We use the expression ``cybernetic control'' to describe the control loops formed by the nervous system, the brain, and the wearable mechanical-electrical communication system. The StartleCam control links these two feedback systems, using an autonomic nervous system response-the skin conductivity startle response-to automatically control the electro-mechanical wearable computer and communicate selective information to the wearer's local digital archives or a remote server on the Internet.
This control loop might also be considered ``cyborgian'' as the term cyborg denotes man in quasi-symbiotic union with electro-mechanical homeostatic control systems. The concept of a cyborg was originally envisioned for astronauts to be able to regulate their homeostatic environment without conscious attention in the hostile environment of space[Cly95]. However, the idea is equally relevant to people on earth attempting to regulate their homeostasis against the stress created by information overload. By using physiological cues, the wearer's computer can react and respond in real time to unexpected events. In this aspect, StartleCam has an advantage over software agents that use a priori belief models of the user preferences [Mae94], although agents may someday use these cues to help train their belief models.
The startle response has been linked to reactions of sudden fright [LeD94], anticipation of bad results [Dam94], and stressful situations [Hel78] [Lev92] that are deviations from homeostasis [LG88]. It is also often referred to as the orienting response, because any interruption in attention can generate it, not jut the extreme startle reaction. By saving images when the startle response is detected, the StartleCam application models the wearer's own capacity for selective memory, according to a theory that memories are formed when survival is threatened [Dam94]. This effect is what is responsible for the the so-called ``flashbulb memory [BK77]'' where extremely arousing events seem to be highlighted and stored with unusual clarity. StartleCam offers three different modes of operation, direct control, automatic logging, and image series capture. The camera can be controlled directly by the wearer, through the input device, to record any data of specific interest as snapshot or an image series. When the startle response is detected, a composite image of recent events can be created by saving several frames from the video camera in rapid succession as shown in Figure 8. By setting a high threshold for the startle detector, StartleCam will record only the most arousing or threatening events. This mode of operation would be most useful for an application such a ``SafetyCam'' in which images of threatening events are transmitted to secure websites of the wearer's SafetyNet of friends and family[Man97]. In the opposite mode, the camera can be set to automatically record images at a set frequency when very few responses have been detected from the wearer, indicating that their attention level has dropped. This mode of operation is useful when the value of a lecture, a meeting or a contact is not immediately apparent and our natural selective memory system designed to survive immediate threats fails us.
StartleCam's cybernetic control loop allows the wearer to control applications on the wearable computer with minimal effort. This allows the wearer to offload tasks for which computers are well suited. For example, human memory tends to be inaccurate for recalling the exact details of events[BK77], and memories tend to change over time, but a computer can easily store a video archive with time stamps to aid recall. The vision of the cyborg has the computer cooperating with the human to manage cognitive load and mediate the flow of and storage of information; however, direct control should never be taken from the user. StartleCam allows the user to directly add or delete images from the video archive and start or stop an automatic process of capturing video images at regular time intervals throughout the day.