Authors:

GARAGNANI, M., SHTYROV, Y., Kujala, T., Wennekers, T. & PULVERMULLER, F.

Reference:

Year of publication:

2008

CBU number:

6881

Abstract:

Recent EEG/MEG studies revealed that brain responses to the same speech sounds differ if the stimuli are presented in different task contexts. When subjects are not paying attention to the auditory input, their mismatch negativity (MMN) brain response is greater for the words than for matched meaningless pseudowords. In contrast, greater late N400 responses to pseudowords than to words emerge in tasks where subjects have to attend to the language stimuli. Current cognitive or neural models are unable to explain these data using a single, unifying account. We investigated the neural mechanisms underlying these opposite effects through neurocomputational and neurophysiological studies. Firstly, we built a neuroanatomically-grounded neural network model of the left perisylvian cortex, and used it to simulate early word acquisition processes by means of synaptic-weight adaptation based on a neurobiologically-realistic (Hebbian) learning rule. The trained networks were able to replicate both the MMN and N400 brain responses. In particular, variation of one of the model’s parameters (the area-specific inhibition, mediating the strength of competition between coactive representations) differently modulated the network response to words and pseudowords: strong inhibition (simulating passive, inattentive listening) produced larger responses to words than to pseudowords, whereas weak inhibition (simulating attentive conditions) produced the opposite result. Thus, the model explains the divergence between MMN and N400 results, providing a first unifying account, at the cortical-circuit level, of these data. In addition, it makes strong predictions on how attention processes modulate event-related brain responses to speech stimuli. Secondly, in order to experimentally test the validity of this model, we recorded neurophysiological responses to words and pseudowords when the subjects were asked to attend to the spoken input or ignore it. The brain responses (peaking at around 120-140ms after critical stimulus onset) were significantly larger to pseudowords than to words when subjects were attending to the stimuli; the opposite pattern (words ≥ pseudowords) emerged (within the same time window) in the ignore condition. In addition, words responses were not significantly modulated by attention, indicating a robust activation of word-specific memory traces. These data confirm the predictions of the model and provide further evidence supporting earlier suggestions that initial stages of lexical processing are not affected by attentional demands and may thus be automatic.