Language comprehension functionally modulates first-order relay thalamic nuclei
Curation statements for this article:-
Curated by eLife
eLife Assessment
This study presents an important exploration of the role of thalamic nuclei in language processing. The results will be of interest to researchers interested in the neurobiology of language. However, the evidence in support of some primary claims is incomplete.
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
Language, a uniquely human higher-order cognitive function, has traditionally been attributed to cortical mechanisms with limited attention given to subcortical contributions. Recent advances in non-invasive neuroimaging have revealed that thalamic activity can be modulated by attention and task demands. Moreover, lesion studies have hinted at the thalamus’s potential role in language processing. Nevertheless, the precise involvement of this structure in language remains unclear. Here, we ask whether language- related modulations can be detected as early as the sensory thalamic stage. Using functional MRI to image 40 human participants (both female and male) while processing linguistic and non-linguistic stimuli of three main language systems (reading, speech comprehension, and speech production), we demonstrate specific activation of first-order nuclei during targeted language system tasks: lateral geniculate (LGN) for reading, medial geniculate (MGN) for speech comprehension and ventrolateral (VLN) for speech production. Notably, we show linguistic versus non-linguistic stimuli exhibit functional modulations during comprehension tasks (reading and speech) of left MGN and, to a lesser extent, left LGN—in line with prior studies of lateralization of language processes. Multi-voxel classification analysis confirmed left-lateralized linguistic modulation in the MGN, but not in the LGN. Given the complexity of thalamic connectivity and its potential role in integrating sensory and cognitive processes, this work offers a comprehensive characterization of first-order thalamic engagement across reading, speech comprehension, and speech production within the same participants, advancing our understanding of thalamic involvement and thalamocortical interactions in language function.
Article activity feed
-
eLife Assessment
This study presents an important exploration of the role of thalamic nuclei in language processing. The results will be of interest to researchers interested in the neurobiology of language. However, the evidence in support of some primary claims is incomplete.
-
Reviewer #1 (Public review):
The manuscript by Mengxing et al., reports an assessment of three thalamic nuclei (auditory, visual, somatosensory) in a 3 x 2 factorial design to test for specificity of responses in thalamic nuclei (as defined by an anatomical atlas, Iglesisas et al. 2018), to linguistic processing particularly in the left hemisphere. The conditions are reading, speech production, and speech comprehension and their respective control conditions. The control condition consists of resynthesised material of the three conditions and therefore differs in task as well as stimulus content. The authors report the following results:
(1) BOLD-response analyses: left MGB linguistic vs non-linguistic significant; left LGN linguistic vs non-linguistic significant. There is a hemisphere x stimulus Interaction for the MGB.
(2) MVPA: left …
Reviewer #1 (Public review):
The manuscript by Mengxing et al., reports an assessment of three thalamic nuclei (auditory, visual, somatosensory) in a 3 x 2 factorial design to test for specificity of responses in thalamic nuclei (as defined by an anatomical atlas, Iglesisas et al. 2018), to linguistic processing particularly in the left hemisphere. The conditions are reading, speech production, and speech comprehension and their respective control conditions. The control condition consists of resynthesised material of the three conditions and therefore differs in task as well as stimulus content. The authors report the following results:
(1) BOLD-response analyses: left MGB linguistic vs non-linguistic significant; left LGN linguistic vs non-linguistic significant. There is a hemisphere x stimulus Interaction for the MGB.
(2) MVPA: left MGB linguistic vs. non-linguistic significant; bilateral VLN linguistic vs. non-linguistic significant; significant lateralisation in MGB (left MGB responses better classified linguistic vs. non-linguistic in contrast to right).
(3) Functional connectivity: there is, in general, connectivity between the thalamic ROIs and the respective primary cortices independent of linguistics.
Strengths:
The study has a clear and comprehensive design and addresses a timely topic. Thalamic nuclei and their interaction with the cerebral cortex regions are likely key to understanding how perception works in a world where one has to compute highly dynamic stimuli often in an instant. Speech and written text are prime examples of ecologically important, extremely dynamic, and complex stimuli. The field of the contribution of cerebral cortex-thalamic loops in speech and language processing is wide open, and the study presents a solid approach to address their role in different language modalities (i.e., reading, comprehension, production). Particularly the left MGB specialisation for linguistic content seems to me a solid finding of the study. This is revealed both in the BOLD-response analysis as well as the MVPA analyses and is unlikely to be driven by a potential confound of differences in the stimulus material as this is accounted for by the interaction (linguistic vs. non-linguistic in left MGB vs. right MGB).
Weaknesses:
(1) The spatial resolution of the fMRI acquisition of 2.4 mm3 pre-empts the claim of modulation of first-order thalamic nuclei for the auditory modality. The MGB is composed of multiple subnuclei that are not all first-order. Resolving the subnuclei would require different procedures and likely higher spatial resolution to avoid partial volume effects (Moerel et al., 2015; Mihai et al., 2019).
(2) The task conditions differed in both stimulus content as well as tasks. The linguistic tasks included single-word reading (visual), speech single-word comprehension (auditory), and single-word speech production (motor); while non-linguistic tasks involved viewing scrambled images (visual), listening to noise sequences (auditory), and producing unintelligible sounds (motor). It therefore remains unclear whether the difference in linguistic conditions are due to task-effects or due to stimulus differences.
Appraisal:
The open question the authors want to address is whether first-order thalamic nuclei "show functional modulations dependent upon whether the stimuli being processed are linguistic or non-linguistic". To address the question the authors want to examine the "engagement of first-order relay thalamic nuclei specifically in [...] three language systems: LGN in reading, MGN in SC [speech comprehension], and VLN in SP [speech production]." In my opinion the study does not yet fully achieve this aim. First, it is unclear whether it is really the linguistic components that modulate the thalamic nuclei, or other differences between the active tasks and the passive listening/viewing or production of unintelligible sounds or differences in stimulus composition that could be equally revealed by non-linguistic meaningful stimuli against resynthesised versions of these. As said above, this criticism does not apply to the interaction found in the auditory modality that reveals stronger modulation for the linguistic condition in contrast to the non-linguistic condition in the left MGB in contrast to the right MGB. Second, to delineate the first-order thalamic nuclei would in my view need other procedures and higher spatial resolution.
