Biological constraints on neural network models of cognitive function

Nature reviews. Neuroscience, Год журнала: 2021, Номер 22(8), С. 488 - 502

Опубликована: Июнь 28, 2021

Язык: Английский

---

Language is primarily a tool for communication rather than thought

Nature, Год журнала: 2024, Номер 630(8017), С. 575 - 586

Опубликована: Июнь 19, 2024

Язык: Английский

---

Structural-functional brain network coupling predicts human cognitive ability

NeuroImage, Год журнала: 2024, Номер 290, С. 120563 - 120563

Опубликована: Март 16, 2024

Individual differences in general cognitive ability (GCA) have a biological basis within the structure and function of human brain. Network neuroscience investigations revealed neural correlates GCA structural as well functional brain networks. However, whether relationship between networks, structural-functional network coupling (SC-FC coupling), is related to individual remains an open question. We used data from 1030 adults Human Connectome Project, derived connectivity diffusion weighted imaging, resting-state fMRI, assessed latent g-factor 12 tasks. Two similarity measures six communication were model possible interactions arising SC-FC was estimated degree which these align with actual connectivity, providing insights into different strategies. At whole-brain level, higher associated coupling, but only when considering path transitivity strategy. Taking region-specific variations strategy account differentiating positive negative associations GCA, allows for prediction scores cross-validated framework (correlation predicted observed scores: r = .25, p < .001). The same also predicts completely independent sample (N 567, .19, Our results propose neurobiological correlate suggest strategies efficient information processing predictive ability.

Язык: Английский

---

A hierarchical atlas of the human cerebellum for functional precision mapping

Nature Communications, Год журнала: 2024, Номер 15(1)

Опубликована: Сен. 27, 2024

Язык: Английский

---

Integrative frontal-parietal dynamics supporting cognitive control

eLife, Год журнала: 2021, Номер 10

Опубликована: Март 2, 2021

Coordinating among the demands of external environment and internal plans requires cognitive control supported by a fronto-parietal network (FPCN). Evidence suggests that multiple systems span FPCN whose operations are poorly understood. Previously (Nee D’Esposito, 2016; 2017), we detailed frontal dynamics support processing, but left open their role in broader cortical function. Here, I show consists an external/present-oriented to internal/future-oriented gradient extending outwardly from sensory-motor cortices. Areas at ends this act segregative manner, exciting areas same level, suppressing different levels. By contrast, middle excite all levels, promoting integration processing. Individual differences integrative predict higher level ability amenability neuromodulation. These data suggest intermediary zone within underlies processing supports control.

Язык: Английский

---

Incremental Language Comprehension Difficulty Predicts Activity in the Language Network but Not the Multiple Demand Network

Cerebral Cortex, Год журнала: 2021, Номер 31(9), С. 4006 - 4023

Опубликована: Март 2, 2021

What role do domain-general executive functions play in human language comprehension? To address this question, we examine the relationship between behavioral measures of comprehension and neural activity "multiple demand" (MD) network, which has been linked to constructs like attention, working memory, inhibitory control, selection, implicated diverse goal-directed behaviors. Specifically, functional magnetic resonance imaging data collected during naturalistic story listening are compared with theory-neutral online difficulty incremental processing load (reading times eye-fixation durations). Critically, ensure that variance these is driven by features linguistic stimulus rather than reflecting participant- or trial-level variability, neuroimaging datasets were nonoverlapping samples. We find no behavioral-neural link functionally localized MD regions; instead, found domain-specific, fronto-temporal "core network," both left-hemispheric areas their right hemispheric homotopic areas. These results argue against strong involvement circuits comprehension.

