How Meaning Shapes a Rule: Neural Signatures of Morphological Learning and Generalization
Abstract
Rule generalization is one of the most useful tools our mind uses to facilitate learning. In language, we apply it to connect what words mean to what they are made of, such as their suffixes. This is a morphosemantic relation, and it helps us process words we already know and easily acquire new words by composing their meaning. How is this relation acquired, and how does it manifest into generalization for understanding new words? In this study, we investigated the brain basis of these questions through an English-based artificial language study, where learners read sentences to learn a suffix-meaning relation of new words like “thinkle”. Words were either meaningful (places or things) or had no clear meaning. Each suffix family had majority and minority meaning words that participants had to keep track of for testing. For meaningful words in training, we found that higher performance scores predicted higher selective engagement in the left inferior frontal gyrus, specifically the pars triangularis and pars opercularis. In testing, both learned and generalized words, both increased lateral prefrontal cortex engagement. Meanwhile, words with no clear meaning had a much more diffuse brain engagement, recruiting bilateral temporoparietal regions. This resembles the processing of abstract words from previous literature. We found no difference between majority and minority meaning words, especially with the variance in participants’ performance strategies, but found differences in what predicted significant brain activity based on strategy difference. This study grounds itself in the reality of word and affixal semantics with previous research on language learning and universals. Ultimately, it lends a learning paradigm that resembles natural language learning across many languages to carefully investigate its brain basis, while addressing an understudied aspect of rule generalization: ambiguity.
Related
- Large language models poorly capture how word order shapes representations of sentence meaning in the human brain.
- How Patients' Inquiry Shapes Meaning in Primary Care Consultations: A Qualitative Study Using Own-Point-of-View Video-Stimulated Recall.
- Algorithmic Geometry Shapes the Biological Meaning of Thalamocortical Structural Covariance
- Evaluation design shapes estimates of cross-donor generalization in single-cell classification of cell-type
- Protocells reproduce morphological and chemical signatures of early microfossils
BibTeX
@misc{azar2026meaning,
title = {{How Meaning Shapes a Rule: Neural Signatures of Morphological Learning and Generalization}},
author = {Marianne Azar and Alec Marantz},
howpublished = {PsyArXiv},
year = {2026},
month = sep,
doi = {10.31234/osf.io/vu5k4_v4},
url = {https://osf.io/preprints/psyarxiv/vu5k4_v4/},
note = {CC0},
}
RIS
TY - UNPB
TI - How Meaning Shapes a Rule: Neural Signatures of Morphological Learning and Generalization
AU - Marianne Azar
AU - Alec Marantz
PY - 2026
DA - 2026/09/16
PB - PsyArXiv
DO - 10.31234/osf.io/vu5k4_v4
UR - https://osf.io/preprints/psyarxiv/vu5k4_v4/
AB - Rule generalization is one of the most useful tools our mind uses to facilitate learning. In language, we apply it to connect what words mean to what they are made of, such as their suffixes. This is a morphosemantic relation, and it helps us process words we already know and easily acquire new words by composing their meaning. How is this relation acquired, and how does it manifest into generalization for understanding new words? In this study, we investigated the brain basis of these questions through an English-based artificial language study, where learners read sentences to learn a suffix-meaning relation of new words like “thinkle”. Words were either meaningful (places or things) or had no clear meaning. Each suffix family had majority and minority meaning words that participants had to keep track of for testing. For meaningful words in training, we found that higher performance scores predicted higher selective engagement in the left inferior frontal gyrus, specifically the pars triangularis and pars opercularis. In testing, both learned and generalized words, both increased lateral prefrontal cortex engagement. Meanwhile, words with no clear meaning had a much more diffuse brain engagement, recruiting bilateral temporoparietal regions. This resembles the processing of abstract words from previous literature. We found no difference between majority and minority meaning words, especially with the variance in participants’ performance strategies, but found differences in what predicted significant brain activity based on strategy difference. This study grounds itself in the reality of word and affixal semantics with previous research on language learning and universals. Ultimately, it lends a learning paradigm that resembles natural language learning across many languages to carefully investigate its brain basis, while addressing an understudied aspect of rule generalization: ambiguity.
ER -