evidence supporting the therapeutic potential of non-invasive brain stimulation approaches as an adjuvant to traditional behavioral speech-language therapy in stroke survivors.

Background: Recovery from chronic post-stroke aphasia is highly heterogeneous and shaped by lesion characteristics, brain integrity, and systemic health. Traditional group-level models struggle to capture this multidimensional, dynamic variability. Digital twin approaches - patient-specific, continually updating models - may enable individualized prediction and counterfactual evaluation of modifiable risk factors. Therefore, the aim was to develop and validate a proof-of-concept digital twin that predicts individual naming outcomes during language treatment and quantifies the estimated impact of modifiable health factors on naming. This study represents the first application of digital twin modeling to aphasia recovery, and we hypothesize that this could constitute a critical first step toward dynamically adaptive, personalized models for aphasia rehabilitation. Methods: We analyzed longitudinal data from 106 chronic stroke survivors with aphasia enrolled in the POLAR randomized clinical trial. For each participant we combined baseline demographic/health variables (age, sex, education, days post-stroke, hypertension, diabetes, BMI), lesion load in left-hemisphere language ROIs (JHU atlas), ROI-level white-matter microstructure (FA), and resting-state functional connectivity restricted to language regions. A continual-learning linear model (River framework; Adam optimizer) was pretrained on baseline data and updated across timepoints. Model performance was assessed by R² at the final timepoint. Counterfactual simulations systematically altered hypertension, diabetes, and BMI to estimate isolated and combined effects on predicted Philadelphia Naming Test (PNT) scores. Results: The digital twin predicted final PNT scores with R² = 0.5848 (explaining approximately 58% of variance). The largest contributors were prior naming performance, age, lesion load in language regions, and white-matter integrity in temporal regions (notably right MTG and STG pole). Counterfactual results estimated modest but consistent effects of health factors, with them collectively accounting for approximately 25% of the variance in treatment gains. The average change in PNT score with counterfactual changes was 7.92 (SD = 16.11). Therefore, diabetic status explained 2% of the variance in treatment gains, hypertensive status explained 4.75%, and increasing BMI explained 18.5%. Conclusions: This study demonstrate the feasibility and clinical potential of applying a digital twin framework to chronic post-stroke aphasia, with the model successfully predicting more than half the variance in naming performance during language treatment. Through counterfactual simulation, we demonstrated that modifiable health factors exert measurable, bidirectional influences on predicted treatment outcomes, underscoring the role of systemic health in shaping language recovery. Although the individual effects of these factors were modest in magnitude, their cumulative influence on treatment gains illustrates how multiple small biological contributors can add up to shape meaningful differences in language outcomes. More broadly, these findings illustrate the potential value of digital twin models for aphasia treatment, particularly as a tool to integrate diverse biological factors and generate individualized, dynamically updated predictions. Graphical