References
Iglesias, J. E., Insausti, R., Lerma-Usabiaga, G., Bocchetta, M., Van Leemput, K., Greve, D. N., van der Kouwe, A.,Fischl, B., Caballero-Gaudes, C., & Paz-Alonso, P. M. (2018). A probabilistic atlas of the human thalamic nuclei combining ex vivo MRI and histology. NeuroImage, 183(July), 314-326.641 https://doi.org/10.1016/j.neuroimage.2018.08.012
Mihai, P. G., Moerel, M., de Martino, F., Trampel, R., Kiebel, S., & von Kriegstein, K. (2019). Modulation of tonotopic ventral medial geniculate body is behaviorally relevant for speech recognition. eLife, 8, e44837. https://doi.org/10.7554/eLife.44837
Moerel, M., De Martino, F., Uğurbil, K., Yacoub, E., & Formisano, E. (2015). Processing of frequency and location in human subcortical auditory structures. Scientific reports, 5, 17048. https://doi.org/10.1038/srep17048
-
Author response:
The following is the authors’ response to the original reviews.
Public Reviews:
Reviewer #1 (Public review):
Summary:
The manuscript by Mengxing et al., reports an assessment of three first-order thalamic nuclei (auditory, visual, somatosensory) in a 3 x 2 factorial design to test for specificity of responses in first-order thalamic nuclei to linguistic processing particularly in the left hemisphere. The conditions are reading, speech production, and speech comprehension and their respective control conditions.
The authors report the following results:
(1) BOLD-response analyses: left MGB linguistic vs non-linguistic significant; left LGN linguistic vs non-linguistic significant. There is no hemisphere x stimulus interaction.
(2) MVPA: left MGB linguistic vs. non-linguistic significant; bilateral VLN …
Author response:
The following is the authors’ response to the original reviews.
Public Reviews:
Reviewer #1 (Public review):
Summary:
The manuscript by Mengxing et al., reports an assessment of three first-order thalamic nuclei (auditory, visual, somatosensory) in a 3 x 2 factorial design to test for specificity of responses in first-order thalamic nuclei to linguistic processing particularly in the left hemisphere. The conditions are reading, speech production, and speech comprehension and their respective control conditions.
The authors report the following results:
(1) BOLD-response analyses: left MGB linguistic vs non-linguistic significant; left LGN linguistic vs non-linguistic significant. There is no hemisphere x stimulus interaction.
(2) MVPA: left MGB linguistic vs. non-linguistic significant; bilateral VLN linguistic vs. non-linguistic significant; significant lateralisation in MGB (left MGB responses better classified linguistic vs. non-linguistic in contrast to right).
(3) Functional connectivity: there is, in general, connectivity between the thalamic ROIs and the respective primary cortices independent of linguistics.
Strengths:
The study has a clear and comprehensive design and addresses a timely topic. First-order thalamic nuclei and their interaction with the respective cerebral cortex area are likely key to understanding how perception works in a world where one has to compute highly dynamic stimuli often in an instant. Speech is a prime example of an ecologically important, extremely dynamic, and complex stimulus. The field of the contribution of cerebral cortex-thalamic loops is wide open, and the study presents a solid approach to address their role in different speech modalities (i.e., reading, comprehension, production).
Weaknesses:
I see two major overall weaknesses in the manuscript in its current form:
(1) Statistics:
Unfortunately, I have doubts about the solidity of the statistics. In the analyses of the BOLD responses, the authors do not find significant hemisphere x stimulus interactions. In my view, such results would pre-empt doing a post-hoc t-test. Nevertheless, the authors motivate their post-hoc t-test by 'trends' in the interaction and prior hypotheses. I see two difficulties with that. First, the origin of the prior hypotheses is somewhat unclear (see also the comment below on hypotheses), and the post-hoc t-test is not corrected for multiple comparisons. I find that it is a pity that the authors did not derive more specific hypotheses grounded in the literature to guide the statistical testing, as I think these would have been available, and the response properties of the MGB and LGN also make sense in light of them. In addition, I was wondering whether the MVPA results would also need to be corrected for the three tests, i.e., the three ROIs.
We thank the reviewer for prompting us to revisit this, as it led us to identify and correct a specification error in the original ROI ANOVAs. The original ROI ANOVAs had been fitted without subjects as a within-subjects factor, treating each subject's observations as independent. For a fully within-subjects design, this is a genuine error: it pools within- and between-subject variance into a single residual and uses an incorrect error term and degrees of freedom. We have re-run all ROI analyses as repeated-measures ANOVAs with subjects as a within-subjects factor, on the same data and contrasts.
For the modality main effects, all three nuclei retain a significant effect of Modality under the corrected model (LGN F(2,76) = 30.02, p < 0.001, η2_p = 0.441; MGN F(2,76) = 4.56, p = 0.014, η2_p = 0.107; VLN F(2,76) = 42.28, p < 0.001, η2_p = 0.527). The corrected F-values are lower than originally reported because the original analyses used an inflated error df (465, from treating observations as independent) rather than the correct within-subjects df (76); the tests are therefore more conservative. We also replaced the original Tukey HSD post-hoc tests, which assume independent groups, with paired comparisons (Holm-corrected within each nucleus). The qualitative pattern is unchanged.
For the Stimuli × Hemisphere analysis, the corrected model yields a significant MGN interaction (F (1,38) = 8.63, p = 0.006, η2_p = 0.185). The LGN (F (1,38) = 1.02, p = 0.318, η2_p = 0.026) and VLN (F (1,38) = 0.52, p = 0.474, η2_p = 0.014) interactions remain non-significant, consistent with the original report. We also now report the Stimuli main effects, as the reviewer requested: the LGN shows a significant overall linguistic effect across hemispheres (F(1,38) = 4.44, p = 0.042), consistent with the reviewer's observation that a main effect of stimuli appeared present in the LGN.
