PAGE 8 OF _ 11

Main Results*

*Error bars are standard errors of the mean.
•Reaction time is the time to correctly respond (i.e., a Hit, if the target is present, or Correct Rejection if the target is absent).
•3 number of items is the 3 late-onset cue condition (3 X's flashed to constrain the search to 3 locations)
•4 number of items is the 4 late-onset cue condition (4 X's flashed to constrain the search to 4 locations)
•"all" number of items is the 0 late-onset cue condition (all the X's flashed signaling to subjects to search all 12 locations)

Comments and observations (for target present trials)

(Note: the circle symbols are feature trials, and the square symbols are conjunction trials. The closed symbols are target-present trials, and open symbols are target-absent trials).

1. Appears to be a clear effect of number of search items on the feature search (contrast this with the Burkell and Pylyshyn's 1996 results where number of search items had no effect in the feature search).

2. The 3 and 4 late-onset cues appear to be effective in selecting a subset of items to search. If this was not the case, then the system would have searched all the items in the display, and this would have always been a conjunction search. This means that response times in the 3 and 4 late-onset cue (or number of search items) feature search-target present conditions would have been as high as the 0 late-onset (or "all" number of search items) conjunction search-target present condition -- but they are not.

3. Feature search time in the "All" number of search items condition is much lower than in the 3 and 4 late-onset cue conditions. 

4. Search time in the 4-item subset conjunction condition appears to take as long as a conjunction search in the "All" number of search items condition.

Next