Structure transfer and consolidation in visual implicit learning

Опубликована: Янв. 24, 2025

Transfer learning, the re-application of previously learned higher-level regularities to novel input, is a key challenge in cognition. While previous empirical studies investigated human transfer learning supervised or reinforcement for explicit knowledge, it unknown whether such occurs during naturally more common implicit and unsupervised and, if so, how related memory consolidation. We compared newly acquired abstract knowledge by extending visual statistical paradigm context. found but with important differences depending on explicitness/implicitness knowledge. Observers acquiring initial could structures immediately. In contrast, observers same amount showed opposite effect, structural interference transfer. However, sleep between phases, observers, while still remaining implicit, switched their behaviour pattern as did. This effect was specific not after non-sleep Our results highlight similarities generalizable relying consolidation restructuring internal representations.

Язык: Английский

---

How much intelligence is there in artificial intelligence? A 2020 update

Intelligence, Год журнала: 2021, Номер 87, С. 101548 - 101548

Опубликована: Май 25, 2021

Schank (1980) wrote an editorial for Intelligence on “How much intelligence is there in artificial intelligence?”. In this paper, we revisit question. We start with a short overview of modern AI and showcase some the breakthroughs four decades since Schank’s paper. follow description main techniques these were based upon, such as deep learning reinforcement learning; two that have roots psychology. Next, discuss how psychologically plausible could become given AI’s ability to learn. then access question intelligent systems actually are. For example, are can solve human tests? conclude Shank's observation, all about generalization not particularly good at this, has, so far, withstood test time. Finally, consider what insights mean study individual differences intelligence. close further research vice versa, look forward fruitful interactions future.

Язык: Английский

---

Towards the next generation of recurrent network models for cognitive neuroscience

Current Opinion in Neurobiology, Год журнала: 2021, Номер 70, С. 182 - 192

Опубликована: Окт. 1, 2021

Язык: Английский

---

Modeling, Replicating, and Predicting Human Behavior: A Survey

ACM Transactions on Autonomous and Adaptive Systems, Год журнала: 2023, Номер 18(2), С. 1 - 47

Опубликована: Янв. 24, 2023

Given the popular presupposition of human reasoning as standard for learning and decision making, there have been significant efforts a growing trend in research to replicate these innate abilities artificial systems. As such, topics including Game Theory, Theory Mind, Machine Learning, among others, integrate concepts that are assumed components reasoning. These serve techniques understand behaviors humans. In addition, next-generation autonomous adaptive systems will largely include AI agents humans working together teams. To make this possible, require ability embed practical models behavior, allowing them not only technique “learn” but also actions users anticipate their so truly operate symbiosis with them. The main objective article is provide succinct yet systematic review important approaches two areas dealing quantitative behaviors. Specifically, we focus on (i) learn model or policy behavior through exploration feedback, such Reinforcement (ii) directly mechanisms reasoning, beliefs bias, without necessarily via trial error.

Язык: Английский

---

Reinforcement learning establishes a minimal metacognitive process to monitor and control motor learning performance

Nature Communications, Год журнала: 2023, Номер 14(1)

Опубликована: Июль 8, 2023

Abstract Humans and animals develop learning-to-learn strategies throughout their lives to accelerate learning. One theory suggests that this is achieved by a metacognitive process of controlling monitoring Although such also observed in motor learning, the aspect learning regulation has not been considered classical theories Here, we formulated minimal mechanism as reinforcement properties, which regulates policy for memory update response sensory prediction error while its performance. This was confirmed human experiments, subjective sense learning-outcome association determined direction up- down-regulation both speed retention. Thus, it provides simple, unifying account variations speeds, where monitors controls process.

Язык: Английский

---

Recurrent networks endowed with structural priors explain suboptimal animal behavior

Current Biology, Год журнала: 2023, Номер 33(4), С. 622 - 638.e7

Опубликована: Янв. 18, 2023

Язык: Английский

---

Specific connectivity optimizes learning in thalamocortical loops

Cell Reports, Год журнала: 2024, Номер 43(4), С. 114059 - 114059

Опубликована: Апрель 1, 2024

Thalamocortical loops have a central role in cognition and motor control, but precisely how they contribute to these processes is unclear. Recent studies showing evidence of plasticity thalamocortical synapses indicate for the thalamus shaping cortical dynamics through learning. Since signals undergo compression from cortex thalamus, we hypothesized that computational depends critically on structure corticothalamic connectivity. To test this, identified optimal promotes biologically plausible learning synapses. We found projections specialized communicate an efference copy output benefit while communicating modes highest variance working memory tasks. analyzed neural recordings mice performing grasping delayed discrimination tasks communication consistent with predictions. These results suggest orchestrates functionally precise manner structured

Язык: Английский

---

Challenges, evaluation and opportunities for open-world learning

Nature Machine Intelligence, Год журнала: 2024, Номер 6(6), С. 580 - 588

Опубликована: Июнь 24, 2024

Язык: Английский

---

AutomaticAI – A hybrid approach for automatic artificial intelligence algorithm selection and hyperparameter tuning

Expert Systems with Applications, Год журнала: 2021, Номер 182, С. 115225 - 115225

Опубликована: Май 26, 2021

Язык: Английский

---

Organizing memories for generalization in complementary learning systems

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2021, Номер unknown

Опубликована: Окт. 15, 2021

ABSTRACT Memorization and generalization are complementary cognitive processes that jointly promote adaptive behavior. For example, animals should memorize a safe route to water source generalize features allow them find new sources, without expecting paths exactly resemble previous ones. Memory aids by allowing the brain extract general patterns from specific instances were spread across time, such as when humans progressively build semantic knowledge episodic memories. This process depends on neural mechanisms of systems consolidation, whereby hippocampal-neocortical interactions gradually construct neocortical memory traces consolidating hippocampal precursors. However, recent data suggest consolidation only applies subset memories; why certain memories consolidate more than others remains unclear. Here we introduce novel network formalization highlights an overlooked tension between transfer generalization, resolve this postulating it generalization. We specifically show unregulated can be detrimental in unpredictable environments, whereas optimizing for generates high-fidelity, dual-system supporting both theory generalization-optimized produces transfers some components neocortex leaves dependent hippocampus. It thus provides normative principle reconceptualizing numerous puzzling observations field insight into how behavior benefits learning specialized memorization

Язык: Английский

---