Brain Computations and Connectivity

Oxford University Press eBooks, Journal Year: 2023, Volume and Issue: unknown

Published: July 11, 2023

Abstract The subject of this book is how the brain works. In order to understand this, it essential know what computed by different systems; and computations are performed. aim elucidate in describe current computational approaches models each these systems computes. Understanding way has enormous potential for understanding ourselves better health disease. Potential applications treatment disease; artificial intelligence which will benefit from knowledge performs many its extraordinarily impressive functions. This pioneering taking approach function: consider our computed. also biologically plausible computation. incorporating evidence on connectivity 360 cortical regions human brain, making highly relevant brain. be interest all scientists interested function works, whethr they neuroscience, or medical sciences including neurology psychiatry, area science machine learning intelligence, areas such as theoretical physics.

Language: Английский

---

A Theory and Model of Scene Representations With Hippocampal Spatial View Cells

Hippocampus, Journal Year: 2025, Volume and Issue: 35(3)

Published: April 28, 2025

ABSTRACT A theory and network model are presented of how scene representations built by forming spatial view cells in the ventromedial visual cortical pathway to hippocampus primates including humans. Layer 1, corresponding V1–V4, connects 2 retrosplenial area uses competitive learning form feature combination neurons for part being fixated, a fixation patch. In 3, parahippocampal hippocampus, patches stitched together whole representations. This is performed with continuous attractor made from overlapping Gaussian receptive fields as head rotates scene. addition, gain modulation gaze direction maps correct representation when saccades made. Each neuron 3 thus cell that responds location viewed based on features The novel conceptual advances this shows may be primates, humans, scenes anchor world (to allocentric, world‐based, space); contributes this. offers revolutionary approach understanding navigation episodic memory

Language: Английский

---

A theory of hippocampal function: New developments

Progress in Neurobiology, Journal Year: 2024, Volume and Issue: 238, P. 102636 - 102636

Published: June 2, 2024

We develop further here the only quantitative theory of storage information in hippocampal episodic memory system and its recall back to neocortex. The is upgraded account for a revolution understanding spatial representations primate, including human, hippocampus, that go beyond place where individual located, location being viewed scene. This fundamental much primate navigation: functions supported humans by pathways build 'where' view feature combinations ventromedial visual cortical stream, separate from those 'what' object face inferior temporal cortex, reward orbitofrontal cortex. Key new computational developments include capacity CA3 attractor network storing whole charts space; how correlations inherent self-organizing continuous impact capacity; can combine discrete representations; roles rewards reach hippocampus later consolidation into long-term part via cholinergic cortex; ways analysing neocortical using Potts networks.

Language: Английский

---

The memory systems of the human brain and generative artificial intelligence

Heliyon, Journal Year: 2024, Volume and Issue: 10(11), P. e31965 - e31965

Published: May 24, 2024

Generative Artificial Intelligence foundation models (for example Pre-trained Transformer - GPT models) can generate the next token given a sequence of tokens. How this 'generative AI' be compared with 'real' intelligence human brain, when for generates whole memory in response to an incomplete retrieval cue, and then further prospective thoughts? Here these two types generative intelligence, artificial machines real brain are compared, it is shown how memories generated by hippocampal recall what computes, computes it, very different from AI. Key differences use local associative learning rules system, non-local backpropagation error Indeed, argued that operation performed computationally differently implemented Moreover, emphasized primate including system includes computations about spatial view where objects people scenes, whereas rodents emphasis on place cells path integration movements between places. This comparison processing has interesting implications development AI neuroscience research.

Language: Английский

---

Slow semantic learning in the cerebral cortex, and its relation to the hippocampal episodic memory system

Cerebral Cortex, Journal Year: 2025, Volume and Issue: 35(5)

Published: May 1, 2025

Abstract A key question is how new semantic representations are formed in the human brain and this may benefit from hippocampal episodic memory system. Here, we describe major effective connectivity between system anterior temporal lobe (ATL) humans. Then, present model a theory of be ATL using slow associative learning attractor networks that receive inputs The hypothesis if one category being processed for several seconds, then short-term trace biologically plausible rule will enable all components during time to associated together network. This benefits binding provided by modeled four-layer network view-invariant visual object recognition, followed layer utilizes form categories based on occur close time, or world.

