In the future, as our wearable computers ``get to know'' us better they could be empowered to help us in the way we would want them to, taking notes for us when we were not paying attention, sending a warning to our loved ones when we were in danger, adapting our flow of information to provide more when we are ready for it and less when our human buffers are full and we show signs of stress. This paper describes a wearable system, including a startle detection algorithm, which is robust across several wearers, and which enables the wearable to automatically respond to events of potential interest to the wearer. Several methods of controlling the flow of video information recorded by StartleCam have been demonstrated including transmitting images surrounding highly arousing events back to a remote server in the Internet as a safety device and the automatic logging of uneventful images when the system notices an unusually low number of responses. Working toward the vision of the cyborg as man and machine working in automatic cooperation, the physiological control of the StartleCam allows humans to offload tasks that are better handled by computers and focus on tasks that are creative and engaging.