Granger causality mapping reveals congruency-dependent directed influences from superior temporal to auditory cortex during audiovisual integration

Nienke van Atteveldt, Alard Roebroeck, Rainer Goebel
Poster
Time: 2009-06-30  09:00 AM – 10:30 AM
Last modified: 2009-06-04

Abstract


Objective
How do sensory-specific and multisensory brain regions interact during multisensory integration? Previous neuroimaging studies provide contradictory evidence, especially with regard to the role of multisensory regions such as the superior temporal sulcus (STS).
Here, we directly examined the role of STS in the audiovisual integration of letters and speech sounds by mapping directed interactions to and from STS over the brain. Based on previous results [1,2] we predicted that directed influence from STS to auditory cortex is stronger for congruent than for incongruent letter-sound pairs.

Methods
10 subjects passively perceived randomly presented letters, speech sounds, and congruent or incongruent letter-sound pairs. Scanning (3T) repetition time was 2s and volume acquisition time 1.5s; stimuli were presented in the silent 500ms delay. Random-effects general linear models (GLMs) were used to create statistical maps of the congruency contrast [congruent vs. incongruent]. Effective connectivity was assessed using Granger Causality Mapping (GCM) [3], which identifies voxels that are sources/targets of directed influences to/from a selected reference region-of-interest (ROI). We selected ROIs in STS in 10 individual subjects using the conjunction of [Auditory > baseline] ∩ [Visual > baseline], which indentifies heteromodally responsive voxels. Using this left STS ROI as reference region, single-subject Granger causality maps (GCMs) were calculated for the congruent and incongruent conditions. Next, a random-effects group GCM was created to test modulation of effective connectivity by congruency (modulation map).

Results
The GCM modulation map demonstrates an influence from the left STS to auditory cortex bilaterally and right occipital cortex that was stronger during congruent letter-sound pairs than incongruent pairs. The auditory cortex regions targeted by STS are the same as in which a congruency effect (congruent > incongruent) was observed using GLM analysis in the present as well as in previous studies [1,2].

Conclusions
The GCM results indicate that the STS provides congruency-dependent feedback to auditory and visual sensory-specific regions during audiovisual integration. This suggests that multisensory STS can be a crucial component during integration of meaningful auditory and visual information. Mapping directed influences to and from the targeted low-level sensory regions is currently in progress, to further confirm and extend the present results.

[1] van Atteveldt et al., Neuron, 2004
[2] van Atteveldt et al., Cerebral Cortex, 2007
[3] Roebroeck et al., Neuroimage, 2005

Conference System by Open Conference Systems & MohSho Interactive Multimedia