Article
Audiology & Speech-Language Pathology
Anthony J. Angwin, Samuel R. Armstrong, Courtney Fisher, Paola Escudero
Summary: This study examines the rapid acquisition of novel word meanings through cross-situational statistical word learning (CSWL) using event-related potentials. The findings show that novel words and familiar words elicit similar N400 effects but with different hemispheric distributions.
BRAIN AND LANGUAGE
(2022)
Article
Neurosciences
Oscar Ferrante, Alexander Zhigalov, Clayton Hickey, Ole Jensen
Summary: Visual attention is affected by past experiences, and expectations about distractor locations can be learned and reduced through statistical learning. Using magnetoencephalography (MEG), it was found that early visual cortex showed reduced neural excitability at retinotopic locations associated with higher distractor probabilities. This suggests that proactive mechanisms of attention are involved in predictive distractor suppression and are associated with altered neural excitability in early visual cortex.
JOURNAL OF NEUROSCIENCE
(2023)
Article
Psychology, Biological
Tomoko Okano, Tatsuya Daikoku, Yoshikazu Ugawa, Kazuaki Kanai, Masato Yumoto
Summary: Statistical learning helps understand structured information in language and music, with the brain predicting future states based on transition probabilities to minimize sensory reactions and derive entropy. Amplitude differences in event-related neural responses reflect statistical learning effects, with transition-probability ratios finely tuning early cognitive processes.
INTERNATIONAL JOURNAL OF PSYCHOPHYSIOLOGY
(2021)
Article
Psychology, Experimental
Rebecca K. Reh, Takao K. Hensch, Janet F. Werker
Summary: Research shows that infants' sensitivity to distributional statistics in speech sounds declines as the period of perceptual attunement closes. EEG studies reveal that neuronal responses to phonemes /ra/ and /la/ are influenced by exposure to bimodal distribution in 5- and 9-month-old infants, but not in 12-month-olds.
Article
Neurosciences
A. Nora, H. Renvall, M. Ronimus, J. Kere, H. Lyytinen, R. Salmelin
Summary: Developmental dyslexia is a specific learning disorder characterized by difficulties in reading and spelling, often associated with phonological processing and learning impairments. Research shows that children with dyslexia have reduced activation in the temporal cortices during the recognition and memory of new word forms, particularly in the left hemisphere phonological areas.
Article
Computer Science, Artificial Intelligence
Neta Shoham, Haim Avron
Summary: The impressive performance of modern machine learning models relies on training them with a large amount of labeled data. However, obtaining such data is often limited or expensive, so curating the training set becomes desirable. Classical experimental design methods are not applicable to overparameterized machine learning models. This paper proposes a design strategy suitable for overparameterized models and demonstrates its applicability in the context of deep learning.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Neurosciences
Neus Ramos-Escobar, Matti Laine, Mariana Sanseverino-Dillenburg, David Cucurell, Clement Francois, Antoni Rodriguez-Fornells
Summary: The study explores the temporal brain dynamics of explicit verbal associative learning between unfamiliar items in humans. It found that N400 amplitude modulations reflect the emergence of novel lexical traces even without semantic information, suggesting a balance between domain-general and language-specific mechanisms in specific word learning steps.
Article
Psychology, Biological
Jiayan Chen, Baoguo Chen
Summary: This study investigates the process of learning new meanings for known words in a second language through the use of ERP technique. It finds that the new meaning needs offline consolidation to be integrated into long-term semantic memory, and that semantic relatedness promotes the integration process by facilitating the update of existing meaning representation.
Article
Neurosciences
Timothy Trammel, Natalia Khodayari, Steven J. Luck, Matthew J. Traxler, Tamara Y. Swaab
Summary: Machine-learning (ML) decoding methods have been found valuable in analyzing information represented in electroencephalogram (EEG) data. However, a systematic comparison of major ML classifiers for EEG decoding in neuroscience studies of cognition is lacking. This study compared three major ML classifiers, SVM, LDA, and RF, using EEG data from visual word-priming experiments. The results showed that SVM outperformed the other methods in both experiments and on all measures.
Article
Behavioral Sciences
Sven Braeutigam, Jessica Clare Scaife, Tipu Aziz, Rebecca J. Park
Summary: This study used MEG to observe the neuronal response of anorexia nervosa patients undergoing deep brain stimulation (DBS) during a food wanting task. The results showed that DBS can modulate brain activity and reduce the psychopathology of eating disorders in patients.
FRONTIERS IN BEHAVIORAL NEUROSCIENCE
(2022)
Article
Neurosciences
Elena Boto, Vishal Shah, Ryan M. Hill, Natalie Rhodes, James Osborne, Cody Doyle, Niall Holmes, Molly Rea, James Leggett, Richard Bowtell, Matthew J. Brookes
Summary: Optically-pumped magnetometers (OPMs) are a viable alternative to superconducting sensors for magnetoencephalography (MEG), with advantages such as flexibility, uniform coverage, better data quality, and lower cost. This study introduces a novel triaxial OPM sensor that can accurately and sensitively measure magnetic fields, showing comparable performance to conventional OPMs. A child-friendly 3D-printed OPM-helmet is also proposed, demonstrating the feasibility of triaxial measurement in pediatric populations.
Article
Neurosciences
Tim M. Tierney, Nicholas Alexander, Stephanie Mellor, Niall Holmes, Robert Seymour, George C. O'Neill, Eleanor A. Maguire, Gareth R. Barnes
Summary: A proposed method models magnetic interference as a spatially homogeneous field, reducing sensor variance and improving statistical power without requiring detailed neuroanatomy or sensor positioning information. The method has been validated in an auditory experiment, showing an increase in sensor SNR by a factor of 3, demonstrating its potential as a powerful preprocessing step for arrays of optically pumped magnetometers.
Article
Psychology, Multidisciplinary
Iris Broedelet, Paul Boersma, Judith Rispens
Summary: This paper investigates the role of distributional learning in learning novel object categories in school-aged children. The results demonstrate that the frequency distribution of input has a significant influence on the formation of novel object categories in the categorization of sensory stimuli.
FRONTIERS IN PSYCHOLOGY
(2022)
Article
Acoustics
Mousumi Malakar, Ravindra B. Keskar
Summary: Phonemes are the smallest distinct sound units in a language, crucial for automatic speech recognition systems. Machine learning techniques play a significant role in overcoming barriers to phoneme recognition and are now favored over traditional methods.
SPEECH COMMUNICATION
(2021)
Article
Chemistry, Analytical
Dimitrios Pantazis, Amir Adler
Summary: This paper presents a deep learning solution for localizing MEG brain signals, demonstrating improved performance over the traditional RAP-MUSIC algorithm in specific scenarios. The deep learning models show robustness to forward model errors and significantly reduce computation time, making real-time MEG source localization possible.