Reading the reading brain: A new meta-analysis of functional imaging data on reading

Over the last 20 years, reading has been the focus of much research using functional imaging. A formal assessment of the implications of this work for a more general understanding of reading processes is still lacking. We performed a new meta-analysis based on an optimized hierarchical clustering algorithm which automatically groups activation peaks into clusters; the functional role of the clusters was assessed on the basis of statistical criteria. We considered the literature from 1992 to 2008, focussing exclusively on experiments based on single words or pseudowords from the following four classes of tasks: reading, lexical decision, phonological decision and semantic tasks. Our analysis was restricted to alphabetic orthographies and was based on 35 studies. We identified three networks: (1) a difficulty modulated network including Broca's area and attention-related brain regions; (2) a word-related network, primarily involving regions of the left temporal lobe and of the anterior fusiform region, known to participate to semantic processes; (3) a pseudoword-related network in the basal occipito-temporal regions and in the left inferior parietal cortex. These subnetworks constitute the basis upon which a plausible functional model of reading is proposed, where orthographic, phonological, and semantic processes are recruited to compute the phonology of a written stimulus based on cooperative and competitive mechanisms. The results of this meta-analysis held face validity when compared with the results of literature published until mid 2010, the time of completion of data collection.

► We aim at clarifying the functional-anatomical picture concerning reading processes.
► We introduce a rigorous methodology for meta-analysis of data reported in the literature.
► A tentative functional model is sketched where word-related areas are interpreted as the substrate of semantic processing.
► Pseudoword-related areas are linked to mapping orthographic representations into phonological ones.
► Difficulty-modulated areas contribute to final phonology by means of integrative and competitive mechanisms.