Investigative Ophthalmology & Visual Science,34:1236. 1993. [ARVO 1993; Sarasota, FL.]
In the beginning (Julesz, 1980; Treisman & Gelade, 1980), "pop out" occurred in visual search and texture segregation tasks only when targets were defined by simple features (e.g., brightness, orientation, size). However, researchers soon found that feature conjunctions and even spatial relations among features would sometimes pop out (e.g., if feature values were very dissimilar, spatial relations coincided with 3D properties, or targets were prespecified for subjects). Some researchers have now provided partial solutions to this conundrum (Duncan & Humphreys, 1989; Enns & Rensink, 1990; Ramachandran, 1988; Wolfe, 1992). In this paper we focus more broadly on the question, "For which visual attributes is popout unconditional?" In a series of experiments we examine the factors of target certainty (1 vs. 3 possible targets) and type of search task (detection vs. shape discrimination) on visual attributes for which the claim of "pop out" has been made. We find that some features pop out unconditionally (e.g., brightness, color) while others pop out only in a detection task (e.g., 3D orientation). These findings support the need to distinguish theoretically among several levels of preattentive vision, each with its own set of operations and rules.