Speakers adapt gestures to addressees' knowledge: implications for models of co-speech gesture | Synapse