The search display items (see Figure 1) appeared in white on a black background (0.53 cd/m2) and was composed of three squares with a small gap (0.15° angle at a viewing distance of 45 cm) along one side, either vertically (i.e. on the top or bottom), or laterally (on the left or right). The target was defined as the only square gaped vertically. The sizes of the three squares were: large (0.73° angle a side, 1.93 cd/m2), medium (0.57° angle, 1.59 cd/m2) and small (0.48° angle, 1.18 cd/m2). A pre-test suggested that size per se did not change the processing times of these items4. The fixation cross was made of two lines of 0.57° angle each. Display items appeared on the circumference of an imaginary circle centred at fixation. The distance between the fixation cross and each item was of 4.8° angle. In each trial, three items were presented, one of each size. The three squares formed an equilateral triangle pointing randomly and equiprobably either upward or downward5, so that all items were equidistant from each other and from fixation. Target sizes as well as target and distractor locations were distributed equally and randomly. All three experiments were conducted in a dimly lit room (0.20 cd/m2). The stimuli were displayed centrally on the colour screen of a Macintosh PowerPC G4, with 700 MHz processor.
Twelve students from the University of Lyon II (4 men, mean age: 21,8; SE.: 1.1) achieved a similar discrimination task. Stimuli and response requirements were similar to those of the Experiment 1, except that search was made unnecessary. The target was either a small or a large square and was presented alone on the screen, randomly at one of the six possible positions reported in Experiment 1. First, a fixation cross appeared for 1000 ms, followed by a 500 ms black background. Then, the search display appeared and remained on screen until the participants responded. Participants performed 96 trials, in two mixed blocks separated by a break. Each target size condition consisted of 48 trials. Mean RTs did not differ significantly between the two conditions (large: 509 ms; small: 515 ms; t(11) = 0.97; p > .17; one-tailed t-test).
On an earlier version of the manuscritpt, Dirk Wentura suggested a possible flaw of the display used in the present experiment. He feared that there could be a natural tendency to reframe two-dimensional pictures into a three-dimensional perspective, and then to process items according to their perceived proximity. Fortunately, some configurations proably suggest 3-D perspective much more strongly than others. For instance, the display with the large item on a bottom location should introduce more perspective than the display with the large square on the top location. We compared these two sets of trials using Student two-tailed t-tests and found no significant difference. Moreover, RTs were longer for large square at bottom in Experiments 1 (t(17)= 0.87, p > .39) and 3 (t(17)= 0.90, p > .38), and for large square at top in Experiment 2 (t(37)= 0.43, p > .66). Even when combining the data of the three experiments, the difference was in the unexpected direction, and failed to reach significance (large at top: 902 ms, large at bottom: 918 ms; t(74)= 0.76, p > .45). Thus, this depth hypothesis could probably not account for the present results, as suggested further by the visual inspection of the display at scale (see Figure 1).