Biology subjects

Flo, A.

Publications and source records attributed to Flo, A..

13 recordsLinked to original sources

bioRxiv2026

Why probe registration is crucial for infant fNIRS and HD-DOT? An evaluation across 49 newborns

Source-detector probe registration, i.e., the mapping of optode positions on the scalp to their true anatomical locations in head/brain space, is crucial for brain imaging with fNIRS or HD-DOT. While individual, MRI-based registration is desirable, it is not always feasible, motivating the development of alternative registration approaches. This study uses the STORM-Net registration protocol and quantifies the variation of cap alignment in a cohort of 49 newborns to evaluate how this affects reconstructed images using HD-DOT. Relative to a centrally aligned template, the cohort showed a median probe-position error of 10.12 mm, while the least and most misaligned subjects exhibited mean errors of 4.41 mm and 20.60 mm, respectively. The results demonstrate that the level of cap misalignment typically observed in newborn imaging can affect the spatial accuracy of reconstructed functional images, particularly in cases with larger probe displacements. These findings provide a quantitative basis for adopting more conservative interpretations of group-level functional brain activity and functional connectivity results, and strongly support the routine use of registration methods in infant HD-DOT studies.

neuroscience↗

bioRxiv2026

APICE-Py: An Open-Source MNE-Python Pipeline for Scalable EEG Preprocessing

Electroencephalography (EEG) is fundamental to cognitive neuroscience as it provides a direct measure of human neural activities with millisecond precision. Its noninvasive nature allows for the study of brain function across diverse age groups and experimental contexts--from newborns to adults, and from tightly controlled laboratory environments to more naturalistic real-world settings. However, EEG signals--especially those recorded from infants--are highly prone to noise and arti-facts, posing significant challenges for data analysis. To address these issues, we present APICE-Py (Automated Preprocessing for Infants Continuous EEG), an open-source preprocessing pipeline originally designed as a matlab toolbox for infant EEG and now re-implemented in Python to support scalable and flexible analysis across developmental and adult datasets. APICE-Py is built upon three core principles: (i) adaptive artifact detection on continuous data using data-driven thresholds rather than fixed cutoffs; (ii) hierarchical artifact correction, combining short-segment correction via Principal Component Analysis (PCA) with broader segment and continuous data correction using Spherical Spline Interpolation (SSI); and (iii)transparent reporting, providing comprehensive quality logs and decision-tracking to ensure reproducibility and informed analysis. We summarize the underlying algorithms, release an implementation compatible with common EEG formats, and demonstrate its use on three datasets spanning neonates, 5-month-old infants, and child-parent hyperscanned data, acquired using high-density wet electrodes and mobile gel-based EEG systems. When benchmarked against the original MATLAB implementation, APICE-Py achieved comparable levels of data quality and trial retention. While the original pipeline was developed for early developmental EEG (e.g., infants), we show that the pipeline further extends its applicability to both children and adult datasets, enabling robust preprocessing across a broad age range. Moreover, it supports data acquired using a variety of EEG configurations and experimental settings, highlighting its flexibility across age groups, hardware systems, and paradigms. The APICE-Py source code and documentation are freely available at https://github.com/neurokidslab/apice-py.

neuroscience↗

bioRxiv2025

From Syllables to Words: EEG Evidence of Different Age Trajectories in Speech Tracking and Statistical Learning in Infants at High and Low Likelihood for Autism

Delayed onset of canonical babbling and first words is often reported in infants later diagnosed with autism spectrum disorder. Identifying the neural mechanisms underlying language acquisition in autism is therefore critical to inform early diagnosis, prognosis, and intervention strategies. In this study, we investigated two speech processing mechanisms previously identified as atypical in children and adults with autism: the neural ability to track syllables; and statistical learning, the capacity to detect speech regularities beneath surface variability. We recorded 83 longitudinal high-density EEGs from 44 infants (2.5-22.6 months) at high (HL) and low (LL) likelihood for autism and assessed their verbal outcomes at 20 months. Neural entrainment was measured at syllable and word frequencies during exposure to a multi-speaker stream of concatenated tri-syllabic words, followed by a word recognition test using ERP recording. Our findings revealed reduced tracking abilities at the syllabic level in HL infants, a measure that correlated with verbal outcomes. While HL infants did not exhibit deficits in statistical learning itself, they displayed reduced novelty orientation during the word recognition test, indicated by a reduced late ERP. By contrast, multi-talker variability temporarily disrupted word segmentation around 12 months in LL infants, but not in HL infants, potentially reflecting decreased sensitivity to human voices variability in the HL group. These results emphasize the importance of longitudinal protocols employing online, implicit measures to track the hierarchical stages of speech processing in both HL and LL infants.