The significant MGN interaction directly addresses the reviewer's concern: the left-MGN comparison is the appropriate decomposition of a reliable interaction, not a test performed after a non-significant omnibus result. This prediction was also motivated a priori by prior work on the left auditory thalamus in speech recognition and developmental dyslexia, which we set out in detail in our response to the Hypotheses comment below, as well as in the revised manuscript version (see pp. 5).
For the LGN, the interaction is not significant, and we now present the left-LGN comparison (t (38) = 2.43, p = 0.020, d_z = 0.39) as a directional, theoretically motivated test rather than as a follow-up to a significant interaction. As detailed in our response to the Hypotheses comment, the prediction has two components: that the LGN is modulated by linguistic visual content at all (well supported), and that this modulation is left-lateralised (motivated more indirectly, and which we therefore treat as the more exploratory element). We interpret the LGN left-lateralisation with corresponding caution.
The hemisphere comparisons are corrected for multiple comparisons using the Holm–Bonferroni procedure across the two tests; both remain significant after correction (left MGN p_Holm = 0.009; left LGN p_Holm = 0.020). The corrected statistics for all nuclei are reported in the revised Methods and Results sections of the manuscript (see pp. 10, 14, 15).
Regarding the analysis, they showed whether the individual classification results are different from chance using a permutation test. Therefore, these are significant tests, but they do not compare results between conditions or different distributions. Here, we just built a null distribution to estimate the performance of the classifier for each ROI. Thus, multiple comparison corrections do not apply to these analyses.
Hypotheses:
In my view, it is relatively unclear where the hypotheses precisely come from. For example, the paragraph on the hypotheses in the introduction (p. 6-7) is devoid of references. I also have the impression that the hypotheses are partly not taking into account previous reports on first-order thalamic nuclei involvement in linguistic vs. non-linguistic processing. For example, the authors test for lateralisation of linguistic vs. non-linguistic responses in all nuclei. However, from previous literature, one could derive the hypothesis that the lateralisation in MGB for speech might be there - previous work shows, for example, that speech recognition abilities consistently correlate with left MGB only (von Kriegstein et al., 2008 Curr Biol; Mihai et al., 2019 eLife). In addition, the involvement of the MGB in speech-in-noise processing is present in the left MGB (Mihai et al., 2021, J Neuroscience). Developmental dyslexia, which is supposed to be based on imprecise phonological processing (Ramus et al., 2004 TiCS), has alterations in left MGB (Diaz et al., 2012 PNAS; Galaburda et al., 1994 PNAS) and left MGB connections to planum temporale (Tschentscher et al., 2019 J Neurosci) as well as altered lateralisation (Müller-Axt et al., 2025 Brain). Conversely, in the LGN, I'm not aware of any studies showing lateralisation for speech. See, for example, Diaz et al., 2018, Neuroimage, where there are correlations of LGN task-dependent modulation with visual speech recognition behaviour in both LGNs. Thus, based on this literature, one could have predicted the result pattern displayed, for example, in Figure 3A at least for MGB and LGN.
In summary, the motivation for the different hypotheses needs to be carved out more and couched into previous literature that is directly relevant to the topic. The above paragraph is, of course, my view on the topic, but currently, the paper lacks different literature as references to fully understand where the hypotheses are derived from.
We appreciate this point and agree that the hypothesis section in the original manuscript needed to be more specifically grounded. We have now revised the Introduction and have better grounded the hypotheses, organising the supporting evidence into two converging lines (pp. 3-5).
The first concerns the left MGN. In neurotypical listeners, task-dependent modulation of the auditory thalamus during speech recognition is bilateral, but its amplitude in the left MGB specifically tracks speech-recognition ability (von Kriegstein et al., 2008; Mihai et al., 2019), and this modulation localises to the left ventral MGB under more demanding speech-in-noise conditions (Mihai et al., 2021). This evidence is complemented by left-lateralised alterations of the auditory thalamus in developmental dyslexia, including reduced left MGB responses during phoneme processing that correlate with diagnostic scores (Diaz et al., 2012) and reduced left MGB-planum temporale structural connectivity (Tschentscher et al., 2019). On this basis, our prediction of left-MGN involvement was directly motivated by prior work, and the pattern shown in Figure 3A, as the reviewer notes, broadly aligns with what this literature would lead one to expect. As discussed in our previous response to the Statistics comment, the corrected Stimuli × Hemisphere analysis confirms this prediction, revealing a significant MGN interaction and a significant simple effect in the left MGN.
The second issue concerns the LGN, for which we distinguish between two predictions supported by different levels of empirical evidence. First, the prediction that the LGN is modulated by the linguistic content of visual stimuli is well motivated. Rather than acting as passive relays, first-order thalamic nuclei are subjected to top-down corticothalamic modulation (Mumford, 1991; Sherman & Guillery, 1996; Sillito et al., 1994), and Díaz et al.’s (2018) study showed that LGN responses are modulated during visual speech recognition. Consistent with this evidence, we observed a significant overall (bilateral) effect of linguistic content in the LGN. By contrast, the more specific prediction that this modulation would be left-lateralised is based on more indirect evidence. It follows from the strong left-lateralisation of reading and language processing at the cortical level, which may extend upstream to the thalamic visual relay, together with findings from the dyslexia literature. Morphometric studies have reported reduced left LGN volume in developmental dyslexia (Giraldo-Chica et al., 2015), although this finding has not been consistently replicated (Müller-Axt et al., 2017). More robust evidence comes from studies reporting left-lateralised alterations in LGN connectivity and function, including reduced left LGN–V5/MT connectivity correlated with rapid naming ability (Müller-Axt et al., 2017) and altered leftward functional lateralisation of the magnocellular LGN (Müller-Axt et al., 2025). However, existing functional studies report bilateral rather than left-lateralised LGN involvement during language processing (Díaz et al., 2018), and Müller-Axt et al. (2025) also note that the hemispheric origin of the magnocellular LGN lateralisation cannot be fully resolved. We therefore present the prediction of left-lateralised LGN modulation as theoretically motivated but exploratory, whereas the prediction of LGN involvement per se rests on considerably stronger empirical support. The revised Introduction now makes this distinction explicit and cites the relevant literature. We consider that the predictability of the lateralisation pattern from prior work, which the reviewer rightly highlights, is not in tension with the contribution of the present study. Our aim was not to establish that first-order thalamic nuclei participate in language processing per se, but rather to provide the first comprehensive examination of all three first-order sensory relay nuclei across reading, speech comprehension, and speech production within the same group of participants.
