MMN reflections of language and attention: a neurocomputational model

Garagnani, M.; Wennekers, Thomas and Pulvermüller, Friedemann. 2009. 'MMN reflections of language and attention: a neurocomputational model'. In: MMN 09 Fifth Conference on Mismatch Negativity (MMN) and its Clinical and Scientific Applications. Budapest, Hungary 4-7 April 2009. [Conference or Workshop Item]

No full text available

Abstract or Description

Recent EEG and MEG studies have 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) response is greater for words than for matched meaningless pseudowords. However, greater late N400 responses to pseudowords than to words emerge in tasks where subjects have to attend to the stimuli. We (Garagnani, Wennekers, & Pulvermüller, 2008) have recently proposed an explanation for these divergent patterns of results based on simulations obtained with a neural-network model mimicking areas of the human left perisylvian language cortex and their connectivity, along with known neurobiological mechanisms of synaptic plasticity. Repeated co-activation of the model’s primary motor and auditory cortex with predetermined patterns led to the spontaneous formation in the network of strongly connected distributed circuits (the possible brain-correlates of words) that exhibited input-specificity and functional discreteness. The resulting model was then used to simulate neurophysiological brain responses to meaningful familiar word and senseless unknown pseudoword stimuli under different amounts of processing resources. Variation of a single parameter, the network’s global inhibition feedback (the model correlate of attention) replicated and explained the divergence between MMN and N400 results, providing the first unifying account for these data. In particular, weak inhibition produced late activation differences, with stronger responses to pseudowords than to words (mirroring the N400 pattern). In contrast, strong inhibition led to early differences, with stronger responses to words than to pseudowords; this closely resembles the pattern seen in the MMN data. These dynamics were largely due to a modulation of the activation produced by the pseudoword stimuli: whereas such responses were strongly affected by the degree of inhibition/attention in the network, word activations were not. The discrete character of the neuronal circuits that developed for words lies at the basis of such differential modulation: changes in the inhibition/attention levels have little influence on the activation dynamics of functionally discrete assemblies; on the other hand, network responses to items (pseudowords) that have not been previously learned – and, therefore, lack a corresponding memory representation – are strongly modulated by the amount of inhibition/attention, for this mediates the amount of competition between the partially activated multiple memory circuits. Novel simulations obtained with the same model using oddball stimulation demonstrate how the MMN response may reflect both mechanisms of automatic change detection and long-term linguistic memory trace activation.

Item Type:

Conference or Workshop Item (Paper)

Identification Number (DOI):

https://doi.org/10.3389/conf.neuro.09.2009.05.041

Departments, Centres and Research Units:

Computing

Dates:

DateEvent
23 March 2009Published

Event Location:

Budapest, Hungary

Date range:

4-7 April 2009

Item ID:

27716

Date Deposited:

13 Dec 2019 10:12

Last Modified:

13 Dec 2019 10:12

URI:

https://research.gold.ac.uk/id/eprint/27716

Edit Record Edit Record (login required)