With childhood hemispherectomy, one hemisphere can support-but is suboptimal for-word and face recognition

Abstract

The right and left cerebral hemispheres are important for face and word recognition, respectively-a specialization that emerges over human development. The question is whether this bilateral distribution is necessary or whether a single hemisphere, be it left or right, can support both face and word recognition. Here, face and word recognition accuracy in patients (median age 16.7 y) with a single hemisphere following childhood hemispherectomy was compared against matched typical controls. In experiment 1, participants viewed stimuli in central vision. Across both face and word tasks, accuracy of both left and right hemispherectomy patients, while significantly lower than controls' accuracy, averaged above 80% and did not differ from each other. To compare patients' single hemisphere more directly to one hemisphere of controls, in experiment 2, participants viewed stimuli in one visual field to constrain initial processing chiefly to a single (contralateral) hemisphere. Whereas controls had higher word accuracy when words were presented to the right than to the left visual field, there was no field/hemispheric difference for faces. In contrast, left and right hemispherectomy patients, again, showed comparable performance to one another on both face and word recognition, albeit significantly lower than controls. Altogether, the findings indicate that a single developing hemisphere, either left or right, may be sufficiently plastic for comparable representation of faces and words. However, perhaps due to increased competition or "neural crowding," constraining cortical representations to one hemisphere may collectively hamper face and word recognition, relative to that observed in typical development with two hemispheres.

Keywords: development; face recognition; hemispherectomy; plasticity; word recognition.

Fig. 1.

Participants viewed sequential pairs of faces and words (not drawn to scale here) in separate blocks. Participants were instructed to indicate, via one of two response keys, whether stimuli in a pair were the same or different. (A) In experiment 1, participants viewed all stimuli at central fixation. (B) In experiment 2, participants viewed the second stimulus in one of two visual hemifields to initially restrict processing to controls' single hemisphere (34), to compare to patients' preserved single hemisphere.

Fig. 2.

Violin plots show the distribution of overall accuracy values on the face and word recognition tasks for each group and stimulus category. Overlaid point plots show the individual values for each participant. The black points show the estimated probability of correct trials per condition, and the error bars represent the 95% confidence interval of these estimates (back transformed from a logarithmic scale). To visualize any effect of age, individual participants' points are separated by the median of the age distribution for all participants in the study (younger versus older than 17.5 y), shown in different shades of blue (controls) and red (patients). Asterisks indicate the data for two patients in the sample who had their hemispherectomies completed/revised as adults. Points are randomly jittered to minimize overlapping data. Note: Axes begin at accuracy = 0.5 because cases in which a participant performed at or below chance on a task were excluded from the analysis.

Fig. 3.

Violin plots show the distribution of overall accuracy values on the face and word recognition tasks for each group and stimulus category, by primary hemisphere used. Overlaid point plots show the individual values for each participant. The black points show the estimated probability of correct trials per condition, and the error bars represent the 95% confidence interval of these estimates (back transformed from a logarithmic scale). To visualize any effect of age, individual participants' points are separated by the median of the age distribution for all participants in the study (younger versus older than 17.5 y), shown in different shades of blue (controls) and red (patients). Asterisks indicate the data for two patients in the sample who had their hemispherectomies completed/revised as adults. Points are randomly jittered to minimize overlapping data. Note: Axes begin at accuracy = 0.5 because cases in which a participant performed at or below chance on a task were excluded from the analysis.

Fig. 4.

Correlation plots displaying each participant's accuracy on experiment 2 (peripheral stimulus presentation; y axis) versus accuracy on experiment 1 (central stimulus presentation; x axis) for faces (Left) and words (Right). The regression line of best fit for each group is overlaid, with 95% confidence intervals.

Fig. 5.

Violin plots show the distribution of overall accuracy values on the face and word recognition tasks for patients, separated by levels of conceivable predictors (per participant and/or guardian report). (A) Cognitive ability. (B) Occurrence of any seizures after patients' most recent surgery. (C) Approximate age at surgery (5 y and younger versus 6 y and older). Some patients underwent multiple surgeries prior to or to complete their hemispherectomy: Age at final surgery was used here. Overlaid point plots show the individual values for each participant. To visualize any effect of age at the time of testing of the study, individual participants' points are separated by the median of the age distribution for all participants in the study (younger versus older than 17.5 y), shown in different shades of red. Points are randomly jittered to minimize overlapping data. Note: Axes begin at accuracy = 0.5 because cases in which a participant performed at or below chance on a task were excluded from the analysis.