Reviewer #2 (Public review):
Summary:
This study investigates the involvement of first-order thalamic nuclei in language-related tasks using task-based fMRI in a 3 × 2 design contrasting linguistic and non-linguistic versions of reading, speech comprehension, and speech production. By focusing on the LGN, MGN, and VLN and combining activation, connectivity, lateralization, and multivariate pattern analyses, the authors aim to characterize modality-specific and language-related thalamic contributions.
Strength:
A major strength of the work is its hypothesis-driven and multimodal analytical approach, and the modality-specific engagement of first-order thalamic nuclei is robust and consistent with known thalamocortical organization. This is a very sound study overall.
Weaknesses:
However, several conceptual issues complicate the interpretation of the results as evidence for linguistic modulation per se. A central concern relates to the operationalization of the linguistic versus non-linguistic contrast. In the present design, linguistic and non-linguistic stimuli differ along multiple dimensions beyond linguistic content. For example, written words and scrambled images differ in spatial frequency structure, edge composition, contrast regularities, and familiarity, while intelligible speech and acoustically scrambled sounds differ substantially in temporal and spectral statistics. This is particularly relevant given that first-order thalamic nuclei such as the LGN are known to be highly sensitive to low-level sensory properties. As a result, observed differences in thalamic responses may reflect sensitivity to stimulus properties rather than linguistic processing per se, and this limits the specificity of claims regarding linguistic modulation.
We thank the reviewer for this comment and agree that stimuli differ in different low-level aspects beyond the linguistic versus non-linguistic nature of the stimuli. Rather than comparing multiple different types of stimuli in each modality to isolate effects, for example, at the visual level of spatial frequency structure, edge composition, contrast regularities and familiarity, the main goal of the present study was to examine within the same sample differential involvement of first-order thalamic nuclei as a function of language modality and the linguistic versus non-linguistic nature of the stimuli. As the reviewer points out below, the contribution of the present work lies less in establishing thalamic involvement in language or a single language modality (reading, speech comprehension, speech production) per se. Here our main goal was to examine these three modalities within the same study and with the same group of participants. And, so, it would have been extremely hard to include all the many types of stimuli needed to contrast or isolate these low-level aspects in every single modality, as we have done in previous work, for example on reading (e.g., Lerma-Usabiaga, Carreiras & Paz-Alonso, 2018). It is also true that the non-linguistic stimuli we used have been widely used as standard control (perceptual, and not lexical) conditions in previous studies (see for instance Lerma-Usabiaga et al., 2018; Mano et al., 2015; Sánchez et al., 2023; Vinckier et al., 2007, for the reading modality) and that the left-lateralized linguistic versus non-linguistic modulation that we found is difficult to reconcile with a purely low-level account, where as far as we know effects would have been bilateral.
Based on this comment from the reviewer, we have now included as a limitation of the study in the revised version of the manuscript that the stimuli differ on other low-level sensory properties beyond their linguistic versus non-linguistic nature and that future studies may want to further investigate all these other aspects in each of the modalities examined in the present work (see p. 22).
Relatedly, although the manuscript frequently refers to effects "depending on the linguistic nature of the stimuli," the statistical evidence for linguistic versus nonlinguistic modulation is uneven across analyses. Whole-brain contrasts collapse across stimulus type and primarily test modality effects. Similarly, the primary ROI analyses of activation amplitude are collapsed across linguistic and nonlinguistic conditions and convincingly demonstrate modality-specific engagement of thalamic nuclei, but do not in themselves provide evidence for linguistic modulation. Linguistic effects emerge only in later, more targeted analyses focusing on hemispheric lateralization and multivariate pattern classification, and these effects are nucleus-, modality-, and analysis-specific rather than general. Taken together, these results suggest that linguistic modulation constitutes a secondary and selective finding, whereas modalityspecific task engagement represents the primary and most robust outcome of the study.
We thank the reviewer for this observation, which in our view reflects the intended logic of the study.
The whole-brain contrasts and the primary ROI activation analyses (across modalities) were not designed to demonstrate linguistic modulation. Their role was twofold: they confirm modality-specific organisation of the first-order relay nuclei (LGN for reading, MGN for speech comprehension, VLN for production), here shown for all three nuclei in the same participants. And they validate that our ROIs were correctly identified and engaged under the same task conditions as the subsequent analyses. As the reviewer notes, this modality-specific engagement is the most robust outcome of these analyses, which it was expected since it follows directly from known thalamocortical organisation, and it is the foundation on which the study's novel question is then tested.
The linguistic modulation is a more specific hypothesis, and it is tested in the analyses designed to address it: the Stimuli × Hemisphere ROI analyses and the multivariate classification. That the linguistic effects emerge in these targeted analyses rather than in the modality contrasts is therefore by design: the two sets of analyses ask different questions. We agree the linguistic effects are more selective than the modality effects. They are concentrated in the comprehension tasks, left lateralised, and strongest in the MGN. But this selectivity forms a coherent and interpretable pattern consistent with left-hemisphere language lateralisation, rather than a diffuse effect.
We have revised the manuscript, and we think that the main different findings of the study are adequately reflected in the abstract and discussion sections. We have now made subtle changes along the revised version of the manuscript to make even clearer that the modality analyses are framed as a combined replication and ROI-validation step, and the linguistic modulation is foregrounded as the study's primary contribution.
