Our lab has also tried replicating this experiment on a few non-naive subjects as a pilot. Anecdotally, I'm aware of at least 2 other groups that have as well, and I believe they aren't the groups cited in the paper (#14,15). We all failed.

Thanks to NIH NIDCD for supporting this line of work and to @urneuroscience.bsky.social for additional and ongoing support.

We discuss several ways in which our position can be tested. We plan to engage vigorously in such testing and welcome parallel efforts from other groups. Also more than happy to take any feedback on this preprint and on our position more generally.

Ultimately, this leads us to conclude that entrained oscillations in specific frequency bands make very limited contributions to the cortical tracking of natural speech and that the contribution of evoked responses to that tracking cannot reasonably be ignored.

6) Natural speech is not a very rhythmic stimulus, suggesting that the contribution to cortical speech tracking of oscillatory attention mechanisms may even be quite limited.

5) There is good evidence for entrained oscillations as a mechanism of attention in the context of rhythmic, predictable stimuli. The extrapolation of this finding to argue for oscillations as a primary contributor to sensory-driven responses to natural speech seems unwarranted.

4) Several features of TRFs suggest they are not explainable as deriving from the entrainment of endogenous oscillations.

3) TRFs share a lot of commonalities with classic ERPs. If the early sensory components of ERPs derive from sensory-driven evoked activity, then TRFs – and the cortical tracking that TRFs model – also likely derive from evoked activity.

2) Temporal Response Functions exist and are consistent with a model based on evoked responses, although not conclusive proof of such a model.

Specifically, we argue that:
1) The evidence for entrained endogenous oscillations as a key mechanism underlying neural measures of natural speech processing is weak.

In this opinion piece, we take a strong position on this debate – by making some falsifiable claims and questioning some existing assumptions and interpretations – as a step towards increased clarity on the topic of cortical speech tracking.

Research on cortical speech tracking can largely be divided into two separate literatures: one that interprets cortical speech tracking through the lens of entrained oscillations, and one that (at least implicitly) frames such tracking as reflecting evoked responses.

We also invite graduate & postdoctoral researchers to submit tutorial proposals as part of the workshop. Successful applicants will also have the opportunity to showcase their related work in a dedicated 20 min presentation.
Deadline: 20 June 2025
Application form: docs.google.com/forms/d/e/1F...