Temporal constraints on the neural signatures of narrative processing

Reviewer #1 (Public review):

Summary:

This paper reports an important MEG study that is interesting from many different angles. By manipulating presentation rate (Fast vs. Slow) and "Structure" (word list vs. sentence list vs. story), the authors revealed the spatiotemporal dynamics of processing constituents at different speeds and semantic-conceptual scales. The methods are solid, and the results would be interesting to both researchers interested in the neural basis of structure building and researchers interested in the consequences of presentation rate, which, as the authors note, is understudied for visual sentence presentation. Overall, I enjoyed reading this paper and agree that it reports valuable findings for language processing, but there are a few points where clarity can be improved for readers to better evaluate and appreciate this work.

Strengths:

This paper studies the effects of presentation rate and linguistic structure building at different scales with MEG, which is perhaps one of the first MEG studies approaching these questions, and MEG is an appropriate technology to investigate spatiotemporal dynamics in the brain. The authors conducted a breadth of analysis, enabling us to fully understand what is shown by their data.

Weaknesses:

Here I would like to point out a few points where clarity can be improved (e.g., analytical details) for readers to better appreciate this paper.

(1) The "mean proficiency" was reported on page 5, but I did not see the details of the proficiency task, which also need to be reported.

(2) I was confused about the baseline correction part of the MEG analysis on page 10, and I would appreciate it if the authors laid out the rationale more clearly. From what I understand, the authors did not perform baseline correction, which is completely understandable, as word-lists, sentence-lists, and stories would have different baselines, and the baselines also keep building up as the trial evolves (which is a part of the study, instead of something to be parsed out). In this context, it becomes confusing why the authors singled out the Fast presentation condition to be unsuitable for baseline correction ("Since the prestimulus period cannot be analyzed during the Fast presentation..."). If the rationale is that the Fast condition did not have a long-enough blank screen between segments, this would not be a problem, as baselining is done in ERP RSVP studies anyway all the time. I am also curious what is meant by "The Fast and Slow presentation conditions were separated and demeaned independently". Would this make comparing the Fast and Slow conditions trickier? Overall, I think the analytical choices are potentially defensible, but more explanations of the rationale are needed.

(3) For the Results section, the authors should check the text again to increase clarity for statistical reporting for readers. For example, on page 14, the p-values were not always reported, and the dfs for t-values were also not always reported. For "pairwise comparisons", one would expect a reporting like "ps < ..." instead of a singular "p < 0.001".

(4) For behavioral results, was the analysis of RT directed at all trials or correct trials only? It would be worth checking whether the RT results hold true when analyzing only correct trials, or whether they were primarily driven by incorrect trials. Note that I don't find it a problem to include all trials for the MEG analyses (which the authors did), as we would believe that the participants were doing the task anyway despite it being challenging, and task difficulty was an inherent component of the research goal instead of something to be parsed out.

Reviewer #2 (Public review):

Summary:

The main contribution of this study is that brain activations related to linguistic context varied as a result of presentation speed, with the main finding that increased activity for coherent stories relative to other conditions was reduced in fast presentation relative to slow. The results thus challenge certain assumptions about the nature of the brain dynamics of language processing, with certain effects even disappearing under faster presentations, which may be related to the processing mode of the participant. The results continue to establish the viability of a parallel presentation design, which generally produces results congruent with those of the literature.

Strengths:

The study contains a somewhat novel presentation method, illustrating its viability. The results are bolstered by a strong sample size (N=33) and robust analytic techniques. The conclusions are measured and appropriate to the results, and the manuscript is exceedingly clearly written and accessible to readers.

Weaknesses:

The spatial specificity of the effects is hampered by the use of MEG, particularly with minimal structural MRIs for participants. Thus, the conclusions of the study in the spatial domain are tentative and more general than might result from other studies.

In addition, the general finding that faster presentation speed reduced activity overall (and eliminated it in the frontal cortex) appears to be somewhat contradictory to existing literature, which finds that sentences which are complex or difficult to process generally produce greater activation, particularly in the frontal cortex. These studies might be reviewed, and this (seeming) contradiction could be addressed.