Iconoclasm Redux: Limits to the Involvement of Iconic Memory in Visual Search
R.A. Rensink, Cambridge Basic Research, Nissan Research & Development, Inc., Cambridge, Massachusetts, USA.

Investigative Ophthalmology & Visual Science, 38:707. 1997. [ARVO 1997; Ft. Lauderdale, FL.]

Purpose. To determine the extent to which visual search can be based on iconic memory (informational persistence).

Methods. Performance was measured using simple stimuli (bars and lines) under flicker conditions, i.e., where displays continually alternated between "on" (exposure times 80-200 ms) and "off" (blank fields; ISIs 120-240 ms). Two variants of search task were used. In the first, targets were fixed, with properties unique to the display (e.g., a vertical line among slanted lines). In the second, targets had nonunique properties that changed between successive displays (e.g., changing from vertical to slanted among fixed vertical and fixed slanted lines). For both variants, the degree to which search was unaffected by off-times reflected the ability of search to be carried out on iconic memory.

 Results. When searching for a fixed target, speeds were largely unaffected by off-times (up to 240 ms), indicating that search could use iconic memory, and that it was as fast as search based on direct visual input. However, results for a changing target were quite different: search slowed down dramatically as off-time was increased, indicating that iconic memory could not be used.

Conclusions. Although the contents of iconic memory are involved in some visual processes, they are not involved in all—processes do exist that operate only when the signal is available in the incoming light. A key factor may be whether the process requires high-level matching (fixed targets) versus low-level readout of stimulus properties at each alternation (changing targets).

Supported by Nissan Research & Development, Inc.

Back to main publications list.