(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Functional reorganization of brain regions supporting artificial grammar learning across the first half year of life [1]

['Lin Cai', 'Department Of Electronics', 'Electrical Engineering', 'Keio University', 'Yokohama', 'Global Research Center For Logic', 'Sensitivity', 'Global Research Institute', 'Tokyo', 'Takeshi Arimitsu']

Date: 2024-10

Pre-babbling infants can track nonadjacent dependencies (NADs) in the auditory domain. While this forms a crucial prerequisite for language acquisition, the neurodevelopmental origins of this ability remain unknown. We applied functional near-infrared spectroscopy in neonates and 6- to 7-month-old infants to investigate the neural substrate supporting NAD learning and detection using tone sequences in an artificial grammar learning paradigm. Detection of NADs was indicated by left prefrontal activation in neonates while by left supramarginal gyrus (SMG), superior temporal gyrus (STG), and inferior frontal gyrus activation in 6- to 7-month-olds. Functional connectivity analyses further indicated that the neonate activation pattern during the test phase benefited from a brain network consisting of prefrontal regions, left SMG and STG during the rest and learning phases. These findings suggest a left-hemispheric learning-related functional brain network may emerge at birth and serve as the foundation for the later engagement of these regions for NAD detection, thus, providing a neural basis for language acquisition.

Introduction

Humans are born with sophisticated auditory abilities, possibly shaped by prenatal experience and a relatively mature auditory system [1,2]. Studies with newborn infants have thus demonstrated impressive abilities in the auditory domain including discrimination of stimuli based on various auditory features [3,4], auditory learning of novel sounds [5,6], and computation of more complex sound sequences [7–10]. The latter is of particular relevance for language acquisition, as human syntax relies on the ability to decode a hierarchical structure from sequentially organized auditory input. The sequential auditory input can involve more or less complex dependency patterns ranging from simple adjacent dependencies, i.e., the relation of 2 consecutive stimuli, to multiple embedded nonadjacent dependencies. Nonadjacent dependencies (NADs) are important for language because they allow the meaningful relation of elements across a distance, enabling the formation of complex and hierarchical sentence structures [11]. For example, the sentence “The baby smiles” that contains an NAD between the verb suffix–s and the noun, can be extended to “The baby who is sitting on her mother’s lap smiles” that contains a further NAD between the suffix–ing and the auxiliary “is,” creating a nested structure of NADs. In order to analyze such sentences, the dependent elements have to be stored and retrieved across variable distances. The basic computational mechanisms underlying the ability to detect NADs have been studied both using linguistic as well as nonlinguistic materials [12–15]. The ability to learn NADs seems to be present from early on, both when encoded in speech [16] and when encoded in computationally simpler sine-tone sequences [12,13]. Even nonhuman animals seem to be able to detect NADs in some cases [17,18] suggesting that the basic ability may not be unique to humans and constitutes an important computational basis for language which is present early in development and potentially rooted in our primate ancestors. What is not known, though, is (i) from which age onwards the ability to detect NADs is present in humans; and (ii) which brain areas are involved during learning and detection of NADs in those early developmental stages.

The ability to learn NADs in the auditory domain and its developmental trajectory have been subject to intense investigations in infants and young children [11,19]. Previous behavioral investigations have shown that a sensitivity to NADs, embedded in natural or artificial language, appears after the first year of life [20–23] (around 15 to 19 months), and the ability to discriminate nonadjacent repetitions (e.g., ABA) from other patterns (e.g., ABB or ABC) is present in 5- to 7-month-old infants [24,25]. However, such findings may be constrained by the restrictions of behavioral measures, i.e., the dependence of those measures on overtly observable responses. Electrophysiological methods, for example, present a more fine-grained approach toward directly probing the recognition of NADs, and have correspondingly validated infants’ earlier sensitivity to NADs [15,26]. For example, an event-related potential (ERP) study using the familiarization-test paradigm suggests that 4-month-old monolingual infants can passively learn NADs from a foreign language, indicated by a late positive ERP effect in response to NAD violations after learning [15]. Several other electrophysiological studies attested NAD learning before the first birthday [12,26,27]. While the learning of adjacent dependencies has been demonstrated already in newborns [9,10], there is no evidence so far for the learning of NADs in newborns.

Unfortunately, electrophysiological evidence alone cannot provide precise information concerning the question of which brain regions are involved in NAD learning. Recent functional magnetic resonance imaging (fMRI) advances in adults reveal that Broca’s region is involved in the processing of NADs [28–30]. Similarly, van der Kant and colleagues [13] found, using functional near-infrared spectroscopy (fNIRS), that NAD violation detection was subserved by a left-hemispheric temporo-fronto-parietal network for linguistic stimuli in 2-year-olds. On the other hand, 3-year-olds, but not 2-year-olds were able to learn NADs consisting of nonlinguistic tone stimuli engaging a bilateral temporo-parietal network. However, there is no evidence that NAD learning in tone stimuli develops later than in language stimuli as demonstrated by a recent ERP evidence showing that infants who were only 5 months of age can track embedded NAD structures between simple sine tone stimuli [12]. A study focusing on the specific role of prosodic cues for NAD learning in 9-month-old infants revealed the contribution of frontotemporal brain regions in the more difficult, monotonous condition and the engagement of temporal brain areas in the presence of prosodic cues [14]. Yet, this study leaves unclear whether the infants learned NADs between 2 specific elements or only positional regularities. Together, these findings indicate that NAD learning may be present early in development and that neural networks that include classic language areas are involved when correct and incorrect NADs are successfully discriminated from early childhood onwards, at least in the case of linguistic stimuli. However, it is not known whether similar neural networks support NAD learning and detection from birth and how they develop across the first half year of life.

