Prior Visual Experience Modulates Learning of Sound Localization Among Blind Individuals

Abstract

Cross-modal learning requires the use of information from different sensory modalities. This study investigated how the prior visual experience of late blind individuals could modulate neural processes associated with learning of sound localization. Learning was realized by standardized training on sound localization processing, and experience was investigated by comparing brain activations elicited from a sound localization task in individuals with (late blind, LB) and without (early blind, EB) prior visual experience. After the training, EB showed decreased activation in the precuneus, which was functionally connected to a limbic-multisensory network. In contrast, LB showed the increased activation of the precuneus. A subgroup of LB participants who demonstrated higher visuospatial working memory capabilities (LB-HVM) exhibited an enhanced precuneus-lingual gyrus network. This differential connectivity suggests that visuospatial working memory due to the prior visual experience gained via LB-HVM enhanced learning of sound localization. Active visuospatial navigation processes could have occurred in LB-HVM compared to the retrieval of previously bound information from long-term memory for EB. The precuneus appears to play a crucial role in learning of sound localization, disregarding prior visual experience. Prior visual experience, however, could enhance cross-modal learning by extending binding to the integration of unprocessed information, mediated by the cognitive functions that these experiences develop.

Keywords: Blindness; Experience modulation; Functional connectivity; Sound localization; fMRI.

Fig. 1

Locations for the low- and high-resolution sound stimuli and definition of correct responses in the sound localization task. The neighboring locations are also regarded as correct responses. The neighboring locations have the same distance or azimuth as the exact correct locations, but with one-step difference in distance or azimuth. The locations indicated as blue circles (e.g.,1.5 m/−30°), each has two neighboring locations; the locations indicated as green triangles (e.g., 2.5 m/−30° or 1.5 m/−15°), each has three neighboring locations; the locations indicated as orange stars (e.g., 2.5 m/+15°), each has four neighboring locations. The chance levels of accuracy for the blue, green and orange locations are 33.33, 25, and 20%, respectively. (Color figure online)

Fig. 2

Behavioral performance on the sound localization task in the EB (n = 11), LB-LVM (n = 6), and LB-HVM (n = 7) groups

Fig. 3

t maps for contrasts testing the main effect of sound localization processing (Localization–Discrimination) on the pre-training and post-training scan. P < 0.001 at the voxel level

Fig. 4

t maps for longitudinal contrasts testing the modulation effect of the training. P < 0.005 at the voxel level

Fig. 5

ROI results showed a significant Group × Training interaction effect. a The left precuneus. b The right precuneus

Fig. 6

PPI results showed training enhanced functional connectivity between a the left precuneus and a distributed network in the EB group, b the left precuneus and occipital regions in the LB-HVM group, and c the right precuneus and occipital regions in the LB-HVM group