Biology subjects

Tseng, C.

Publications and source records attributed to Tseng, C..

4 recordsLinked to original sources

bioRxiv2025

Optimizing Language Model Embeddings to Voxel Activity Improves Brain Activity Predictions

Recent studies have shown that contextual semantic embeddings from language models can accurately predict human brain activity during language processing. However, most studies use contextual embeddings with the same context length and model layer for all voxels, potentially overlooking meaningful variations across the brain. In this study, we investigate whether optimizing contextual embeddings for individual voxels improves their ability to predict brain activity during reading. We optimize embeddings for each voxel by selecting the best-predicting context length, model layer, or both. We perform this optimization with two different types of stimuli (isolated sentences and narratives), and quantify the performance gains of optimized embeddings over standard fixed embeddings. Our results show that voxel-specific optimization substantially improves the prediction accuracy of contextual semantic embeddings. These findings demonstrate that voxel-specific contextual tuning provides a more accurate and nuanced account of how the contextual semantic information is represented across the cortex.

neuroscience↗

bioRxiv2024

Genome Instability Precedes Viral Integration in Human Papilloma Virus Transformed Tonsillar Keratinocytes

Approximately 70% of oropharyngeal squamous carcinomas (OPSCC) are associated with human papillomavirus (HPV). While HPV-positive (HPV+) OPSCC responds better to standard therapies and patients with HPV+ tumor generally have better outcomes than those with HPV-negative (HPV-) tumors, a subset of HPV+ patients do have poor outcomes. Our previous work suggested that tumors with integrated virus exhibit significantly greater genome wide genomic instability than those with only episomal viral genomes and patients with HPV+ OPSCC with episomal viral genomes had better outcomes. To explore the causal relation between viral integration and genomic instability, we have examined the time course of viral integration and genetic instability in tonsillar keratinocytes transformed with HPV16. HPV-infected human tonsil keratinocyte cell lines were continuously passaged and every fifth passage some cells were retained for genomic analysis. Whole genome sequencing and optical genomic mapping confirmed that virus integrated in five of six cell lines while remaining episomal in the sixth. In all lines genome instability occurred during early passages, but essentially ceased following viral integration but continued to occur later passages in the episomal line. To test tumorigenicity of the cell lines, cells were injected subcutaneously into the flanks of nude mice. A cell line with the integrated virus induced tumors following injection in the nude mouse while that with the episomal virus did not. We conclude that genomic instability is not the result of viral integration but likely promotes integration. Moreover, those transformants with episomal virus appear to be less tumorigenic than those with integrated virus.

cancer biology↗

bioRxiv2024

Bilingual language processing relies on shared semantic representations that are modulated by each language

Billions of people throughout the world are bilingual, and they can extract meaning from multiple languages. While some evidence suggests that there is a shared system in the human brain for processing semantic information from native and non-native languages, other evidence suggests that semantic processing is language-specific. We conducted a study to determine how semantic information for different languages is represented in the brains of bilinguals. Functional magnetic resonance imaging (fMRI) was used to record brain responses while participants read several hours of natural narratives in their native (Chinese) and non-native (English) languages. These data were then used to compare semantic representations between the two languages. We find that semantic representations are largely shared between languages, but that there are fine-grained differences in the representation of some semantic categories across languages. These results reconcile current competing theories of bilingual language processing. Significance StatementBilinguals understand the meaning of words in multiple languages. Whether this capacity reflects a shared brain system for processing both native and non-native languages, or whether processing is language-specific is still unclear. Here, we examine whether and how semantic representations in the brain support shared and/or language-specific processing. We recorded brain activity from participants reading narratives in their native (Chinese) and non-native (English) languages, and modeled how their brains represent word meaning in each language. We show that semantic representations are similar between the two language conditions, and that these representations are systematically modulated between native and non-native language comprehension.

neuroscience↗

bioRxiv2021

Semantic representations during language comprehension are affected by context

The meaning of words in natural language depends crucially on context. However, most neuroimaging studies of word meaning use isolated words and isolated sentences with little context. Because the brain may process natural language differently from how it processes simplified stimuli, there is a pressing need to determine whether prior results on word meaning generalize to natural language. fMRI was used to record human brain activity while four subjects (two female) read words in four conditions that vary in context: narratives, isolated sentences, blocks of semantically similar words, and isolated words. We then compared the signal-to-noise ratio (SNR) of evoked brain responses, and we used a voxelwise encoding modeling approach to compare the representation of semantic information across the four conditions. We find four consistent effects of varying context. First, stimuli with more context evoke brain responses with higher SNR across bilateral visual, temporal, parietal, and prefrontal cortices compared to stimuli with little context. Second, increasing context increases the representation of semantic information across bilateral temporal, parietal, and prefrontal cortices at the group level. In individual subjects, only natural language stimuli consistently evoke widespread representation of semantic information. Third, context affects voxel semantic tuning. Finally, models estimated using stimuli with little context do not generalize well to natural language. These results show that context has large effects on the quality of neuroimaging data and on the representation of meaning in the brain. Thus, neuroimaging studies that use stimuli with little context may not generalize well to the natural regime. Significance StatementContext is an important part of understanding the meaning of natural language, but most neuroimaging studies of meaning use isolated words and isolated sentences with little context. Here we examined whether the results of neuroimaging studies that use out-of-context stimuli generalize to natural language. We find that increasing context improves the quality of neuroimaging data and changes where and how semantic information is represented in the brain. These results suggest that findings from studies using out-of-context stimuli may not generalize to natural language used in daily life.

neuroscience↗