An experiment somewhat similar to the one Pat describes has been done (a
Master's thesis at Florida State University by Doug Joseph). Doug did not
use a partial report procedure used by Sperling. Instead, he looked at
accuracy of picture recognition as a function of number and durations of
exposures to pictures using exposures with tachistoscopic durations. (This
work is available only as the unpublished thesis.) I don't recall how
exactly how he measured recognition. As I recall his lab set-up I'm
inclined to say that he probably used a yes/no recognition test (he had only
one projector, so he could only present one picture at a time, even at
test). I worked with the same professor after Doug had left and we used
two-alternative forced-choice recognition tests in which the foils consisted
of pictures of the same scene (they differed from the old pictures in
detailed ways that defied verbal description). So the recognition of these
pictures was probably mediated by fine-grained details (in various
locations) that participants managed to encode during these brief exposures
(in this sense, we might argue that the test assesses how much information
was registered just as Sperling's unpredictable cues assessed how many of
the letters in the matrix were registered). One of the subjective qualities
associated with tachistoscopic viewing of pictures of real-world scenes (vs.
displays of letters) is that the scenes have a coherence and continuity that
matrices of letters don't have. Perhaps the pattern recognition process
involved in the extraction of individual letters from an array of random
letters places limits on the amount and apparent spatial extent of
information that is registered (being focused on relatively small spatial
locations and operating sequentially on the letters in the array). Pehaps
there are more global processes that contribute to visual processing of
pictures (or even words?) but are useless when we are dealing with a random
matrix of letters.
>Pat Cabe writes on 22 Jan 99,:
>
>I have always thought that it was a large inferential jump from being able
>to report some letters in a row to saying that the sensory register
>contains a veridical picture of the visual sensation. I have wondered
>why someone hasn't taken this further and put up some kind of complex
>photograph for a time and then remove the picture and turn on a shape
>on the screen about 250 msec after the picture goes off and see if
>people can report what was contained within the shape. If this requires
>too much processing for sensory memory, how about just reporting the
>color of that part of the picture. I know that Phillips in 1974 reported
>some work using, according to Reed (1996), "patterns made by
>randomly filled cells in a square matrix" (p. 22) but Reed doesn't report
>the size of the matrix. I am guessing that there were quite a few cells in
>the matrix but I would like to know how many.
>
>Maybe my complex picture experiment has been done (or has not been
>done for good reason) and I am not aware of it. If so, I would like to
>know what happened.
>
___________________________________________________
Claudia J. Stanny, Ph.D. e-mail: [EMAIL PROTECTED]
Department of Psychology Phone: (850) 474 - 3163
University of West Florida FAX: (850) 857 - 6060
Pensacola, FL 32514 - 5751
