September 29, 2026
Using high-density electrocorticography in nine patients undergoing awake brain-tumor surgery with language mapping, a team from Fudan University's Huashan Hospital, ShanghaiTech University, NYU Shanghai, and collaborating institutions tracked how cortex encodes the articulatory movement trajectories of syllables that were either said aloud or only imagined. Supramodal populations in middle premotor, subcentral, and postcentral–supramarginal regions coded both modes during planning, while populations specific to imagery or to articulation sat interleaved in primary sensorimotor cortex. A decoder built on these signals identified imagined syllables with a median accuracy of 80.4% (chance 16.7%), close to the 78.3% it reached for spoken ones, which the authors say bears on imagery-based brain–computer interfaces.
Nature Neuroscience →September 24, 2026
Kim Nguyen, working with Nora Newcombe while at Temple University, used fMRI to compare children aged 8–13 with young adults as they viewed objects they had first encountered in real life, either seated in a room or on a tour of an unfamiliar academic building, mixed with new foil objects. For objects from the tour, activity in the entorhinal and perirhinal cortices differed between the two age groups, with the early adolescents appearing to need stronger communication between the two regions to retrieve the episodic memory. The hippocampal activation the team expected did not appear. Published in the Journal of Cognitive Neuroscience.
Cognitive Neuroscience Society →September 22, 2026
Recording from lateral prefrontal cortex in two animals trained on a two-target value-based choice task, Earl K. Miller's lab at MIT's Picower Institute found that neural ensembles first code the options by the order they were presented, then regroup them into chosen versus unchosen once the decision is made. Individual neurons joined different ensembles from trial to trial. Rather than fixed circuits, the representations were flexible assemblies kept in separate neural subspaces, which the authors argue prevents options from interfering with one another.
MIT News →September 18, 2026
A team led by Mitra Javadzadeh at Cold Spring Harbor Laboratory, with collaborators at the University of Cambridge and University College London, recorded simultaneously from primary visual cortex (V1) and the higher-order lateromedial area (LM) in mice discriminating grating orientations — 194 V1 and 228 LM neurons across seven mice — while briefly silencing one area with optogenetics. Circuit models fit to the data showed that reciprocal excitatory connections between the two areas act as an approximate line attractor: activity patterns the areas agree on decay slowly, while inconsistent ones are suppressed quickly. The authors call this "dynamic consensus building" and propose it as a general principle of how specialized cortical areas reconcile conflicts into coherent percepts and decisions.
Nature Neuroscience →September 17, 2026
Fabian-Xosé Fernandez, Sara N. Burke, and Lynn Nadel of the University of Arizona argue in Perspectives on Psychological Science that late-life cognition is an adaptive reorganization rather than decline: memory shifts in emphasis from the hippocampus, which captures detail, to the frontal cortex, which processes broader meaning. They propose this supports narrative coherence, emotional regulation, and knowledge-sharing roles such as mentoring, and that Alzheimer's disease may derail the normal shift. It is a hypothesis and framework paper, not a new experiment.
Perspectives on Psychological Science →September 15, 2026
Using magnetoencephalography and time-resolved decoding, researchers showed that covertly producing a syllable recruits speech motor cortex for articulatory features before the imagined onset, then auditory cortex for acoustic features just afterward — the same temporal staging seen in overt speech. Participants silently produced consonant-vowel syllables designed to tease apart articulatory and acoustic contrasts, and the neural signatures tracked both dimensions with sub-phonemic precision.
PNAS →September 15, 2026
Recording from stereotactic EEG electrodes in the orbitofrontal cortex (OFC) of six patients implanted for clinical reasons, a UCSF and UC Berkeley team led by Clara Kwon Starkweather, with Edward F. Chang and Robert T. Knight jointly supervising, had participants play a video-game-style task, deciding whether to approach treasure chests in a corridor that also held bombs. Before approaching, activity rose in the medial OFC and fell in the lateral OFC. In the moments before each decision, these signals switched rapidly between discrete pro-approach and pro-avoidance states. The authors propose this as a new functional architecture for how the human OFC weighs reward against potential punishment.