A growing body of research demonstrates that neonates possess an adult-like ventral pathway for language, which connects the anterior temporal lobe with the ventrolateral prefrontal cortex by the extreme capsule [31]. Likewise, the part of the dorsal pathway connecting the temporal cortex to the premotor cortex is also already present at birth. These structural connections may serve as a neural basis for rule learning [31], maternal speech perception [32], and phonological learning [5] in neonates. By contrast, the part of the dorsal pathway connecting the temporal cortex to Broca’s area develops much later, but it catches up during the first postnatal months [33]. Note that this connection has been argued to be involved in parsing more complex syntactic structures during childhood [31,34]. Taking into account both the early functionality of the prefrontal cortex and its relative immaturity, a very recent review [35] proposed that infants learn actively from their environment early on and that this is supported by their (anatomically) immature prefrontal cortex. Taken together, these studies reveal that an intriguing functional brain network consisting of several distributed key brain regions for language processing has emerged since birth. Further, they suggest that the remarkable learning ability of NADs in older infants reported above could be supported by the ventral and dorsal pathways despite the relative immaturity of the prefrontal cortex (and its dorsal connection to posterior brain areas).

Because NAD learning seems to be present early in development and distributed across various vertebrate species and also relevant for core processes of language, as outlined above, we hypothesize that (i) humans are born with the ability to track NADs; and (ii) the brain regions that support this process may include areas that are also involved in early language processing including left hemispheric fronto-temporal networks. In the current study, we examined the neurodevelopmental origins of NAD learning and detection using nonlinguistic tone sequences in order to further understand the ontogenesis of human language. We chose nonlinguistic stimuli as we aimed to test the ability to learn NADs from auditory input in an experience-independent manner and as we further wanted to be able to compare the results to previous findings in nonhuman animals [17]. Furthermore, it is well established from previous studies that sequential learning in infancy works with both linguistic and nonlinguistic auditory stimuli [12,36,37]. The NADs in the current study were realized as dependencies between nonadjacent tone categories mimicking syntactic relations between word categories, as for example between nouns and verbs with agreement marking. We adopted the fNIRS technique to test neonates (1 to 5 days old) in Experiment 1 and infants (6 to 7 months old) in Experiment 2 using an auditory artificial grammar learning paradigm (Fig 1B). We included neonates to test neural processes involved in learning and detection of NADs, as outlined above, at birth. Infants aged 6 to 7 months were chosen because a more stable global neural system for language processing appears to be established at this age compared to younger ages, as judged by their resting-state connectivity from the language areas [38]. Additionally, we needed an infant group older than 3 to 4 months who would show a reliable response to detect NADs [15,26] to compare the neonate group. Note that our experimental design with the chosen participant groups allowed the assessment of brain regions during NAD detection in both age groups, while the assessment of regions involved during learning could be assessed in neonates.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. Stimuli and experimental design. (A) The pitch contours of 6 acoustic categories. (B) Experimental design. In the Learning phase, standard triplets were repeated 60 times. In the Test phase, 2 types of target trials (Correct and Incorrect conditions) were separated by a jittering baseline condition in which the same standard triplets with the Learning phase were presented. (C) Measurement phases for 2 experiments. https://doi.org/10.1371/journal.pbio.3002610.g001

In both experiments, the stimuli were composed of a series of perceptually simple tone sequences, each of which comprised 3 frequency-modulated sine tones from 6 categories of pitch contour (A, B, C, D, X1, and X2; see Fig 1A). Each sine tone category had various pitch variants which were used to avoid stimulus-specific learning. In the learning (Learning) phase, participants were exposed to 60 standard tone triplets conforming to NAD rules (e.g., AXB or CXD grammar) for approximately 6 min. Subsequently, during the test (Test) phase, they were presented with the familiar standard sequences and sequences comprised of novel pitch variants of the acoustic categories heard during the Learning phase arranged into either “Correct” triplets (e.g., AXB or CXD) or “Incorrect” triplets (e.g., AXD or CXB), thereby testing whether individuals were able to generalize the NAD rules from the Learning phase and detect violations. In Experiment 1, we additionally recorded the hemodynamic activities of neonates during the pre-task resting-state (Pre-Rest) phase, the Learning phase, and the post-task resting-state (Post-Rest) phase. Pre-Rest and Post-Rest phases were inserted before and after Learning and Test phases, respectively. The 2 resting-state phases allowed us to examine changes in functional connectivity (FC) from Pre-Rest to Learning phases or from Pre-Rest to Post-Rest phases, as well as the relationship between the FC changes and the activation to NAD violations during the Test phase (Fig 1C). In Experiment 2, we exposed 6- to 7-month-olds to Learning and Test phases without resting-state phases but we performed fNIRS scans only during the Test phase (Fig 1C). The decision to limit scan time in this experiment was based on the increased level of physical activity and thus potentially decreased compliance with the procedure in this age group.

Thus, the first aim of Experiment 1 was to reveal whether human neonates can extract NAD relations and generalize them to novel tone sequences, which would be indicated by a greater neural response to incorrect sequences than to correct ones during the Test phase. The second aim of Experiment 1 was to identify, in the case of successful NAD violation detection, the brain networks underlying NAD learning at birth by examining FC changes (Learning minus Pre-Rest or Post-Rest minus Pre-Rest) and their correlations to cerebral responses to learned NAD relations as measured during the Test phase. The aim of Experiment 2 was to examine what brain networks underlie the successful detection of NADs in 6- to 7-month-old infants. By combining these 2 experiments, we could shed light on the emergence and development of the functional brain network underlying the detection of NADs across the first half year of life.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002610

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.