Learning new perceptual skills: Individual differences in the computations that integrate novel sensory cues into depth perception

Sensory substitution and augmentation rely on the brain’s ability to integrate novel sensory cues into its perceptual repertoire. However, the flexibility of the computations that support augmented perception is still not fully known. Here, we contrasted how a novel depth cue is processed, as compared with familiar depth cues, following one hour of training. Observers (N=78) made forced-choice comparisons of surface distances (depths), while we assessed three markers of integration with familiar and newly learned cues: (1) cue combination, predicting precision benefits; (2) re-weighting, predicting reliability-weighted biases; and (3) congruence sensitivity, predicting increased sensitivity for the learned cue mapping. We found that, (1) while familiar cues (size and binocular disparity) were combined near-optimally, there was little evidence for a novel cue (auditory pitch) being combined with binocular disparity. Measures of repeatability across two sessions suggest that this was due to reliable inter-individual differences in combination with the novel cue. In contrast, (2) both familiar and novel cue pairs were re-weighted by their relative reliabilities and (3) showed sensitivity to incongruence. These direct comparisons show that, while a novel cue was rapidly mapped onto depth and weighted in line with its relative reliability, it was not integrated into the native perceptual repertoire by everyone. Reliable individual differences suggest that abilities to combine novel sensory cues may vary between people. These findings provide insight into the flexibility of human sensory processing and suggest that learning to interpret new sensory information depends in part on individual flexibility in perceptual computations.