An additional interpretational issue concerns task engagement and attention. The tasks differ substantially in cognitive demands (e.g., passive reading and listening versus overt speech production), and linguistic and non-linguistic blocks may differ systematically in salience or engagement. This is particularly important given prior evidence, cited by the authors, that LGN and MGN activity can be modulated by task demands and attention. In the absence of behavioral measures indexing task engagement or compliance, it is difficult to determine whether differences between linguistic and non-linguistic conditions reflect linguistic processing per se or are mediated by attentional factors.
We thank the reviewer, and we acknowledge that we did not collect behavioural measures of task engagement, and that linguistic and non-linguistic stimuli may differ in attention or engagement as well as in linguistic content. However, attentional allocation is in part intrinsic to linguistic processing. Comprehending a meaningful word entails allocating attention in a way that processing meaningless input does not. Cognitive functions are not a unitarity phenomenon, and it is basically impossible to tease them apart into every single element in it. For instance, reading involves visual perception, orthographic processing, phonological processing, semantics, attention, and memory. The main goal of the present study is not to isolate every single possible element. At the level of our claim, that the thalamus is differentially engaged depending on the linguistic nature of the stimuli, these attentional differences are part of the phenomenon rather than a confound to be removed. Isolating more specific levels of linguistic processing, for instance, lexical processing, via word versus pseudoword contrasts, or orthographic processing, via consonant strings, would require designs targeting those finer distinctions, which we note in the Discussion section as a direction for future work; these address narrower questions than the linguistic-versus-non-linguistic distinction examined here.
A further consideration is that some of the effects we found are left-lateralised. A general difference in attention or engagement between conditions would not necessarily predict a specifically left-lateralised thalamic response. In our view, the asymmetry instead points to language-related lateralised processing rather than to generic differences in engagement. But of course, more studies examining more conditions and aspects in each modality are needed. The present study represents just one contribution towards that goal.
We added some lines discussing this point in the Discussion section of the revised manuscript (pp. 20-22).
Finally, while the manuscript emphasizes the novelty of evaluating thalamic involvement in language, thalamic contributions to language have been documented previously in both lesion and functional imaging studies. The contribution of the present work, therefore, lies less in establishing thalamic involvement in language per se, and more in its focus on specific first-order nuclei, its multimodal design, and its combination of univariate, connectivity, and multivariate analyses. Moderating claims of novelty would help place the findings more clearly within the existing literature.
We thank the reviewer for this comment and the time dedicated to this review. We have moderated our novelty claims. We removed statements asserting priority for thalamic involvement in language, including the claim that no prior study had reported linguistic modulation at the level of these nuclei (also raised by Reviewer #1), and we no longer frame the study as the "first" to examine these questions.
We now frame the contribution as a comprehensive examination of all three first-order relay nuclei (LGN, MGN, VLN) across reading, speech comprehension, and speech production in the same participants, combining univariate activation, functional and structural connectivity, and multivariate pattern analyses. Within this design, the differential engagement of these nuclei depending on the linguistic nature of the stimuli, left-lateralised, is the finding we foreground, situated within rather than asserted as priority over the existing literature on thalamic involvement in language.
We have revised our abstract and conclusion to reflect this change (p. 2, p. 23)
Recommendations for the authors:
Reviewer #1 (Recommendations for the authors):
There are a couple of things that could be done to strengthen the manuscript besides the ones that I've already outlined in the public review.
(1) ROIS
In our experience, the Iglesias et al. (2018) ROIs are unfortunately spatially imprecise, and it would be reassuring to test the results with other ROIs, e.g., Sitek ROIs for MGB and Müller-Axt ROIs for LGN.
We thank the reviewer for this suggestion. In the present study ROI analyses were conducted in each participant's native anatomical space. Adopting the Sitek or Müller-Axt ROIs would require warping externally defined templates into native space, which introduces interpolation and registration errors that can be particularly consequential for structures of this size. Working from native-space segmentations therefore avoids a source of spatial distortion.
We developed Iglesias et al’s (2018) atlas and its companion segmentation in our lab (Iglesias, … & Paz-Alonso, 2018), which has been reviewed, validated, and widely used in the neuroimaging literature. Although we can agree that for some nuclei, especially the smaller ones, there can be some spatial imprecisions (as in practically any available atlas of thalamic nuclei), given the type of analyses and resolution used in the present work, as well as its main goal (i.e., examining the involvement of three first-order nuclei in three main language modalities within the same study sample) we have retained the native-space from Iglesias et al.’s 2018 segmentation and note the choice in the Methods section.
Nevertheless, we agree with the reviewer that, for investigations specifically using higher functional resolution and targeting further fine-grained subregions within these ROIs (e.g., vMGB, dMGB, and mMGB; VLa and VLp), a different segmentation strategy and, where feasible, ultrahigh-field (7 T) MRI would be ideal for more spatially precise delineations.
We provided details in the ROI definition section to show the volumes of our ROIs and the comparison with relevant reports from other labs (p. 10).
(2) Introduction
It has now been decades since the 'relay nuclei' are not only relay (e.g., Sillito 1994; Sherman&Guillery 1996). So, as someone in the field, an introduction based on 'first-order thalamic nuclei are not a relay' seems a bit old-fashioned. Conversely, it would be great to review in depth the relevant literature on the topic. In my view, this would also streamline the introduction. In that context, I would appreciate a more thorough treatment of previous literature. For example, it is wrong that "Similar attention-related effects have also been observed in the MGN (e.g., Mihai et al., 2021; von Kriegstein et al., 2008)". These are not attention-related effects; in fact, the two tasks are in many studies specifically designed to have the same task difficulty to avoid attention-related effects. In the end, these tasks tap into the difference between linguistic speech recognition tasks vs. control tasks (paralinguistic, such as voice recognition or completely different, such as loudness recognition). At another point, the same studies are cited as studying speech comprehension, which, in my view, is also wrong, as I don't think that one can label one-back syllable recognition tasks as 'speech comprehension'.
We have revised the Introduction to situate the work within the contemporary corticothalamic framework from the outset, drawing on the relevant literature (e.g. Sillito et al., 1994; Sherman & Guillery, 1996), rather than framing it against a passive-relay model.
