Relation between brain activation and lexical performance

Abstract

Functional magnetic resonance imaging (fMRI) was used to determine whether performance on lexical tasks was correlated with cerebral activation patterns. We found that such relationships did exist and that their anatomical distribution reflected the neurocognitive processing routes required by the task. Better performance on intramodal tasks (determining if visual words were spelled the same or if auditory words rhymed) was correlated with more activation in unimodal regions corresponding to the modality of sensory input, namely the fusiform gyrus (BA 37) for written words and the superior temporal gyrus (BA 22) for spoken words. Better performance in tasks requiring cross-modal conversions (determining if auditory words were spelled the same or if visual words rhymed), on the other hand, was correlated with more activation in posterior heteromodal regions, including the supramarginal gyrus (BA 40) and the angular gyrus (BA 39). Better performance in these cross-modal tasks was also correlated with greater activation in unimodal regions corresponding to the target modality of the conversion process (i.e., fusiform gyrus for auditory spelling and superior temporal gyrus for visual rhyming). In contrast, performance on the auditory spelling task was inversely correlated with activation in the superior temporal gyrus possibly reflecting a greater emphasis on the properties of the perceptual input rather than on the relevant transmodal conversions.

Figure 1

Activation maps showing correlation of activation with accuracy for the spelling (1st row) and rhyming (2nd row) tasks in the visual modality. Red indicates a positive correlation, greater activation associated with higher accuracy (A: fusiform gyrus; B: supramarginal gyrus; C: superior temporal gyrus), and green indicates a negative correlation, greater activation associated with lower accuracy (D: fusiform gyrus). Letters label regions of interest in the left hemisphere significant at P < 0.001 (yellow) and P < 0.01 (cyan). The figure displays all significant activation at the P < 0.01 level for presentation purposes. See Table V for number of voxels significant at P < 0.001 level.

Figure 2

Activation maps showing correlation of activation with accuracy for spelling (1st row) and rhyming (2nd row) in the auditory modality. Red indicates a positive correlation, greater activation associated with higher accuracy (A: supramarginal gyrus; B: angular gyrus; C: fusiform gyrus; D: superior temporal gyrus), and green indicates a negative correlation, greater activation associated with lower accuracy (E: superior temporal gyrus). Letters label regions of interest in the left hemisphere significant at P < 0.001 (yellow) and P < 0.01 (cyan). The figure displays all significant activation at the P < 0.01 level for presentation purposes. See Table VI for number of voxels significant at P < 0.001 level.

Figure 3

Overlap of activation (blue) between accuracy and reaction time for auditory spelling (1st row), auditory rhyming (2nd row, left) and visual rhyming (2nd row, right). The positive correlation for accuracy (higher accuracy correlation represented in red) overlaps with the positive correlation for reaction time (slower reaction time correlation represented in green) for auditory spelling (A: angular gyrus; B: fusiform gyrus). The positive correlation for higher accuracy (red) also overlaps with the negative correlation for reaction time (faster reaction time correlation represented in green) for auditory and visual rhyming (C: superior temporal gyrus). Letters label regions of interest in the left hemisphere significant at P < 0.001 (yellow) and P < 0.01 (cyan). The figure displays all significant activation at the P < 0.01 level for presentation purposes. (See Tables V and VI for number of voxels significant at P < 0.001 level.)

Figure 4

Positive correlations between accuracy and signal intensity for the intramodal tasks. Percent of variance (r²) explained is indicated for the auditory rhyming task for the left superior temporal gyrus (LSTG; ‐51,‐24,‐3) and for the visual spelling task in the left fusiform gyrus (LFG: ‐39,‐57,‐9). Dashed line and open circles, rhyming, LSTG (Rsq = 0.5406); solid line and open squares, spelling LFG (Rsq = 0.5557).

Figure 5

Positive correlations between accuracy and signal intensity for the cross‐modal auditory spelling task. Percent of variance (r²) explained is indicated for the left fusiform gyrus (LFG: ‐30,‐51,‐6), the left angular gyrus (LAG; ‐51,‐63,12) and the left supramarginal gyrus (LSG; ‐30,‐48,42). Dashed line and diamonds, LFG (Rsq = 0.4409); solid lines and circles, LAG (Rsq = 0.8326); Dotted line and squares, LSG (Rsq = 0.3428).

Figure 6

Positive correlations between accuracy and signal intensity for the cross‐modal visual rhyming task. Percent of variance (r²) explained is indicated for the left superior temporal gyrus (LSTG: ‐48,‐6,6), the right superior temporal gyrus (RSTG; 51,‐39,9) and the left supramarginal gyrus (LSG; ‐57,‐36,18). Dashed line and diamonds, LSTG (Rsq = 0.4660); Dotted line and circles, RSTG (Rsq = 0.7215); solid line and squares, LSG (Rsq = 0.6903).

Figure 7

Negative correlations between accuracy and signal intensity for the cross‐modal tasks. Percent of variance (r²) explained is indicated for the visual rhyming task for the left fusiform gyrus (LFG: ‐33,‐60,‐21), and for the auditory spelling task for the left superior temporal gyrus (LSTG; ‐60,‐21,9) and the right superior temporal gyrus (RSTG; 63,‐15,12). Dotted line and diamonds, rhyming, LFG (Rsq = 0.0894); solid line and circles, spelling, LSTG (Rsq = 0.5779); dotted line and square, Spelling, RSTG (Rsq = 0.4913).