Structure transfer and consolidation in visual implicit learning DOI Open Access
Dominik Garber, József Fiser

Published: Aug. 27, 2024

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 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, switched their behaviour pattern as did while still remaining implicit. This effect was specific not after non-sleep Our results highlight similarities generalizable relying consolidation restructuring internal representations.

Language: Английский

A theory of memory for binary sequences: Evidence for a mental compression algorithm in humans DOI Creative Commons
Samuel Planton, Timo van Kerkoerle, Leïla Abbih

et al.

PLoS Computational Biology, Journal Year: 2021, Volume and Issue: 17(1), P. e1008598 - e1008598

Published: Jan. 19, 2021

Working memory capacity can be improved by recoding the memorized information in a condensed form. Here, we tested theory that human adults encode binary sequences of stimuli using an abstract internal language and recursive compression algorithm. The predicts psychological complexity given sequence should proportional to length its shortest description proposed language, which capture any nested pattern repetitions alternations limited number instructions. Five experiments examine predict adults’ for variety auditory visual sequences. We probed violation paradigm participants attempted detect occasional violations otherwise fixed sequence. Both subjective ratings objective detection performance were well predicted our theoretical measure complexity, simply reflects weighted sum elementary instructions digits formula captures language. While simpler transition probability model, when as single predictor statistical analyses, accounted significant variance data, goodness-of-fit with data significantly language-based was included while explained model largely decreased. Model comparison also showed provides better fit than six alternative previously models encoding. support hypothesis that, beyond extraction knowledge, coding relies on language-like structures.

Language: Английский

Citations

72
Structure transfer and consolidation in visual implicit learning DOI Open Access
Dominik Garber, József Fiser

Published: Jan. 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.

Language: Английский

Citations

1
Humans parsimoniously represent auditory sequences by pruning and completing the underlying network structure DOI Creative Commons
Lucas Benjamin, Ana Fló, Fosca Al Roumi

et al.

eLife, Journal Year: 2023, Volume and Issue: 12

Published: May 2, 2023

Successive auditory inputs are rarely independent, their relationships ranging from local transitions between elements to hierarchical and nested representations. In many situations, humans retrieve these dependencies even limited datasets. However, this learning at multiple scale levels is poorly understood. Here, we used the formalism proposed by network science study representation of higher-order structures interaction in sequences. We show that human adults exhibited biases perception elements, which made them sensitive high-order such as communities. This behavior consistent with creation a parsimonious simplified model evidence they receive, achieved pruning completing elements. observation suggests brain does not rely on exact memories but world. Moreover, bias can be analytically modeled memory/efficiency trade-off. correctly accounts for previous findings, including transition probabilities well structures, unifying sequence across scales. finally propose putative implementations bias.

Language: Английский

Citations

14
Statistical learning occurs during practice while high-order rule learning during rest period DOI Creative Commons
Romain Quentin, Lison Fanuel,

Mariann Kiss

et al.

npj Science of Learning, Journal Year: 2021, Volume and Issue: 6(1)

Published: July 1, 2021

Abstract Knowing when the brain learns is crucial for both comprehension of memory formation and consolidation developing new training neurorehabilitation strategies in healthy patient populations. Recently, a rapid form offline learning during short rest periods has been shown to account most procedural learning, leading hypothesis that mainly between practice periods. Nonetheless, several subcomponents not disentangled previous studies investigating dynamics, such as acquiring statistical regularities task, or else high-order rules regulate its organization. Here we analyzed 506 behavioral sessions implicit visuomotor deterministic probabilistic sequence tasks, allowing distinction general skill rule learning. Our results show temporal dynamics apparently simultaneous processes differ. While acquired offline, evidenced online. These findings open avenues on short-scale reveal fundamental former benefiting from online evidence accumulation latter requiring consolidation.

Language: Английский

Citations

26
Brain-imaging evidence for compression of binary sound sequences in human memory DOI Creative Commons
Fosca Al Roumi, Samuel Planton, Liping Wang

et al.

eLife, Journal Year: 2023, Volume and Issue: 12

Published: Nov. 1, 2023

According to the language-of-thought hypothesis, regular sequences are compressed in human memory using recursive loops akin a mental program that predicts future items. We tested this theory by probing for 16-item made of two sounds. recorded brain activity with functional MRI and magneto-encephalography (MEG) while participants listened hierarchy variable complexity, whose minimal description required transition probabilities, chunking, or nested structures. Occasional deviant sounds probed participants’ knowledge sequence. predicted task difficulty would be proportional complexity derived from length our formal language. Furthermore, should increase learned sequences, decrease deviants. These predictions were upheld both fMRI MEG, indicating sequence highly dependent on structure become weaker delayed as increases. The proposed language recruited bilateral superior temporal, precentral, anterior intraparietal, cerebellar cortices. regions overlapped extensively localizer mathematical calculation, much less spoken written processing. propose these areas collectively encode repetitions variations their composition into

Language: Английский

Citations

10
Neurophysiological and functional neuroanatomical coding of statistical and deterministic rule information during sequence learning DOI
Ádám Takács, Andrea Kóbor, Zsófia Kardos

et al.

