Since the inception of modern simulation the designers and users of training devices have attempted to replicate as many physical and functional stimuli as possible in the training device. There are three primary impediments to this activity: our frequent inability to specify the kinds of stimuli that are required, our technological difficulty in replicating some stimuli, and the cost of replicating stimuli.

The constraints cited above have led the training device community to develop the concept of "selective fidelity", meaning that we have to be very selective about the stimuli that we choose to replicate. This paper presents arguments that our definitions of selective fidelity now need to be altered to fit recent behavioral and engineering developments. Over the years we have improved our ability through research and analysis to define the important stimuli. Also, our engineering capability to replicate formerly difficult stimuli has improved significantly. Finally, there have been dramatic decreases in the cost of providing high fidelity simulation. In this paper we discuss our belief that while the concept of selective fidelity will remain important to the training device community, the definition of selective fidelity will be more focused on trainee learning requirements than on analytical and technological shortcomings.