Translating Affective Touch into Text

This paper presents a game-like experience that translates tactile input into text, which captures the emotional qualities of that touch. We describe the experience and the system that generates it: a plush toy instrumented with pressure sensors, a machine learning method that acquires a mapping from touch data into a feature vector of affect values, and a mechanism that transcribes that feature vector into text. We conclude by discussing the range of novel interactions that such a nuanced tactile interface can support.