Poster C30, Friday, August 17, 10:30 am – 12:15 pm, Room 2000AB

Neural correlates of automatic semantic priming revealed by an event-related fMRI

Maria Varkanitsa1, Erin Meier2, Yue Pan2, David Caplan1, Swathi Kiran2;1Massachusetts General Hospital, Harvard Medical School, 2Boston University, Sargent College of Health & Rehabilitation Sciences

Introduction: Semantic priming occurs when a subject is faster in recognizing a target word when it is preceded by a related word-prime, compared to when it is preceded by an unrelated one (Neely, 1991). Previous fMRI studies (Copland et al. 2003; Rissman et al. 2003) have shown that automatic priming is reflected in reduced BOLD signal for related compared to unrelated prime-target pairs. In this study, we employed event-related fMRI to investigate brain activation associated with automatic semantic priming during a lexical decision (LD) task. Methods: Thirteen right-handed healthy subjects (5 female) participated in the study. Mean age was 60 years, with a range from 41 to 77 years. Participants were scanned while performing a LD. Auditory stimuli consisted of pairs of real words and nonwords, in which the first member of the pair is the prime and the second member is the target. There were three stimulus conditions. In the related condition, a real word target was preceded by a semantically related prime, which was a single word semantic feature (REL:32 trials). In the unrelated condition, a real word target was preceded by a semantically unrelated prime (UNREL:32 trials). In the nonword condition, a nonword target was preceded by a real word prime (NONW:64 trials). Each trial consisted of a 500ms fixation, the prime (~500ms), a 50ms interval, and the target (~500ms). T1 images were acquired with the following parameters: 176 sagittal slices, 1x1x1mm voxels, TR=2300ms. BOLD data were collected with the following parameters: 40 axial slices, 2x2x3mm voxels, TR=2570ms, TE=30ms. Functional images were co-registered to structural images and normalized to MNI space. Given the tight contrasts of interest, a threshold of p<0.01 (uncorrected) and a minimum cluster size of 10 was adopted to enable the identification of activation patterns across the three conditions. Results: Participants showed faster RT latencies for REL (375.77ms) than UNREL (412.59ms) or NONW (460.94ms). Accuracy was high across conditions (between 97%-98% correct). Considering the imaging data, stimulus pairs with word (related and unrelated) targets elicited greater activation compared to stimulus pairs with nonword targets in regions including: a) left hemisphere superior and middle frontal gyrus, superior and inferior parietal lobule, superior temporal gyrus, inferior frontal triangularis and opercularis, and orbitofrontal cortex, b) bilateral inferior and middle temporal gyrus, cuneus, precuneus, paracentral lobule, middle cingulate, and c) right hemisphere angular gyrus, caudate, and posterior cingulate. The contrast between unrelated and related targets elicited greater activation for the unrelated targets in regions including left middle temporal gyrus, bilateral middle frontal gyrus and inferior frontal triangularis, and right rectus, right inferior frontal orbitalis, and lateral and posterior orbitofrontal cortex. Greater activation associated with related targets was found in left middle frontal gyrus and inferior frontal triangularis. Summary: A priming related decreased BOLD was observed in LMTG and bilateral IFG consistent with Copland et al. (2003). However, priming related increased BOLD signal was observed in LMFG and LIFGtri for related compared to unrelated targets indicating a potential semantic facilitation effect, a finding in opposition to previous studies.

Topic Area: Meaning: Lexical Semantics