ABSTRACT
Recent advances in neuroscience offer unprecedented insight into human perception and cognition. Powerful neuroimaging methods, such as functional magnetic rcsonance imaging (fMRl) and electrocncephalography (EEG) allow scientists to directly measure dynamic and subtle brain states . I t is easy to imagine how this work could lead to far-rcaching improvements in medical diagnosis and treatmcnt. The implications for revolutionary advances in the way that humans and machines interact in the real world are less obvious. In this paper, we describe an application of neurotechnology that combines the complementary power of the brain and the computer to process complex natural imagery. The goal of this work is a system that can rapidly scan a very wide field-of-view for threats and alert the user to their presence before they can harm him. An automated algorithm is able to quickly process images over a wide area to identify potential threat obj ects . However, these algorithms also detect many uninteresting obj ects (i. e. , a bush moving in the wind or rock that looks like a truck) . If there are too many of these false alarms, the operator is perpetually alerted and the system is useless . By incorporating the human visual system to sort through the objects that the algorithm identifies, the number of algorithm-generated false alarms can be reduced to an acceptable level .
