Dissociation of reading and naming in ventral occipitotemporal cortex

Abstract

Lesions in the language-dominant ventral occipitotemporal cortex (vOTC) can result in selective impairment of either reading or naming, resulting in alexia or anomia. Yet, functional imaging studies that show differential activation for naming and reading do not reveal activity exclusively tuned to one of these inputs. To resolve this dissonance in the functional architecture of the vOTC, we used focused stimulation to the vOTC in 49 adult patients during reading and naming, and generated a population-level, probabilistic map to evaluate if reading and naming are clearly dissociable within individuals. Language mapping (50 Hz, 2829 stimulations) was performed during passage reading (216 positive sites) and visual naming (304 positive sites). Within the vOTC, we isolated sites that selectively disrupted reading (24 sites in 11 patients) or naming (27 sites in 12 patients), and those that disrupted both processes (75 sites in 21 patients). The anteromedial vOTC had a higher probability of producing naming disruption, while posterolateral regions resulted in greater reading-specific disruption. Between them lay a multi-modal region where stimulation disrupted both reading and naming. This work provides a comprehensive view of vOTC organization-the existence of a heteromodal cortex critical to both reading and naming, along with a causally dissociable unimodal naming cortex, and a reading-specific visual word form area in the vOTC. Their distinct roles as associative regions may thus relate to their connectivity within the broader language network that is disrupted by stimulation, more than to highly selective tuning properties. Our work also implies that pre-surgical mapping of both reading and naming is essential for patients requiring vOTC resections, as these functions are not co-localized, and such mapping may prevent the occurrence of unexpected deficits.

Keywords: alexia; anomia; cortical stimulation mapping; dyslexia; language; reading.

Figure 1

Experimental design and patient coverage. (A) Schematic examples of the two tested language tasks. (B) Representative single-patient examples of subdural electrode (SDE) and stereo-electroencephalography (sEEG) coverage. (C) Surface-based coverage map of stimulation sites on a standardized inflated pial surface.

Figure 2

Co-localized stimulation sites. (A) Population-normalized N27 inflated pial surface highlighting the co-localized stimulation sites, showing positive (red) and negative (white) sites for each task (Reading: 216 positive, 783 negative; Naming: 304 positive, 1540 negative). (B and C) Surface-based representation of (B) the task disruption probability, the probability of disrupting the task given stimulation at that location, and (C) the chi-squared statistic, representing regions where the probability of task disruption was greater than chance. Regions with coverage <2 patients are shown in grey.

Figure 3

Comparative spatial topography of reading and naming. (A) Conjunction maps and (B) log odds ratio map of task disruption comparing the reading and visual naming tasks. Conjunction maps show only sites tested with both tasks within a single session. Odds ratio maps compare node-wise probabilities of disruption at any given location across all sessions, constrained to areas showing a significant disruption to the function of either reading or naming (see Fig. 2C).

Figure 4

Relationship with functional imaging of reading. Prior published coordinates for functionally-word-selective areas in ventral occipitotemporal cortex from functional MRI (fMRI; red) and intracranial recording (blue) studies (see Supplementary Table 1), compared with the centroid of all causally-reading-selective electrodes (yellow) and all causally-naming-selective electrodes (orange) from the present study. The cortical surface is coloured to denote the relative probabilities of reading and naming disruption (see Fig. 3B).