Abstract
Sensory substitution and augmentation depend on the brain's capacity to incorporate new sensory cues into perception, but the flexibility of underlying computations remains unclear. We compared how a newly learned depth cue is integrated, relative to familiar cues, after 1 h of training. Using forced-choice depth judgments in 78 observers, we assessed three indicators of integration for familiar and novel cues: (1) cue combination, predicting precision gains; (2) re-weighting, predicting reliability-based biases; and (3) congruence sensitivity, predicting increased sensitivity for learned mappings. Familiar cues were combined near-optimally, while only some people combined the novel and familiar cues. Test-retest measures suggested that this reflected reliable individual differences. Nonetheless, both familiar and novel cue pairs were re-weighted by reliability and showed sensitivity to incongruence. These results indicate that although novel cues can be rapidly mapped to depth and weighted appropriately, their integration varies across individuals, suggesting that individuals may vary in perceptual flexibility.