The first 250 ms in visual word recognition: who is talking to whom

Muriel Lobier, Peter Hansen, Katherine Wheat, Sylviane Valdois, Piers Cornelissen

Research output: Contribution to conferencePosterpeer-review

Abstract

Visual word recognition results from a dynamic interplay between multiple nodes in a distributed cortical and sub-cortical network. To fully understand how it is achieved, we need to identify not only the necessary and sufficient compliment of nodes that comprise this network, but also how information flows through the nodes over time. Of particular interest are those parts of the network which support phonological access during visual word recognition. Recent MEG studies of reading have highlighted early (~100 ms after target onset) neural activity in the Left Inferior Frontal Gyrus (LIFG) (Cornelissen et al., 2009), activity which can be modulated by phonological priming (Wheat et al., 2010). Whilst this is consistent with rapid phonological access driven by activity in higher order visual areas, the alternative possibility of modulation driven by top-down control processes cannot be excluded. We therefore explored directed connectivity between visual areas and the LIFG during reading. If LIFG activity reflects early access to phonology, we hypothesized that there should be early directed coupling from visual areas to LIFG. We used Partial Directed Coherence (PDC), a frequency domain measure of Granger Causality, to measure effective connectivity between MEG time-series. Not only is PDC able to identify direction of information flow, it can distinguish between direct and indirect connectivity (Baccala, 2001). Furthermore, estimation of time varying PDC has been shown to reliably identify rapidly changing connectivity relationships between brain signals (Astolfi, 2011). Participants carried out a silent single word reading task in MEG. Beamformer source-space analysis was used to identify six local maxima (nodes) for each participant and individual time-series were reconstructed for each one. Using a resampling method, PDC time-frequency distributions for each node pair were estimated. In order to identify reading-specific connectivity, average active phase PDC for each time-frequency bin was compared to time-averaged passive phase PDC for each frequency bin . Statistical significance was estimated using a permutation approach, with a t-statistic distribution calculated for each time-frequency bin. From this distribution, a threshold t-value was defined and a corresponding thresholded PDC time-frequency plot produced. Results confirm a significant early connectivity peak from the Left Middle Occipital Gyrus to the LIFG with no significant reciprocal connectivity. This result supports the hypothesis that visual areas directly drive early LIFG activation, consistent with early bottom-up access to phonological information in reading.

Keywords

  • language
  • magnetoencephalography

Fingerprint

Dive into the research topics of 'The first 250 ms in visual word recognition: who is talking to whom'. Together they form a unique fingerprint.

Cite this