Science Department, Laboratory of Robot and Science, of the Japanese Meiji University can show 36 different facial expressions as a reaction to various words. For example, it'll look scared if it hears words like 'war', 'bomb' or 'rocket'.
Future Vision by Erwin Van Lun on this article
The translation of emotion->facial expression is shaping up nicely. The translation of word->emotion is a lot more difficult. Who’s speaking? Where? What context? What are the memories associated with this word? That’s a lot more complicated. What this does demonstrated is how we’re capable of shaping emotions in a rubber face. Expressions that we see more and more often with brand agents, artificial and virtual characters that represent brands. We’ll perfect this technique in the coming decades in the virtual world and then bring it back into the physical world, to robots.