Язык: Английский

---

Robust Effects of Working Memory Demand during Naturalistic Language Comprehension in Language-Selective Cortex

Journal of Neuroscience, Год журнала: 2022, Номер 42(39), С. 7412 - 7430

Опубликована: Авг. 24, 2022

To understand language, we must infer structured meanings from real-time auditory or visual signals. Researchers have long focused on word-by-word structure building in working memory as a mechanism that might enable this feat. However, some argued language processing does not typically involve rich building, and/or apparent effects are underlyingly driven by surprisal (how predictable word is context). Consistent with alternative, recent behavioral studies of naturalistic control for surprisal shown clear effects. In fMRI study, investigate range theory-driven predictors demand during comprehension humans both sexes under rigorous controls. addition, address related debate about whether the mechanisms involved specialized domain general. do so, each participant, functionally localize (1) language-selective network and (2) “multiple-demand” network, which supports across domains. Results show robust surprisal-independent no effect multiple-demand network. Our findings thus support view involves computationally demanding operations memory, addition to any prediction-related mechanisms. Further, these appear be primarily conducted same neural resources store linguistic knowledge, evidence involvement brain regions known SIGNIFICANCE STATEMENT This study uses signatures (WM) story listening, using broad theoretically motivated estimates WM demand. strong distinct predictability. demands register regions, rather than previously been associated nonlinguistic core role incremental processing, language.

Язык: Английский

---

Precision fMRI reveals that the language-selective network supports both phrase-structure building and lexical access during language production

Cerebral Cortex, Год журнала: 2022, Номер 33(8), С. 4384 - 4404

Опубликована: Авг. 25, 2022

A fronto-temporal brain network has long been implicated in language comprehension. However, this network's role production remains debated. In particular, it unclear whether all or only some regions contribute to production, and which aspects of these support. Across 3 functional magnetic resonance imaging experiments that rely on robust individual-subject analyses, we characterize the response high-level demands. We report novel results. First, sentence spoken typed, elicits a strong throughout network. Second, responds both phrase-structure building lexical access demands, although is stronger more spatially extensive, present every region. Finally, contra proposals, find no evidence regions-within outside network-that selectively support relative Instead, respond strongly during than comprehension, suggesting incurs greater cost for Together, results align with idea comprehension draw same knowledge representations, are stored distributed manner within language-selective used interpret generate linguistic utterances.

Язык: Английский

---

A hierarchical atlas of the human cerebellum for functional precision mapping

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2023, Номер unknown

Опубликована: Сен. 14, 2023

ABSTRACT The human cerebellum is activated by a wide variety of cognitive and motor tasks. Previous functional atlases have relied on single task-based or resting-state fMRI datasets. Here, we present atlas that integrates information from 7 large-scale datasets, outperforming existing group atlasses. new has three further advantages: First, the allows for precision mapping in individuals: integration probabilistic with an individual localizer scan results marked improvement prediction boundaries. Second, provide both asymmetric symmetric versions atlas. version, which obtained constraining boundaries to be same across hemispheres, especially useful studying lateralization. Finally, regions are hierarchically organized 3 levels, allowing analyses at appropriate level granularity. Overall, important resource study interdigitated organization health disease.

Язык: Английский

---

Language models, like humans, show content effects on reasoning tasks

PNAS Nexus, Год журнала: 2024, Номер 3(7)

Опубликована: Июнь 28, 2024

reasoning is a key ability for an intelligent system. Large language models (LMs) achieve above-chance performance on abstract tasks but exhibit many imperfections. However, human also imperfect. Human affected by our real-world knowledge and beliefs, shows notable "content effects"; humans reason more reliably when the semantic content of problem supports correct logical inferences. These content-entangled patterns are central to debates about fundamental nature intelligence. Here, we investigate whether models-whose prior expectations capture some aspects knowledge-similarly mix into their answers logic problems. We explored this question across three tasks: natural inference, judging validity syllogisms, Wason selection task. evaluate state art LMs, as well humans, find that LMs reflect same qualitative these tasks-like answer accurately task parallels reflected in accuracy patterns, lower-level features like relationship between LM confidence over possible response times. cases behave differently-particularly task, where perform much worse than large models, distinct error pattern. Our findings have implications understanding contributors cognitive effects, factors influence model performance.

Язык: Английский

---