Dissociable laminar profile of “what” and “when” predictions in the human auditory cortex

2496 

Abstract Submission 

Caspar Schwiedrzik¹, Ryszard Auksztulewicz², Daniel Fabo³, Johanna Szabo³, Istvan Ulbert³, Wittner Lucia³, Thomas Thesen⁴, Werner Doyle⁵, Orrin Devinsky⁵, Lucia Melloni⁶

¹European Neuroscience Institute, Göttingen, Germany, ²Freie Universität Berlin, Berlin, Berlin, ³National Institute of Mental Health, Neurology and Neurosurgery, Budapest, Hungary, ⁴New York University Comprehensive Epilepsy Center, New York, NY, USA, ⁵New York University School of Medicine, New York, NY, USA, ⁶Max Planck Institute for Empirical Aesthetics, Frankfurt am Main, Germany

Caspar Schwiedrzik  
European Neuroscience Institute
Göttingen, Germany

Ryszard Auksztulewicz  
Freie Universität Berlin
Berlin, Berlin

Daniel Fabo  
National Institute of Mental Health, Neurology and Neurosurgery
Budapest, Hungary

Johanna Szabo  
National Institute of Mental Health, Neurology and Neurosurgery
Budapest, Hungary

Istvan Ulbert  
National Institute of Mental Health, Neurology and Neurosurgery
Budapest, Hungary

Wittner Lucia  
National Institute of Mental Health, Neurology and Neurosurgery
Budapest, Hungary

Thomas Thesen  
New York University Comprehensive Epilepsy Center
New York, NY, USA

Werner Doyle  
New York University School of Medicine
New York, NY, USA

Orrin Devinsky  
New York University School of Medicine
New York, NY, USA

Lucia Melloni  
Max Planck Institute for Empirical Aesthetics
Frankfurt am Main, Germany

Predictions based on statistical regularities in the environment are fundamental for adaptive behaviour. However, while predictions of different stimulus features may facilitate sensory processing, they are likely mediated by dissociable neural mechanisms. In previous work [Auksztulewicz et al., 2018], based on direct electrophysiological recordings from the cortical surface in humans, we could elucidate the dissociable neural mechanisms of predicting the contents ("what") and timing ("when") of auditory stimuli. While some (e.g., frontal and premotor) cortical regions were preferentially modulated by either "what" or "when" predictability, the superior temporal gyrus (STG), a higher-order auditory region, integrated predictability of both stimulus features. A Bayesian model fitted to the empirical data suggested that "what" and "when" predictability can be linked to complementary gain modulation mechanisms in the STG.

Here, we tested whether these functional dissociations can be mapped onto different laminar activity profiles in the STG. We analysed local field potentials (LFPs) and broadband activity recorded invasively from the STG using laminar probes. Recordings were conducted in five patients with pharmacologically intractable epilepsy undergoing surgical monitoring. Patients were asked to perform a speeded categorisation task on auditory syllables preceded by visual stimuli, allowing for a formation of "what" and/or "when" predictability (Fig. 1A). The laminar probes spanned the entire cortical sheet and signals were realigned across patients using the current source density profiles of auditory-evoked activity (Fig. 1B).

"What" and "when" predictability was linked to dissociable laminar profiles of activity in the STG (Fig. 1C). Specifically, predictability of stimulus contents modulated activity in both supragranular and infragranular layers (receiving projections from higher-order regions), consistent with a recent study in macaque monkeys [Lakatos et al., 2020]. Predictability of stimulus timing had a more selective effect on infragranular layers, suggesting a partial laminar dissociation between the two types of predictions. Conversely, both types of predictions interactively modulated neural activity in infragranular layers but also in granular layers (receiving projections from lower-order regions), corroborating earlier theoretical proposals of how "what" and "when" predictions may be integrated in the cortex [Musacchia et al., 2014].

Taken together, our results provide direct evidence for functional dissociations between different types of predictability in terms of their effects on cortical microcircuits.

Microcircuitry and Modules ²

Perception: Multisensory and Crossmodal ¹

Cortical Layers

ELECTROCORTICOGRAPHY

Perception

Other - cross-modal; audition; prediction; current source density analysis