neuroscience↗

bioRxiv2025

Verbal Episodic Processing in Newborns

During the first period of life, human infants rapidly and effortlessly acquire the languages they are exposed to. Although memory is central to this process, the nature of early verbal memory systems and the factors that determine retention and forgetting remain largely unknown. Behavioural and brain measures have demonstrated memory formation in newborns. However, word traces fade in the face of acoustic overlap, leading to interference and forgetting. Here, we investigate whether speakers identity changes facilitate the separation into distinct acoustic episodes and the creation of non-overlapping verbal memories. Newborns (0-4 days-old) were tested in a familiarization-interference-test protocol, while neural cortical activity was recorded using functional Near-Infrared Spectroscopy (fNIRS). The results showed higher neural activation to novel words than to familiar ones during the test phase, indicating that the infants recognized the familiar words despite potentially interfering sounds. The recognition response was measured over the left inferior frontal gyrus (IFG) and superior temporal gyrus (STG) areas known to be crucial for encoding auditory information and language processing. The neural response also included the right IFG and STG, involved in interpreting vocal social cues and speaker recognition. The results indicate that speaker identity is a key feature in the formation of verbal memories from birth, facilitating separability, possibly through early source-content binding (i.e., what-who), a precursor to fully mature episodic memory. Impact StatementSpeaker identity is a distinguishing feature at birth and highlights the episodic nature of humans first-stored verbal memories.

neuroscience↗

bioRxiv2025

A proposal for unifying statistical learning at different scales: Long-Horizon Associative Learning

Sensory inputs are rich with temporal patterns that unfold across multiple timescales. Discovering these regularities is essential for predicting future events and navigating the environment efficiently. Numerous models have been proposed to account for learning at specific temporal scales; however, they are often designed in isolation and rely on narrowly tuned statistical measures, limiting their generalizability to other paradigms. In contrast, humans typically learn without prior knowledge of the underlying structure or the relevant timescale at which regularities occur. Here, we present a unifying account of statistical learning that spans a wide range of temporal dependencies, from adjacent and non-adjacent transitions to complex network structures. This model, Long-Horizon Associative Learning (L-HAL), offers a biologically grounded implementation of the successor representation, or equivalently, the Free Energy Minimization Model. Reanalyzing data from 11 previously published studies, we show that a single neural mechanism, based on graded temporal overlap of associative traces and governed by a single free parameter ({beta}), captures both local statistical regularities and higher-order structural properties. This initial domain-general associative learning process, emerging from the graded structure of associations, may later scaffold to higher-level operations such as grouping, categorization, rule abstraction, and memory formation. Overall, this framework offers a conceptual synthesis that bridges disparate strands of the statistical learning literature and reframes apparent paradigm-specific effects as different expressions of a common underlying computation

neuroscience↗

bioRxiv2024

Statistical learning beyond words in human neonates

Interest in statistical learning in developmental studies stems from the observation that 8-month-olds were able to extract words from a monotone speech stream solely using the transition probabilities (TP) between syllables (Saffran et al., 1996). A simple mechanism was thus part of the human infants toolbox for discovering regularities in language. Since this seminal study, observations on statistical learning capabilities have multiplied across domains and species, challenging the hypothesis of a dedicated mechanism for language acquisition. Here, we leverage the two dimensions conveyed by speech -speaker identity and phonemes- to examine (1) whether neonates can compute TPs on one dimension despite irrelevant variation on the other and (2) whether the linguistic dimension enjoys an advantage over the voice dimension. In two experiments, we exposed neonates to artificial speech streams constructed by concatenating syllables while recording EEG. The sequence had a statistical structure based either on the phonetic content, while the voices varied randomly (Experiment 1) or on voices with random phonetic content (Experiment 2). After familiarisation, neonates heard isolated duplets adhering, or not, to the structure they were familiarised with. In both experiments, we observed neural entrainment at the frequency of the regularity and distinct Event-Related Potentials (ERP) to correct and incorrect duplets, highlighting the universality of statistical learning mechanisms and suggesting it operates on virtually any dimension the input is factorised. However, only linguistic duplets elicited a specific ERP component, potentially an N400 precursor, suggesting a lexical stage triggered by phonetic regularities already at birth. These results show that, from birth, multiple input regularities can be processed in parallel and feed different higher-order networks. HIGHLIGHTSO_LIHuman neonates are sensitive to regularities in speech, encompassing both phonetic content and voice dimension. C_LIO_LIThere is no observed advantage for statistical computations of linguistic content over voice content. C_LIO_LIBoth speech dimensions are processed in parallel by distinct networks. C_LIO_LIPhonetic regularities evoked a specific ERP component in the post-learning phase, suggesting activations along the pathway to the lexicon. C_LI

neuroscience↗

bioRxiv2024

Long-horizon associative learning explains human sensitivity to statistical and network structures in auditory sequences