We thank the reviewer for pointing out the “attention-related” clarification regarding Mihai et al.’s (2021) and von Kriegstein et al.’s (2008) studies. We have revised the relevant sentence to describe these studies as demonstrating task-dependent modulation during speech recognition relative to non-speech control tasks, and removed the earlier framing that grouped them with attention effects. Also, we have revised the Introduction to distinguish our passive listening paradigm from the recognition-based paradigms (e.g., one-back syllable tasks) used in prior work, reserving "speech comprehension" for the former (see pp. 3-4).
(3) MGB or MGN
The MGB is called MGB because it contains not only first-order, but also second-order structures. With the resolution of 2.4 mm isotropic, it is impossible to dissociate the first-order MGB in my view.
We thank the reviewer for this point and agree that at 2.4 mm isotropic resolution the ROI cannot be restricted to the first-order ventral subdivision alone. We have addressed this directly in the Methods section (see pp. 9-10) where the ROI is defined, noting that the segmentation derives from Iglesias et al. (2018) and that, at our functional resolution, contributions from higher-order subdivisions cannot be excluded; our findings therefore pertain to the MGN as a whole. On terminology, the manuscript uses MGN throughout, consistent with the atlas and segmentation labelling.
(4) Reporting details
(a) I was missing important information for the results, such as tables for coordinates (ROIs and results), volumes for the ROIs, and comparison of ROI volumes to volumes previously reported in the literature.
As our ROIs were defined in each participant's native anatomical space rather than in a common template, MNI coordinates are not applicable. We have, however, added a sentence reporting the mean ± SD ROI volume (mm³) for each nucleus across participants, derived from the individual FreeSurfer segmentations, which characterises the size and consistency of the ROIs used. We included the ROI volumes and comparison with previous reports on p. 10.
(b) Another point of the statistics is that the results should be reported more comprehensively: For example, it looks a bit as if there were a main effect of linguistics in the LGN, but main effects were not reported (why?) as far as I can see.
We have added the Stimuli main effect and the Stimuli × Hemisphere interaction from the corrected ANOVA for all three nuclei to the revised Results. In brief: the LGN shows a significant overall linguistic advantage (linguistic > non-linguistic) across both hemispheres, with no significant interaction; the MGN shows no significant main effect but a significant Stimuli × Hemisphere interaction, indicating a linguistic advantage that is strongly hemisphere-specific (present in the left, absent in the right) and directly validating the left MGN comparison. The full statistics are now reported in the revised version of the manuscript (see p. 15).
(c) Please report effect sizes throughout the study (e.g., page 7, bottom, please check the whole manuscript).
We have added partial η2 for all ANOVA effects and Cohen's d for all planned comparisons throughout the revised manuscript.
(d) In my view, it is impossible to understand how the control stimuli were created based on the current description. For example, what does 'temporally shifting the auditory signals' mean? How much? In what direction? It would be helpful to show example stimuli.
We agree the original description was insufficient. The non-linguistic auditory stimuli were created using a custom matlab script, which was uploaded along with the dataset to osf repository upon finalization of the manuscript. For each real word, the original waveform was temporally randomized by shuffling the data points of the audio vector in a pseudo-random permutation (randperm). This process destroyed the phonological and semantic structure of the speech while perfectly preserving the duration, sampling rate, and total acoustic energy of the original stimuli. We have clarified this procedure in the Methods (p. 6).
(e) How far was it assessed that the reading task in the SP was not confounded by eye movements?
We thank the reviewer. Two points address this concern. First, within the reading modality, both conditions required the same directed fixation on the screen (words in the linguistic condition, scrambled images in the non-linguistic condition), so a systematic eye-movement difference between conditions is not expected. Second, and more directly, the speech comprehension and speech production tasks were performed with eyes closed, yet the left MGN showed a reliable linguistic effect; an eye-movement account cannot explain a linguistic modulation arising under eyes-closed conditions. We have clarified this in the Methods (pp. 7).
(f) Motivation for choosing the particular functional connectivity method unclear. Why not DCM? Why not PPI?
We thank the reviewer and have added a methodological justification of the functional connectivity method choice to the Methods section (p. 11). Our connectivity question concerned whether each first-order nucleus maintains coupling with its corresponding primary cortex during the relevant language task, that is, the magnitude of within-condition coupling, for which the beta-series correlation method (Rissman et al., 2004) provides a direct estimate of absolute connectivity within a given condition. PPI or gPPI, by contrast, estimates the change in coupling attributable to a psychological variable, i.e., condition-specific differences in connectivity rather than the absolute coupling within a condition; it was therefore less directly suited to our primary question. We additionally tested the linguistic-versus-non-linguistic difference using simple-effects comparisons on the subject-level z-transformed beta-series values, so our design does not preclude detecting condition differences (beta-series and PPI are well suited to comparable condition contrasts; Di et al., 2021). DCM would be the appropriate framework for questions of directionality and effective connectivity; however, we had no a priori directional model of thalamo-cortical interactions to motivate the strong model specification that DCM requires. We accordingly note in the Discussion section that the absence of linguistic modulation in beta-series coupling does not preclude its detection with effective-connectivity approaches such as DCM, which we flag as a direction for future work (p. 22).
(g) Green and red in figures can be difficult for the ca. 5% of the population that is red-green blind.
We have revised all relevant figures to use colourblind-accessible palettes.
(h) In multiple instances, references are missing. One point is the paragraph on the hypotheses I already mentioned, but also, for example, page 19 'language models'. Unclear which ones these are. There are multiple instances of this in the text, which I have not listed here for the sake of time, but proofreading is needed.
We have added the missing references, including in the hypothesis paragraph and for the "language models" reference, and have proofread the manuscript in full to address remaining instances (e.g. pp. 19, 21).
(i) Page 21 'yet no study has previously reported the modulation effect of linguistic processing at the level of the thalamic nuclei'. I don't think that this is correct. Linguistic modulation of first-order thalamic nuclei was first shown in two experiments in von Kriegstein et al., 2008, Curr Biol. I also don't understand why the authors emphasise so much 'the first' on this aspect. I think that the manuscript here does not need to be a 'first' on this aspect, as the design is fantastic and very comprehensive and could be an important novel contribution to the field in my view.