Human Brain Mapping, Journal Year: 2021, Volume and Issue: 42(10), P. 3182 - 3201

Published: April 2, 2021

Abstract Humans are capable of acquiring multiple types information presented in the same stream. It has been suggested that at least two parallel learning processes important during sequential patterns—statistical and rule‐based learning. Yet, neurophysiological underpinnings these not fully understood. To differentiate between simultaneous mechanisms single trial level, we apply a temporal EEG signal decomposition approach together with sLORETA source localization method to delineate whether distinct statistical codes can be distinguished data related functional neuroanatomical structures. We demonstrate concomitant but aspects coded N2 time window play role mechanisms: mismatch detection response control underlie learning, respectively, albeit different levels time‐sensitivity. Moreover, effects temporally decomposed clusters neural activity also differed from each other sources. Importantly, right inferior frontal cortex (BA44) was specifically implicated visuomotor confirming its acquisition transitional probabilities. In contrast, associated prefrontal gyrus (BA6). The results show how operate level orchestrated by cortical areas. current findings deepen our understanding on humans stimulus stream fashion.

Language: Английский

Citations

21
Lexical surprisal shapes the time course of syntactic structure building DOI Open Access
Sophie Slaats, Antje S. Meyer, Andrea E. Martin

et al.

Published: Jan. 18, 2024

When we understand language, recognize words and combine them into sentences. In this paper, explore the hypothesis that listeners use probabilistic information about to build syntactic structure. Recent work has shown lexical probability structure both modulate delta-band (0-4 Hz) neural signal. Here, investigated whether encoding of changes as a function distributional properties word. To end, analyzed MEG data 24 native speakers Dutch who listened three fairytales with total duration 49 minutes. Using temporal response functions cumulative model-comparison approach, evaluated contributions features variance in This revealed surprisal values (a feature), well bottom-up node counts feature) positively contributed model Subsequently, compared responses feature between high- low values. delay consequence value word: high were associated delayed by 150 190 milliseconds. The was not affected word duration, did have origin. These findings suggest brain uses infer structure, highlight an importance for role time process.

Language: Английский

Citations

3
Sequence-dependent predictive coding during the learning and rewiring of skills DOI
Ádám Takács, Teodóra Vékony, Felipe Pedraza

et al.

Cerebral Cortex, Journal Year: 2025, Volume and Issue: 35(2)

Published: Feb. 1, 2025

Abstract In the constantly changing environment that characterizes our daily lives, ability to predict and adapt new circumstances is crucial. This study examines influence of sequence knowledge adaptiveness on predictive coding in skill learning rewiring. Participants were exposed two different visuomotor sequences with overlapping probabilities. By applying temporal decomposition multivariate pattern analysis, we dissected neural underpinnings across levels signal coding. The provides neurophysiological evidence for shaping coding, revealing these are intricately linked predominantly manifest at abstract motor levels. These findings challenge traditional view a competitive relationship between context knowledge, suggesting instead hierarchical integration where their properties processed simultaneously. facilitates adaptive reuse existing face learning. shedding light mechanisms sequences, this research contributes deeper understanding how brain navigates adapts environmental changes, offering insights into foundational processes underlie adaptation dynamic contexts.

Language: Английский

Citations

0
Structure transfer and consolidation in visual implicit learning DOI Open Access
Dominik Garber, József Fiser

Published: March 11, 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.

Language: Английский

Citations

0
Auditory facilitation in deterministic versus stochastic worlds DOI Creative Commons

Berfin Bastug,

Urte Roeber, Erich Schröger

et al.

Cognitive Neuroscience, Journal Year: 2025, Volume and Issue: unknown, P. 1 - 7

Published: April 29, 2025

The brain learns statistical regularities in sensory sequences, enhancing behavioral performance for predictable stimuli while impairing unpredictable stimuli. While previous research has shown that violations of non-informative hinder task performance, it remains unclear whether but task-irrelevant structures can facilitate performance. In a tone duration discrimination task, we manipulated the pitch dimension by varying transition probabilities (TP) between successive frequencies. Participants judged duration, sequences were either deterministic (a rule-governed pattern, TP = 1) or stochastic (no discernible 1/number levels). was and did not predict duration. Results showed reaction times (RTs) significantly faster suggesting predictability still facilitates RTs also two-tone compared to eight-tone likely due reduced memory load. These findings suggest learning benefits extend beyond task-relevant dimensions, supporting predictive coding framework which integrates input optimize cognitive processing.

Language: Английский

Citations

0