http://worldcat.org/entity/work/id/137488205

A Mathematical Framework for an Improved Search Model

Search is currently modeled for DoD applications by a single exponential function. The two adjustable parameters are the time constant, t, characterizing the exponential; and the long time detection probability, P. Deficiencies of the classical model are: human performance data cannot typically be fit with a single exponent model; the probability of detection for short times is less than that predicted by the classical model; the effects of multiple targets and clutter can only be included by adjusting the two-model parameters, which is performed in an ad hoc manner and over-constrains the model. This paper introduces a neoclassical model that includes three processes: attending to the target, random wandering around the scene and attending to other targets/clutter. An expression involving three exponents associated with the three processes is derived and special cases are described. The new model provides uniform treatment of multiple targets and false detections and allows for the separate descriptions of multiple times scales within the search process. Searches can be separated into single region, field-of-view search, and multiple region, field-of-regard search. Field-of-view search can be further subdivided into long searches during which the observer may examine many targets and short searches which are completed after a few target examinations.

Open All Close All