Networks are a useful mathematical tool for capturing the complexity of the world. In a previous behavioral study, we showed that human adults were sensitive to the high-level network structure underlying auditory sequences, even when presented with incomplete information. Their performance was best explained by a mathematical model compatible with associative learning principles, based on the integration of the transition probabilities between adjacent and non-adjacent elements with a memory decay. In the present study, we explored the neural correlates of this hypothesis via magnetoencephalography (MEG). Participants passively listened to sequences of tones organized in a sparse community network structure comprising two communities. An early difference (~150 ms) was observed in the brain responses to tone transitions with similar transition probability but occurring either within or between communities. This result implies a rapid and automatic encoding of the sequence structure. Using time-resolved decoding, we estimated the duration and overlap of the representation of each tone. The decoding performance exhibited exponential decay, resulting in a significant overlap between the representations of successive tones. Based on this extended decay profile, we estimated a long-horizon associative learning novelty index for each transition and found a correlation of this measure with the MEG signal. Overall, our study sheds light on the neural mechanisms underlying human sensitivity to network structures and highlights the potential role of Hebbian-like mechanisms in supporting learning at various temporal scales.

neuroscience↗

bioRxiv2024

The role of conscious attention in statistical learning: evidence from patients with impaired consciousness

The debate over whether conscious attention is necessary for statistical learning has produced mixed and conflicting results. Testing individuals with impaired consciousness may provide some insight, but very few studies have been conducted due to the difficulties associated with testing such patients. In this study, we examined the ability of patients with varying levels of consciousness disorders (DOC), including coma, unresponsive wakefulness syndrome, minimally conscious patients, and emergence from minimally conscious state patients, to extract statistical regularities from an artificial language composed of four randomly concatenated pseudowords. We used a methodology based on frequency tagging in EEG, which was developed in our previous studies on speech segmentation in sleeping neonates. Our study had two main objectives: firstly, to assess the automaticity of the segmentation process and explore correlations between the level of covert consciousness and the abilities to extract statistical regularities, second, to explore a potential new diagnostic indicator to aid in patient management by examining the correlation between successful statistical learning markers and consciousness level. We observed that segmentation abilities were preserved in some minimally conscious patients, suggesting that statistical learning is an inherently automatic low-level process. Due to significant inter-individual variability, word segmentation may not be a sufficiently robust candidate for clinical use, unlike temporal accuracy of auditory syllable responses, which correlates strongly with coma severity. Therefore, we propose that frequency tagging of an auditory stimulus train, a simple and robust measure, should be further investigated as a possible metric candidate for DOC diagnosis.

neuroscience↗

bioRxiv2022

Humans parsimoniously represent auditory sequences by pruning and completing the underlying network structure

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

neuroscience↗

bioRxiv2021

Tracking transitional probabilities and segmenting auditory sequences are dissociable processes in adults and neonates

Since speech is a continuous stream with no systematic boundaries between words, how do pre-verbal infants manage to discover words? A proposed solution is that they might use the transitional probability between adjacent syllables, which drops at word boundaries. Here, we tested the limits of this mechanism by increasing the size of the word-unit to 4 syllables, and its automaticity by testing asleep neonates. Using markers of statistical learning in neonates EEG, compared to adult behavioral performances in the same task, we confirmed that statistical learning is automatic enough to be efficient even in sleeping neonates. But we also revealed that: 1) Successfully tracking transition probabilities in a sequence is not sufficient to segment it 2) Prosodic cues, as subtle as subliminal pauses, enable to recover segmenting capacities 3) Adults and neonates capacities are remarkably similar despite the difference of maturation and expertise. Finally, we observed that learning increased the similarity of neural responses across infants, providing a new neural marker to monitor learning. Thus, from birth, infants are equipped with adult-like tools, allowing to extract small coherent word-like units within auditory streams, based on the combination of statistical analyses and prosodic cues.

neuroscience↗

bioRxiv2021

From computing transition probabilities to word recognition in sleeping neonates, a two-step neural tale