We thank the reviewer and have revised this accordingly. We have removed the sentence claiming that no prior study had reported linguistic modulation at the level of these nuclei. We retain the more circumscribed and accurate statement that the main goal of this study is to demonstrate simultaneous modulation of all three first-order relay nuclei within a single experimental paradigm spanning reading, speech comprehension, and speech production. We have shifted the framing away from priority claims toward the comprehensive cross-modality design as the contribution (pp. 2 and 23).
(j) "Study is in accordance with the Declaration of Helsinki". This statement is incorrect, given that the study has not been preregistered as far as I can see.
We thank the reviewer. We have revised the Ethics statement and dropped the statement about the Declaration of Helsinki (p. 5).
(j) References (that might be a bit harder to find)
Sherman SM, Guillery RW. Functional organization of thalamocortical relays. J Neurophysiol. 1996 Sep;76(3):1367-95. doi: 10.1152/jn.1996.76.3.1367. PMID: 8890259.
Sillito AM, Jones HE, Gerstein GL, West DC. Feature-linked synchronization of thalamic relay cell firing induced by feedback from the visual cortex. Nature. 1994 Jun 9;369(6480):479-82. doi: 10.1038/369479a0. PMID: 8202137.
Thank you!
-
-
-
-
-
eLife Assessment
This study presents a potentially valuable exploration of the role of thalamic nuclei in language processing. The results will be of interest to researchers interested in the neurobiology of language. However, the evidence is incomplete to support robust conclusions at this point.
-
Reviewer #1 (Public review):
Summary:
The manuscript by Mengxing et al., reports an assessment of three first-order thalamic nuclei (auditory, visual, somatosensory) in a 3 x 2 factorial design to test for specificity of responses in first-order thalamic nuclei to linguistic processing particularly in the left hemisphere. The conditions are reading, speech production, and speech comprehension and their respective control conditions. The authors report the following results:
(1) BOLD-response analyses: left MGB linguistic vs non-linguistic significant; left LGN linguistic vs non-linguistic significant. There is no hemisphere x stimulus interaction.
(2) MVPA: left MGB linguistic vs. non-linguistic significant; bilateral VLN linguistic vs. non-linguistic significant; significant lateralisation in MGB (left MGB responses better classified …
Reviewer #1 (Public review):
Summary:
The manuscript by Mengxing et al., reports an assessment of three first-order thalamic nuclei (auditory, visual, somatosensory) in a 3 x 2 factorial design to test for specificity of responses in first-order thalamic nuclei to linguistic processing particularly in the left hemisphere. The conditions are reading, speech production, and speech comprehension and their respective control conditions. The authors report the following results:
(1) BOLD-response analyses: left MGB linguistic vs non-linguistic significant; left LGN linguistic vs non-linguistic significant. There is no hemisphere x stimulus interaction.
(2) MVPA: left MGB linguistic vs. non-linguistic significant; bilateral VLN linguistic vs. non-linguistic significant; significant lateralisation in MGB (left MGB responses better classified linguistic vs. non-linguistic in contrast to right).
(3) Functional connectivity: there is, in general, connectivity between the thalamic ROIs and the respective primary cortices independent of linguistics.
Strengths:
The study has a clear and comprehensive design and addresses a timely topic. First-order thalamic nuclei and their interaction with the respective cerebral cortex area are likely key to understanding how perception works in a world where one has to compute highly dynamic stimuli often in an instant. Speech is a prime example of an ecologically important, extremely dynamic, and complex stimulus. The field of the contribution of cerebral cortex-thalamic loops is wide open, and the study presents a solid approach to address their role in different speech modalities (i.e., reading, comprehension, production).
Weaknesses:
I see two major overall weaknesses in the manuscript in its current form:
(1) Statistics:
Unfortunately, I have doubts about the solidity of the statistics. In the analyses of the BOLD responses, the authors do not find significant hemisphere x stimulus interactions. In my view, such results would pre-empt doing a post-hoc t-test. Nevertheless, the authors motivate their post-hoc t-test by 'trends' in the interaction and prior hypotheses. I see two difficulties with that. First, the origin of the prior hypotheses is somewhat unclear (see also the comment below on hypotheses), and the post-hoc t-test is not corrected for multiple comparisons. I find that it is a pity that the authors did not derive more specific hypotheses grounded in the literature to guide the statistical testing, as I think these would have been available, and the response properties of the MGB and LGN also make sense in light of them. In addition, I was wondering whether the MVPA results would also need to be corrected for the three tests, i.e., the three ROIs.
Hypotheses:
In my view, it is relatively unclear where the hypotheses precisely come from. For example, the paragraph on the hypotheses in the introduction (p. 6-7) is devoid of references. I also have the impression that the hypotheses are partly not taking into account previous reports on first-order thalamic nuclei involvement in linguistic vs. non-linguistic processing. For example, the authors test for lateralisation of linguistic vs. non-linguistic responses in all nuclei. However, from previous literature, one could derive the hypothesis that the lateralisation in MGB for speech might be there - previous work shows, for example, that speech recognition abilities consistently correlate with left MGB only (von Kriegstein et al., 2008 Curr Biol; Mihai et al., 2019 eLife). In addition, the involvement of the MGB in speech in noise processing is present in the left MGB (Mihai et al., 2021, J Neuroscience). Developmental dyslexia, which is supposed to be based on imprecise phonological processing (Ramus et al., 2004 TiCS), has alterations in left MGB (Diaz et al., 2012 PNAS; Galaburda et al., 1994 PNAS) and left MGB connections to planum temporale (Tschentscher et al., 2019 J Neurosci) as well as altered lateralisation (Müller-Axt et al., 2025 Brain). Conversely, in the LGN, I'm not aware of any studies showing lateralisation for speech. See, for example, Diaz et al., 2018, Neuroimage, where there are correlations of LGN task-dependent modulation with visual speech recognition behaviour in both LGNs. Thus, based on this literature, one could have predicted the result pattern displayed, for example, in Figure 3A at least for MGB and LGN.