Language: Английский

---

Visual cortical networks for “What” and “Where” to the human hippocampus revealed with dynamical graphs

Cerebral Cortex, Journal Year: 2025, Volume and Issue: 35(5)

Published: May 1, 2025

Abstract Key questions for understanding hippocampal function in memory and navigation humans are the type source of visual information that reaches human hippocampus. We measured bidirectional pairwise effective connectivity with functional magnetic resonance imaging between 360 cortical regions while 956 Human Connectome Project participants viewed scenes, faces, tools, or body parts. developed a method using deterministic dynamical graphs to define whole networks flow both directions their over timesteps after signal is applied V1. revealed ventromedial “Where” network from V1 via retrosplenial medial parahippocampal scene areas hippocampus when scenes viewed. A ventrolateral “What” V2–V4, fusiform face cortex, lateral region TF faces/objects There major implications computations vs rodent navigation: primates fovea highly processing process about location objects, landmarks whereas rodents representations system mainly place where individual located self-motion places.

Language: Английский

---

A ventromedial visual cortical ‘Where’ stream to the human hippocampus for spatial scenes revealed with magnetoencephalography

Communications Biology, Journal Year: 2024, Volume and Issue: 7(1)

Published: Aug. 25, 2024

Abstract The primate including the human hippocampus implicated in episodic memory and navigation represents a spatial view, very different from place representations rodents. To understand this system humans, computations performed, pathway for view information to reach was analysed humans. Whole-brain effective connectivity measured with magnetoencephalography between 30 visual cortical regions 150 other using HCP-MMP1 atlas 21 participants while performing 0-back scene task. In ventromedial stream, V1–V4 connect ProStriate region where retrosplenial area is located. has VMV1–3 VVC. These medial parahippocampal PHA1–3, which, VMV regions, include area. have entorhinal cortex, perirhinal hippocampus. contrast, when viewing faces, more through ventrolateral stream via fusiform face cortex inferior temporal TE2p TE2a. A ‘Where’ scenes supported by diffusion topography 171 HCP at 7 T.

Language: Английский

---

Selective Brain Activations and Connectivities Related to the Storage and Recall of Human Object‐Location, Reward‐Location, and Word‐Pair Episodic Memories

Human Brain Mapping, Journal Year: 2024, Volume and Issue: 45(15)

Published: Oct. 15, 2024

ABSTRACT Different cortical systems to the hippocampus were activated using fMRI during different types of episodic memory task. For object with scene location memory, activations high in involved spatial processing, including ventromedial visual and medial parahippocampal system. These for system higher right hemisphere. The face processing ventrolateral stream regions FFC, PIT, V8 TE2p object‐location task than reward‐location task, also hippocampus, but selectively storage key reward (ventromedial prefrontal 10r, 10v, 10d; pregenual anterior cingulate d32, p24, p32, s32; orbitofrontal cortex reward‐related pOFC, 11l, OFC). word‐pair lower stream, language‐related Broca's area (44, 45, 47l), left hemisphere these many highly connected inferior frontal gyrus Further, effective connectivity analyses tasks showed that direction was from early V2–V4 VMV1–3 VVC processing; hippocampal system; FFC/V8/PIT temporal cortex, which has lateral TF, turn forward hippocampus.

Language: Английский

---

Hippocampal Discoveries: Spatial View Cells, Connectivity, and Computations for Memory and Navigation, in Primates Including Humans

Hippocampus, Journal Year: 2024, Volume and Issue: 35(1)

Published: Dec. 17, 2024

Two key series of discoveries about the hippocampus are described. One is discovery hippocampal spatial view cells in primates. This opens way to a much better understanding human episodic memory, for memory prototypically involves where people or objects rewards have been seen locations "out there" which could never be implemented by place that encode location rat mouse. Further, valuable navigation using vision and viewed landmarks, provide richer, vision-based, than self-motion update performed rats mice who live dark underground tunnels. Spatial thus offer revolution our functions humans other primates with well-developed foveate vision. The second describes computational theory hippocampal-neocortical system includes only quantitative how information recalled from neocortex. It shown foundations this research were reward neurons food reward, non-reward, primate orbitofrontal cortex, representations value including monetary cortex; face identity expression inferior temporal visual cortex they represent transform-invariant information. illustrates order understand brain computation, whole integrated interdisciplinary needed build operation each neural system.

Language: Английский

---

The orbitofrontal cortex, amygdala, reward value, emotion, and decision-making

Oxford University Press eBooks, Journal Year: 2023, Volume and Issue: unknown, P. 475 - 563

Published: July 11, 2023

Abstract The orbitofrontal cortex receives from the ends of all sensory processing systems, and converts these representations what stimulus is into their reward value. therefore a key brain region in emotions, which can be defined as states elicited by rewards punishers. Indeed, activations are linearly related to subjectively reported pleasantness stimuli. then projects this value information other structures, implement behavioural output, such anterior cingulate cortex, basal ganglia. A computational capacity one-trial object-reward associations, rule-based, enable primates including humans change rewarded behaviour very rapidly. Decision-making using attractor neural networks described. Reward signals have connectivity with cholinergic neurons may thereby influence hippocampal episodic memory consolidation.

Language: Английский

---