Framing effects in reading: an fMRI study

• ¹ Università degli studi di Milano-Bicocca, Psychology Department, Italy
• ² University of Trento, Center for Mind/Brain Sciences, Italy
• ³ Niguarda Ca’ Granda Hospital,, Neuroradiology Department, Italy
• ⁴ IRCCS Galeazzi, Italy

Introduction: According to the dual-route model, a printed string of letters can be processed through either a grapheme-to-phoneme conversion route or a lexical-semantic route. While meta-analyses of the imaging literature (Cattinelli, Borghese, Gallucci, & Paulesu, 2013; Taylor, Rastle, & Davis, 2012) have supported distinct and interacting procedures, individual neuroimaging studies exploring such neural correlates by means of reading, phonological and semantic tasks yielded results that remain not conclusive. Aims: To investigate the neural correlates of the lexical and sublexical reading procedures, we used a frame-manipulation paradigm whereby subjects read disyllabic Italian words embedded in lists of either irregular words, inducing lexical processing of the target words, or pseudowords, inducing sublexical reading processing of the target words. Possible linguistic confounders as word frequency, phonological complexity, orthographic neighborhood size and imageability were thus perfectly controlled. Methods: Thirty-three university subjects participated in the behavioral study and 22 were included in the fMRI study. Two reading tasks were created: in a first task, disyllabic word targets were embedded in filler lists (frames) formed of either loanwords or pseudowords, while, in the second task, disyllabic word targets were embedded in filler lists made of either trisyllabic words or pseudowords with a consonant-vowel structure. Moreover, at behavioral level, all subjects read a list of disyllabic target words (block condition). From fMRI data, we extracted: a) the lexical effect, computed as the activation pattern in the loanword frame and in the trisyllabic-word frame, while excluding areas showing the weakest trend for activation in the sublexical condition; b) the sublexical effect, computed as the activation pattern in the pseudoword frame, once weakest trend for activation in the lexical frame was excluded. It is worth emphasizing that the BOLD signal analyzed was always for reading real disyllabic words dispersed (40%) in the two different frames. Results and discussion: A significant main effect of the list condition emerged from behavioral data: participants were faster when reading disyllabic words embedded in a lexical frame than when reading targets embedded in a sublexical frame. Moreover, participants were faster when reading targets in a block condition than in either the lexical or the sublexical frame. Anatomofunctional results (Figure 1) showed that the left occipital, the anterior and posterior temporal regions, and the left intraparietal sulcus were specifically activated when reading targets in a lexical frame. The left posterior inferior temporal and inferior parietal regions were activated in sublexical condition. Finally, reading along the two routes commonly activated the Visual Word Form Area, the premotor cortex, the left frontal areas and the left SMA, suggesting an involvement of these regions in early-input (early orthographic processing) and late-output processes (phonological output buffer and articulatory programming). Conclusions: These results represent a new fine-grained description of the neurofunctional correlates of the dual route model partially supporting the recent anatomical investigations in patients with specific forms of acquired dyslexia (Ripamonti, Aggujaro, Molteni, Zonca, Ghirardi, & Luzzatti, 2014).