In summary, the motivation for the different hypotheses needs to be carved out more and couched into previous literature that is directly relevant to the topic. The above paragraph is, of course, my view on the topic, but currently, the paper lacks different literature as references to fully understand where the hypotheses are derived from.
-
Reviewer #2 (Public review):
Summary:
This study investigates the involvement of first-order thalamic nuclei in language-related tasks using task-based fMRI in a 3 × 2 design contrasting linguistic and non-linguistic versions of reading, speech comprehension, and speech production. By focusing on the LGN, MGN, and VLN and combining activation, connectivity, lateralization, and multivariate pattern analyses, the authors aim to characterize modality-specific and language-related thalamic contributions.
Strength:
A major strength of the work is its hypothesis-driven and multimodal analytical approach, and the modality-specific engagement of first-order thalamic nuclei is robust and consistent with known thalamocortical organization. This is a very sound study overall.
Weaknesses:
However, several conceptual issues complicate the …
Reviewer #2 (Public review):
Summary:
This study investigates the involvement of first-order thalamic nuclei in language-related tasks using task-based fMRI in a 3 × 2 design contrasting linguistic and non-linguistic versions of reading, speech comprehension, and speech production. By focusing on the LGN, MGN, and VLN and combining activation, connectivity, lateralization, and multivariate pattern analyses, the authors aim to characterize modality-specific and language-related thalamic contributions.
Strength:
A major strength of the work is its hypothesis-driven and multimodal analytical approach, and the modality-specific engagement of first-order thalamic nuclei is robust and consistent with known thalamocortical organization. This is a very sound study overall.
Weaknesses:
However, several conceptual issues complicate the interpretation of the results as evidence for linguistic modulation per se. A central concern relates to the operationalization of the linguistic versus non-linguistic contrast. In the present design, linguistic and non-linguistic stimuli differ along multiple dimensions beyond linguistic content. For example, written words and scrambled images differ in spatial frequency structure, edge composition, contrast regularities, and familiarity, while intelligible speech and acoustically scrambled sounds differ substantially in temporal and spectral statistics. This is particularly relevant given that first-order thalamic nuclei such as the LGN are known to be highly sensitive to low-level sensory properties. As a result, observed differences in thalamic responses may reflect sensitivity to stimulus properties rather than linguistic processing per se, and this limits the specificity of claims regarding linguistic modulation.
Relatedly, although the manuscript frequently refers to effects "depending on the linguistic nature of the stimuli," the statistical evidence for linguistic versus non-linguistic modulation is uneven across analyses. Whole-brain contrasts collapse across stimulus type and primarily test modality effects. Similarly, the primary ROI analyses of activation amplitude are collapsed across linguistic and non-linguistic conditions and convincingly demonstrate modality-specific engagement of thalamic nuclei, but do not in themselves provide evidence for linguistic modulation. Linguistic effects emerge only in later, more targeted analyses focusing on hemispheric lateralization and multivariate pattern classification, and these effects are nucleus-, modality-, and analysis-specific rather than general. Taken together, these results suggest that linguistic modulation constitutes a secondary and selective finding, whereas modality-specific task engagement represents the primary and most robust outcome of the study.
An additional interpretational issue concerns task engagement and attention. The tasks differ substantially in cognitive demands (e.g., passive reading and listening versus overt speech production), and linguistic and non-linguistic blocks may differ systematically in salience or engagement. This is particularly important given prior evidence, cited by the authors, that LGN and MGN activity can be modulated by task demands and attention. In the absence of behavioral measures indexing task engagement or compliance, it is difficult to determine whether differences between linguistic and non-linguistic conditions reflect linguistic processing per se or are mediated by attentional factors.
Finally, while the manuscript emphasizes the novelty of evaluating thalamic involvement in language, thalamic contributions to language have been documented previously in both lesion and functional imaging studies. The contribution of the present work, therefore, lies less in establishing thalamic involvement in language per se, and more in its focus on specific first-order nuclei, its multimodal design, and its combination of univariate, connectivity, and multivariate analyses. Moderating claims of novelty would help place the findings more clearly within the existing literature.
-
Author response:
We acknowledge the concerns raised by both reviewers and plan to address them in our revision:
Regarding Reviewer #1's comments: We will strengthen the statistical framework and address the concerns about multiple comparison corrections. We will also expand our literature review to better motivate our hypotheses, particularly incorporating the work on lateralization patterns in MGN/LGN and the existing evidence on first-order thalamic nuclei in linguistic processing.
Regarding Reviewer #2's comments: We acknowledge the valid concern that linguistic and non-linguistic stimuli differ beyond linguistic content, including some low-level sensory properties. We will elaborate on the creation and properties of these stimuli in the Methods section and upload stimuli examples to an online repository to provide transparency about …
Author response:
We acknowledge the concerns raised by both reviewers and plan to address them in our revision:
Regarding Reviewer #1's comments: We will strengthen the statistical framework and address the concerns about multiple comparison corrections. We will also expand our literature review to better motivate our hypotheses, particularly incorporating the work on lateralization patterns in MGN/LGN and the existing evidence on first-order thalamic nuclei in linguistic processing.
Regarding Reviewer #2's comments: We acknowledge the valid concern that linguistic and non-linguistic stimuli differ beyond linguistic content, including some low-level sensory properties. We will elaborate on the creation and properties of these stimuli in the Methods section and upload stimuli examples to an online repository to provide transparency about differences. We will also add a discussion of this limitation in the Discussion section, acknowledging that disentangling effects of linguistic processing from low-level stimulus properties will require further testing in future research. Additionally, we will moderate part of our claims and reorganize the presentation of results as suggested, and clarify our contribution relative to existing literature.
-
Language comprehension functionally modulates first-order relay thalamic nuclei https://www.biorxiv.org/content/10.1101/2025.10.30.685533v1
-
Language comprehension functionally modulates first-order relay thalamic nuclei https://www.biorxiv.org/content/10.1101/2025.10.30.685533v1
-