Schemas, reinforcement learning and the medial prefrontal cortex

Nature reviews. Neuroscience, Год журнала: 2025, Номер unknown

Опубликована: Янв. 7, 2025

Язык: Английский

---

Neural representation dynamics reveal computational principles of cognitive task learning

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

Опубликована: Июнь 29, 2023

During cognitive task learning, neural representations must be rapidly constructed for novel performance, then optimized robust practiced performance. How the geometry of changes to enable this transition from performance remains unknown. We hypothesized that practice involves a shift compositional (task-general activity patterns can flexibly reused across tasks) conjunctive (task-specific specialized current task). Functional MRI during learning multiple complex tasks substantiated dynamic representations, which was associated with reduced cross-task interference (via pattern separation) and behavioral improvement. Further, we found conjunctions originated in subcortex (hippocampus cerebellum) slowly spread cortex, extending memory systems theories encompass representation learning. The formation hence serves as computational signature reflecting cortical-subcortical dynamics optimize human brain.

Язык: Английский

---

Memory’s gatekeeper: The role of PFC in the encoding of congruent events

Proceedings of the National Academy of Sciences, Год журнала: 2024, Номер 121(30)

Опубликована: Июль 17, 2024

Theoretical models conventionally portray the consolidation of memories as a slow process that unfolds during sleep. According to classical Complementary Learning Systems theory, hippocampus (HPC) rapidly changes its connectivity wakefulness encode ongoing events and create memory ensembles are later transferred prefrontal cortex (PFC) However, recent experimental studies challenge this notion by showing new information consistent with prior knowledge can be consolidated in PFC lesions disrupt encoding congruent HPC. The contributions have therefore largely been overlooked. Moreover, most theoretical frameworks assume random uncorrelated patterns representing memories, disregarding correlations between our experiences. To address these shortcomings, we developed HPC-PFC network model simulates interactions HPC (awake stage), subsequent (sleeping stage) examine each region novel memories. Our results show uses stored "schemas" previous experiences identify inputs evoke activity, quickly integrate it into network, gate which components encoded More specifically, GABAergic long-range projections inhibit neurons input correlated previously "schema," eliciting sparse hippocampal activity exposure events, has experimentally observed.

Язык: Английский

---

Predictions enable top-down pattern separation in the macaque face-processing hierarchy

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

Опубликована: Авг. 21, 2024

Abstract Distinguishing faces requires well distinguishable neural activity patterns. Contextual information may separate representations, leading to enhanced identity recognition. Here, we use functional magnetic resonance imaging investigate how predictions derived from contextual affect the separability of patterns in macaque face-processing system, a 3-level processing hierarchy ventral visual cortex. We find that presence predictions, early stages this exhibit separable and high-dimensional geometries resembling those at top hierarchy. This is accompanied by systematic shift tuning properties higher lower areas, endowing areas with higher-order, invariant representations instead their feedforward properties. Thus, top-down signals dynamically transform into geometries. Our results provide evidence predictive context transforms flexible representational spaces optimally computational resources provided cortical hierarchies for better faster distinction facial identities.

Язык: Английский

---

A neural network model of differentiation and integration of competing memories

eLife, Год журнала: 2024, Номер 12

Опубликована: Сен. 25, 2024

What determines when neural representations of memories move together (integrate) or apart (differentiate)? Classic supervised learning models posit that, two stimuli predict similar outcomes, their should integrate. However, these have recently been challenged by studies showing that pairing with a shared associate can sometimes cause differentiation, depending on the parameters study and brain region being examined. Here, we provide purely unsupervised network model explain other related findings. The exhibit integration differentiation amount activity allowed to spread competitors — inactive are not modified, connections moderately active weakened (leading differentiation), highly strengthened integration). also makes several novel predictions most importantly, occurs as result this mechanism, it will be rapid asymmetric, give rise anticorrelated in is source differentiation. Overall, modeling results computational explanation for diverse set seemingly contradictory empirical findings memory literature, well new insights into dynamics at play during learning.

Язык: Английский

---

Hippocampal-cortical connectivity relates to inter-individual differences and training gains in distinguishing similar memories

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2025, Номер unknown

Опубликована: Янв. 9, 2025

Abstract Mnemonic discrimination (MD) is the ability to distinguish current experiences from similar memories. Research on brain correlates of MD has focused how regional neural responses are linked MD. Here we go beyond this approach investigate inter-regional functional connectivity patterns related MD, its inter-individual variability and training-related improvement. Based prior work medial temporal lobe (MTL), prefrontal cortex (PFC) visual regions. We used fMRI determine between these regions before after 2-weeks web-based cognitive training. identified a signature involving MTL-PFC-visual areas during successful found that hippocampal-PFC was negatively associated with interindividual in performance across two different tasks. Hippocampal-PFC decrease also post-training Additionally, training led increased lateral occipital pole area. Our results point pattern which reliable, task-invariant, marker performance. This further gains providing causal evidence for relevance distinguishing Overall, show constitutes resource enables improvements could be targeted future research enhance cognition.

Язык: Английский

---

How to form brain-like memory in spiking neural networks with the help of frequency-induced mechanism

Neurocomputing, Год журнала: 2025, Номер unknown, С. 129361 - 129361

Опубликована: Янв. 1, 2025

Язык: Английский

---

Impaired hippocampal circuitry and memory dysfunction in schizophrenia

Nature Mental Health, Год журнала: 2025, Номер unknown

Опубликована: Янв. 15, 2025

Язык: Английский

---

Contextual Influence on Pattern Separation During Encoding

NeuroSci, Год журнала: 2025, Номер 6(1), С. 13 - 13

Опубликована: Фев. 6, 2025

Pattern separation is considered a crucial process that allows us to distinguish among the highly similar and overlapping experiences constitute our episodic memory. Not only do different episodes share common features, but it often case they context in which occurred. While there have been great number of studies investigating pattern its behavioral counterpart, known as mnemonic discrimination, surprisingly, research exploring influence on or discrimination has less common. The available evidence shows items with contexts led failure due high similarity triggers overlap between events. On other hand, others shown can take place even under these conditions, allowing humans events contexts, hippocampal subfields would play complementary roles enabling both completion. In present study, we were interested testing how stability influenced separation. Despite fact by definition an encoding computation, existing literature focused retrieval phase. Here, used subsequent memory paradigm manipulated during visual objects selected from diverse categories. Thus, encoded each object category (four within category), so some categories had same context. This approach allowed test not presented also include conditions entail greatest demand After 20 min period, participants performed task differentiate old, similar, new providing one three options for tested item. Similarly previous studies, found no interaction judgments able discriminate old lure at level conditions. Moreover, when averaging ERPs all category, significant SME emerged hits misses, false alarms alarms. These results suggest item recognition emerges subsequently information, just strength processes.

Язык: Английский

---

Menstrual cycle and perceived stress predict performance on the mnemonic similarity task

PLoS ONE, Год журнала: 2025, Номер 20(5), С. e0322652 - e0322652

Опубликована: Май 2, 2025

A growing body of literature demonstrates strong effects ovarian hormones on the hippocampus and adjacent structures. However, resulting impacts human cognition remain unclear. Addressing this gap, we examine pattern separation ability, a core hippocampal process, across menstrual cycle using mnemonic similarity task as behavioral index (N = 183). We find non-linear effect cycle, with performance peaking in high-estradiol, late follicular phase reaching its lowest point during mid-luteal phase, which is characterized by moderate estradiol high progesterone levels. Additionally, that perceived stress may facilitate performance. These results to importance for cognition, reveal novel performance, provide preliminary evidence possible neural pathways involved separation.

Язык: Английский

---