Nature Neuroscience →September 10, 2026
Tal Boger and Chaz Firestone at Johns Hopkins ran 11 experiments (N ≈ 1,500) showing that complexity is represented as a unified cognitive kind. In a reward-transfer paradigm, outcomes learned for one stimulus class — visual shapes — generalized to dot arrays, melodies, letter strings, mathematical expressions, and tactile forms without additional training. Follow-up experiments showed the transfer is automatic (intruding on task-irrelevant judgements) and that individual differences in complexity preference are stable across domains: a person who finds simple shapes aesthetically pleasing tends to prefer simpler melodies and expressions as well.
Nature Human Behaviour →September 10, 2026
A University of Chicago team led by Cheng Xue and Marlene Cohen found that when subjects are uncertain which task rule applies, visual features that are normally encoded by separate neuronal populations collapse into the same ones — "feature interference." Combining rhesus monkey electrophysiology, artificial neural networks, and psychophysics in over 200 human participants, the study frames the cost of multitasking as a structural limit on how neurons represent information, not a failure of effort.
Nature Neuroscience →September 5, 2026
The Cognitive Neuroscience Society's next annual meeting is set for Boston, Massachusetts, with invited symposia, poster sessions, a keynote, and award lectures for the George A. Miller Award, Distinguished Career Contributions Award, and Young Investigators Award. Poster and symposium submissions are already open; see the Events page here for the listing.
Cognitive Neuroscience Society →September 5, 2026
This domain's first version dates back to 1998; this is its first official full relaunch since then. The rebuilt site is currently undergoing review, and we're looking for volunteers to help check listings, flag missing programs and labs, and give feedback before a wider announcement.
cognitivescience.org →September 5, 2026
The Cognitive Science Society's flagship annual meeting returns to a fully in-person format next year, at the Euskalduna Bilbao Convention Centre. The Society's three-year rotation made 2026's Rio de Janeiro meeting the hybrid year; 2027 goes back to a single in-person site, following Western Europe in the location cycle. Program chairs are named on the official page, though submission and registration deadlines aren't posted yet — see the Events page here for the listing.
Cognitive Science Society →September 5, 2026
CogSci Grove funds native tree planting at Trees for Life's Dundreggan estate near Loch Ness in the Scottish Highlands, offsetting carbon from the Society's conference travel and open to members offsetting their own professional or personal activities. Running since 2020, it's planted over 2,406 trees so far — including 206 the Society funded directly to offset CogSci 2025's footprint — at roughly six trees per tonne of CO2, £6 per tree.
Cognitive Science Society →September 1, 2026
Earl K. Miller, Scott L. Brincat, and Jefferson E. Roy of MIT's Picower Institute argue in the Journal of Neuroscience that the standard "circuit" metaphor is incomplete: physical connections store information, but traveling waves of rhythmic activity coordinate millions of neurons and carry out analog computation. They call the mechanism spatiotemporal computing, in which wave interference runs many calculations in parallel, and propose that consciousness arises when those wave patterns grow large enough to unify cortex — a prediction they tie to earlier anesthesia work showing three drugs with different molecular targets all disrupting wave dynamics.
MIT News →August 31, 2026
Caitriona Douglas and colleagues at Dalhousie University used EEG hyperscanning — recording two people at once during a scripted conversation — to test how the N400 lexical frequency effect behaves outside a single-subject lab task. Taking part in the conversation reduced the neural processing normally associated with less frequent words, while a control group that only watched a recorded conversation showed the larger effect. The team also worked out how to run a hyperscanning study on a single EEG system, removing the usual requirement to own two, and built the conversation task on techniques borrowed from improv and theatre.
Cognitive Neuroscience Society →August 30, 2026
In a randomized crossover trial (the HAFTRAP study), 119 adults performed about 12% faster on the Trail Making Test after a month with a HEPA air purifier at home versus a sham unit — a benefit seen only in participants 40 and older, and roughly comparable in size to the cognitive benefits linked to increased exercise.
Scientific Reports →August 25, 2026
Researchers at the University of Edinburgh and the UK Dementia Research Institute found that oligodendrocytes — the brain's myelin-producing cells — become dysfunctional with age and actively contribute to cognitive decline rather than just failing to prevent it. Using the Lothian Birth Cohort 1936 (866 participants with cognitive testing from age 70 to 82) plus mouse experiments, they traced the effect to reduced NRF2 protein levels, pointing at existing MS drugs as candidates to repurpose for preserving cognition in aging.
