Modeling performance in observer theory (Q1802047)

In constructing perceptual models it is necessary to distinguish between competence and performance. Competence theory describes perceptual models in the absence of noise, whereas performance models take account of noise. Thus a competence model would be able to distinguish a rotation that is too small for a performance based model to detect. In this sense a performance based model will consider ``fewer'' interpretations for a class of data than would a competence based model. This paper extends the formalization of an observer (a competence based model) to a performance observer as an instantiation of an observer subject to noise.