Left inferior parietal lobe and auditory cortex jointly contribute to sound knowledge retrieval

1002 

Abstract Submission 

Philipp Kuhnke^1,2, Johannah Voeller², Vincent Cheung³, Ole Numssen², Konstantin Weise^2,4, Markus Kiefer⁵, Gesa Hartwigsen^1,2

¹Leipzig University, Leipzig, Saxony, Germany, ²Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Saxony, Germany, ³Sony Computer Science Laboratories, Tokyo, Japan, ⁴Leipzig University of Applied Sciences, Faculty of Engineering, Leipzig, Saxony, Germany, ⁵Ulm University, Ulm, Baden-Württemberg, Germany

Philipp Kuhnke  
Leipzig University|Max Planck Institute for Human Cognitive and Brain Sciences
Leipzig, Saxony, Germany|Leipzig, Saxony, Germany

Johannah Voeller  
Max Planck Institute for Human Cognitive and Brain Sciences
Leipzig, Saxony, Germany

Vincent Cheung  
Sony Computer Science Laboratories
Tokyo, Japan

Ole Numssen  
Max Planck Institute for Human Cognitive and Brain Sciences
Leipzig, Saxony, Germany

Konstantin Weise  
Max Planck Institute for Human Cognitive and Brain Sciences|Leipzig University of Applied Sciences, Faculty of Engineering
Leipzig, Saxony, Germany|Leipzig, Saxony, Germany

Markus Kiefer  
Ulm University
Ulm, Baden-Württemberg, Germany

Gesa Hartwigsen  
Leipzig University|Max Planck Institute for Human Cognitive and Brain Sciences
Leipzig, Saxony, Germany|Leipzig, Saxony, Germany

Conceptual knowledge is central to human cognition. Previous neuroimaging studies suggest that conceptual processing relies on the joint contribution of modality-specific perceptual-motor and multimodal brain regions (Kuhnke et al. 2023). In particular, the multimodal left inferior parietal lobe (IPL) coupled with auditory cortex during sound knowledge retrieval and with somatomotor cortex during action knowledge retrieval (Kuhnke et al. 2021). However, as neuroimaging is correlational, it remains unknown whether the interaction between modality-specific and multimodal cortices is causally relevant for conceptually-guided behavior. To tackle this issue, we applied inhibitory transcranial magnetic stimulation (TMS) over modality-specific cortex (somatomotor, auditory, or sham), before 24 healthy participants received TMS over multimodal cortex (IPL, or sham) during action and sound judgment tasks on written words (Figure 1A).

To optimize the subject-specific coil position and intensity for each stimulation target, we performed a priori computational electric field (e-field) simulations (Figure 1B). Specifically, we determined the coil positions that maximize the e-field magnitude in each target, and matched the stimulation intensities to the subject-specific cortical stimulation thresholds (i.e. the e-field magnitude in primary motor cortex at resting motor threshold; Numssen, Kuhnke et al. 2023).
Behavioral data were analyzed using a drift diffusion model (DDM), which models response times and accuracies collectively, accounting for potential speed–accuracy tradeoffs and response biases (Voss and Voss 2007). The DDM assumes that during binary decision processes, the subject accumulates evidence for a certain decision from a stimulus, beginning at a starting point (z), until a decision boundary (a = "yes", or 0 = "no") is reached (Figure 2A). The key parameter-of-interest is the drift rate (v), the average rate of evidence accumulation. Thus, after the DDM was fit to the data, we normalized the drift rates for all TMS conditions to sham stimulation and performed Bayesian Wilcoxon signed-rank tests on the sham-normalized drift rates. Finally, Bayesian linear regression tested for relationships between drift rates and e-field strength in each target.

We found that combined stimulation of the left auditory cortex and IPL selectively impaired drift rates for sound judgments on low sound–low action words (Figure 2B). The data were >5 times more likely under the hypothesis that auditory + IPL TMS impaired sound judgments than under the null hypothesis of no TMS effect (BF10 = 5.316). Post-hoc paired comparisons provided evidence for anatomical specificity: Combined auditory + IPL TMS impaired sound judgments more strongly than auditory-only TMS (BF10 = 4.244), somatomotor + IPL TMS (BF10 = 1.854), somatomotor-only TMS (BF10 = 1.887), and IPL-only TMS (BF10 = 1.426). Moreover, we found evidence for task specificity: TMS over auditory cortex + IPL induced a stronger impairment of sound judgments than action judgments on the same words (BF10 = 1.499). Crucially, stronger stimulation of left auditory cortex was associated with worse performance on sound judgments under auditory + IPL TMS (Figure 2C; BF_M = 4.189). No other condition showed convincing evidence for a TMS effect.

Our results indicate that the joint contribution of multimodal IPL and auditory cortex is causally relevant for sound knowledge retrieval: The functional relevance of left IPL depends on the integrity of the auditory cortex, and vice versa, as single perturbation of either region did not disrupt performance. However, our data do not provide evidence for a functional relevance of the joint contribution of multimodal IPL and somatomotor cortex to action knowledge retrieval. These findings suggest substantial robustness of the conceptual system to disruption, which could be further studied by combining non-invasive brain stimulation with neuroimaging.

TMS ²

Language Comprehension and Semantics ¹

Bayesian Modeling

Perception: Visual

Cognition

Cortex

Language

Modeling

Motor

Transcranial Magnetic Stimulation (TMS)

Other - Semantics; Concepts; Auditory; Multimodal