Nature Medicine →August 24, 2026
Madhura Lotlikar and senior author Marc Roig at McGill University kept 54 healthy young adults awake for 30 consecutive hours, then randomly assigned them to a 20-minute moderate-to-vigorous cycling session, a 90-minute nap, or an inactive control condition. Tested on image recognition three days later, both the exercise and nap groups scored about 22% higher than the control group. Brain recordings suggested the two interventions worked through different mechanisms: napping appeared to let the brain recharge for taking in new information, while exercise helped it use remaining cognitive resources more efficiently without reducing fatigue.
McGill University →August 24, 2026
An MIT and Boston University team led by Evelina Fedorenko and Swathi Kiran compared fMRI responses in adults aged 17–39 and 41–80 and found the language network essentially unchanged with age — same topography, same left-lateralization, same selectivity for language over other tasks, and same within-network functional connectivity. The multiple-demand (executive) network in the same participants declined on nearly every measure. Older adults also retained strong sensitivity to linguistic difficulty, arguing against any fundamental change in how they process language.
MIT News →August 24, 2026
Across three studies covering seven datasets and more than 880,000 texts, a University of Southern California team led by Zhivar Sourati found that when large language models polish or rewrite text they keep the core content but homogenize style, reducing writing-complexity variance by a statistically significant 21–50% across datasets and models. The effect amplifies patterns tied to dominant author characteristics and suppresses others, which the authors flag as a problem for any field that reads identity, psychological state, or social context off of language — diagnostics, personalization, and hiring assessment among them.
Nature Human Behaviour →August 20, 2026
Ryssa Moffat, Guillaume Dumas, and Emily S. Cross tracked 31 intergenerational and 30 same-generation pairs (91 younger adults 18–35, 31 older adults 69+, all in Zurich) through a six-week collaborative drawing program, recording fNIRS hyperscanning over bilateral inferior frontal and temporoparietal regions. Same-generation pairs' brain synchrony rose across sessions in theory-of-mind regions, while intergenerational pairs' synchrony fell in the same regions — even though both groups reported growing socially closer. The authors read the split as a prediction effect: familiarity lets intergenerational partners anticipate each other more easily, while younger same-age pairs may sustain higher synchrony by staying less predictable with each other.
PLOS Biology →August 20, 2026
Yingxu Liu, Leon Aksman, and Arthur Toga at USC's Stevens Neuroimaging and Informatics Institute analyzed brain scans and cognitive assessments from 459 community-dwelling adults aged 60+ across India. Gray-matter atrophy remained the strongest predictor of cognitive decline, but healthier superficial white matter — the thin layer of short, curved nerve fibers just beneath the cortex — was linked particularly to stronger language performance, including word recognition and verbal fluency. Among people with comparable gray-matter loss, those with healthier superficial white matter showed smaller cognitive deficits, suggesting this local wiring can buffer against gray-matter atrophy's effects.
USC Keck School of Medicine →August 8, 2026
Researchers at Tsinghua University, TU Graz, and Italy's National Research Council, led by Wolfgang Maass, built an AI system that plans using brain-inspired 'cognitive maps' instead of standard deep neural networks — motivated by the brain's roughly 20-watt power budget against the energy cost of current AI. Published in Nature Machine Intelligence, with hardware-implementation collaboration already underway with Intel and a US startup.
Nature Machine Intelligence →August 1, 2026
The Cognitive Science Society's flagship annual meeting returns as a hybrid event this year, with simultaneous Spanish and Portuguese translation on select sessions. Standard registration closes July 1, 2026; see the Events page here for the full date breakdown.
Cognitive Science Society →July 28, 2026
A free, peer-reviewed, continuously growing successor to the print-era MIT Encyclopedia of the Cognitive Sciences (MITECS) — now with over 200 cross-referenced articles, each carrying a DOI and written by researchers in the field. Worth bookmarking as background reading alongside anything on this site.
Open Encyclopedia of Cognitive Science (MIT) →July 24, 2026
The Cognitive Science Society named Herbert H. Clark, Professor Emeritus of Psychology at Stanford University, as the recipient of the twenty-seventh David E. Rumelhart Prize. The society's citation credits Clark with recasting language as a fundamentally collaborative activity — treating speaking and understanding as forms of joint action — and traces that influence through cognitive psychology, linguistics, psycholinguistics, artificial intelligence, and human-computer interaction. The prize carries a bronze medal, a certificate, a citation, and $100,000.
Cognitive Science Society →No entries match that search. Try a wider date range or a different keyword.