Considerations for interpreting chronometric TMS studies investigating language production

Abstract

Based on a comprehensive meta-analysis, Indefrey & Levelt (2004) provided spatial and temporal correlates to the various stages of word production as described in the model by Levelt and colleagues (Levelt et al., 1999). Specifically, lemma retrieval occurs in the left mid portion of the middle temporal gyrus (MTG) from 200-275ms; phonological code retrieval in the left posterior middle/superior temporal gyri (STG) from 275-355ms; and syllabification in the left posterior inferior frontal gyrus (IFG) from 355-455ms. Over the years, chronometric TMS studies have been conducted in attempts to test these temporal and spatial estimates. The present study aimed to carry out a modified replication of one of these chronometric TMS studies, namely that of Schuhmann et al. (2012). Participants received triple-pulse TMS to one of three stimulation sites (posterior MTG, posterior STG & IFG) in one of five time-windows (225-275; 300-350; 375-425; 450-500; 525-575 relative to picture onset) as well as a No-TMS condition while they named pictures. The study comprised two sessions, one in which pictures with monosyllabic names were used and the other with bisyllabic names. We found that IFG stimulation around 450-500ms post-stimulus onset significantly delays naming latencies during monosyllabic naming, while not significantly disrupting bisyllabic naming. pMTG stimulation in the 450-500ms time-window delayed monosyllabic naming while facilitating bisyllabic naming at 225-350ms. pSTG slowed monosyllabic naming when stimulated at 375-425ms and at 450-500ms during bisyllabic naming. However, throughout testing it became obvious that some participants were naming pictures faster than others, which is problematic given the assumption of a relatively fixed mean naming latency underlying the choice of stimulation time-windows. For this reason, we also performed a response-locked analysis to better control for variability in individual naming latencies. Across both analyses, only IFG seems to show some consistency in effects around 400-500ms post-stimulus onset. With the lack of consistency across sessions and analyses, we call into question the validity of our results and provide a detailed discussion on the methodological confounds that need to be overcome in our own as well as other chronometric TMS studies before theoretical conclusion can be drawn from such studies.

Publication
Presentation at the International Workshop on Language Production, 2021
Adrian Jodzio
Adrian Jodzio
PhD candidate

Related