This research examines human-human cooperation during the driving task to inform the design of automated vehicle (AV) human-machine interactions. It investigates how passengers intuitively infer drivers' intentions through body language and contextual cues, bypassing explicit communication. A mixed-method study introduced the "Driving Intention Grammar" (DiG), which defines key cues, sources, and sequences shaping passenger understanding. These insights informed the Intention Awareness (IA) Model, extending Endsley’s Situation Awareness framework to guide human-machine interface (HMI) design. The findings provide a foundation for anthropomorphic, context-aware, and multimodal HMIs, aiming to enhance AV transparency, predictability, and trust.