Multiple sources of sensory information are combined to develop hand posture percepts in the intact system, but the combination of multiple artificial somatosensory percepts by human prosthesis users has not been studied. Here, we report on a case study in which a person with transradial amputation identified prosthetic hand postures using artificial somatosensory feedback. He successfully combined five artificial somatosensory percepts to achieve above-chance performance of 95.0% and 75.7% in identifying four and seven postures, respectively. We studied how artificial somatosensation and the extant hand representation are combined in the decision-making process by providing two mappings between the prosthetic sensor and the location of the sensory percept: (1) congruent, and (2) incongruent. The participant’s ability to combine and engage with the sensory feedback significantly differed between the two conditions. The participant was only able to successfully generalize prior knowledge to novel postures in the congruent mapping. Further, he learned postures more accurately and quickly in the congruent mapping. Finally, he developed an understanding of the relationships between postures in the congruent mapping instead of simply memorizing each individual posture. These experimental results are corroborated by a Bayesian decision-making model which tracked the participant’s learning.