Extracting statistical regularities from the environment is a primary learning mechanism, which might support language acquisition. While it is known that infants are sensitive to transition probabilities between syllables in continuous speech, the format of the encoded representation remains unknown. Here we used electrophysiology to investigate how 31 full-term neonates process an artificial language build by the random concatenation of four pseudo-words and which information they retain. We used neural entrainment as a marker of the regularities the brain is tracking in the stream during learning. Then, we compared the evoked-related potentials (ERP) to different triplets to further explore the format of the information kept in memory. After only two minutes of familiarization with the artificial language, we observed significant neural entrainment at the word rate over left temporal electrodes compared to a random stream, demonstrating that sleeping neonates automatically and rapidly extracted the word pattern. ERPs significantly differed between triplets starting or not with the correct first syllable in the test phase, but no difference was associated with later violations in transition probabilities, revealing a change in the representation format between segmentation and memory processes. If the transition probabilities were used to segment the stream, the retained representation relied on syllables ordinal position, but still without a complete representation of the words at this age. Our results revealed a two-step learning strategy, probably involving different brain regions.

neuroscience↗

bioRxiv2021

Automated Pipeline for Infants Continuous EEG (APICE): a flexible pipeline for developmental studies

Infant electroencephalography (EEG) presents several challenges compared with adult data. Recordings are typically short. Motion artifacts heavily contaminate the data. The EEG neural signal and the artifacts change throughout development. Traditional data preprocessing pipelines have been developed mainly for event-related potentials analyses, and they required manual steps, or use fixed thresholds for rejecting epochs. However, larger datasets make the use of manual steps infeasible, and new analytical approaches may have different preprocessing requirements. Here we propose an Automated Pipeline for Infants Continuous EEG (APICE). APICE is fully automated, flexible, and modular. Artifacts are detected using multiple algorithms and adaptive thresholds, making it suitable to different age groups and testing procedures without redefining parameters. Artifacts detection and correction of transient artifacts is performed on continuous data, allowing for better data recovery and providing flexibility (i.e., the same preprocessing is usable for different analyses). Here we describe APICE and validate it using two infant datasets of different ages tested in different experimental paradigms. We also tested the combination of APICE with common data cleaning methods such as Independent Component Analysis and Denoising Source Separation. APICE uses EEGLAB and compatible custom functions. It is freely available at https://github.com/neurokidslab/eeg_preprocessing, together with example scripts.

neuroscience↗

bioRxiv2021

Orthogonal neural codes for phonetic features in the infant brain

Creating invariant representations from an ever-changing speech signal is a major challenge for the human brain. Such an ability is particularly crucial for preverbal infants who must discover the phonological, lexical and syntactic regularities of an extremely inconsistent signal in order to acquire language. Within visual perception, an efficient neural solution to overcome signal variability consists in factorizing the input into orthogonal and relevant low-dimensional components. In this study we asked whether a similar neural strategy grounded on phonetic features is recruited in speech perception. Using a 256-channel electroencephalographic system, we recorded the neural responses of 3-month-old infants to 120 natural consonant-vowel syllables with varying acoustic and phonetic profiles. To characterize the specificity and granularity of the elicited representations, we employed a hierarchical generalization approach based on multivariate pattern analyses. We identified two stages of processing. At first, the features of manner and place of articulation were decodable as stable and independent dimensions of neural responsivity. Subsequently, phonetic features were integrated into phoneme-identity (i.e. consonant) neural codes. The latter remained distinct from the representation of the vowel, accounting for the different weights attributed to consonants and vowels in lexical and syntactic computations. This study reveals that, despite the paucity of articulatory motor plans and productive skills, the preverbal brain is already equipped with a structured phonetic space which provides a combinatorial code for speech analysis. The early availability of a stable and orthogonal neural code for phonetic features might account for the rapid pace of language acquisition during the first year. SIGNIFICANCE STATEMENTFor adults to comprehend spoken language, and for infants to acquire their native tongue, it is fundamental to perceive speech as a sequence of stable and invariant segments despite its extreme acoustic variability. We show that the brain can achieve such a critical task thanks to a factorized representational system which breaks down the speech input into minimal and orthogonal components: the phonetic features. These elementary representations are robust to signal variability and are flexibly recombined into phoneme-identity percepts in a secondary processing phase. In contradiction with previous accounts questioning the availability of authentic phonetic representations in early infancy, we show that this neural strategy is implemented from the very first stages of language development.

neuroscience↗