Cortical mechanisms of seeing and hearing speech

Abstract

In face-to-face communication speech is perceived through eyes and ears. The talker's articulatory gestures are seen and the speech sounds are heard simultaneously. Whilst acoustic speech can be often understood without visual information, viewing articulatory gestures aids hearing substantially in noisy conditions. On the other hand, speech can be understood, to some extent, by solely viewing articulatory gestures (i.e., by speechreading).

In this thesis, electroencephalography (EEG), magnetoencephalography (MEG) and functional magnetic resonance imaging (fMRI) were utilized to disclose cortical mechanisms of seeing and hearing speech.

One of the major challenges of modern cognitive neuroscience is to find out how the brain integrates inputs from different senses. In this thesis, integration of seen and heard speech was investigated using EEG and MEG. Multisensory interactions were found in the sensory-specific cortices at early latencies and in the multisensory regions at late latencies.

Viewing other person's actions activate regions belonging to the human mirror neuron system (MNS) which are also activated when subjects themselves perform actions. Possibly, the human MNS enables simulation of other person's actions, which might be important also for speech recognition. In this thesis, it was demonstrated with MEG that seeing speech modulates activity in the mouth region of the primary somatosensory cortex (SI), suggesting that also the SI cortex is involved in simulation of other person's articulatory gestures during speechreading.

The question whether there are speech-specific mechanisms in the human brain has been under scientific debate for decades. In this thesis, evidence for the speech-specific neural substrate in the left posterior superior temporal sulcus (STS) was obtained using fMRI. Activity in this region was found to be greater when subjects heard acoustic sine wave speech stimuli as speech than when they heard the same stimuli as non-speech.