Millisecond Forums

I am not aware of any existing implementation. However, implementing it should be relatively straightforward.

You have an array of 8x6 = 48 possible possible positions (plus some random jitter). Each display features 12 objects (1 target, 11 distractors), i.e. 12 of those positions are used / assigned to objects. There are some additional rules (pick 24 target locations, use 12 for the "old" displays, 12 for the "new" ones).

"Configurations were generated separately for different observers" (Chun & Jiang, 1998, p. 34), which may suggests they were *hardcoded* into the experiment, not generated randomly on-the-fly at runtime.

For the "old" displays, those configurations remain the same throughout the entire task, i.e., they are re-used in each block. For the "new" displays, different configurations are used in each block. There are 24 trials per block (12 "old" configurations, 12 "new" configurations).

If you tell me what exactly you are having trouble imagining implementation-wise, I can perhaps provide some useful pointers.

Three main points come to mind:
1) How to create this grid and get the stimuli to appear in the right place
2) how to make the stimuli - the targets are T's but the distractors are L-like letters (to make them harder to distinguish from T's)
3) how to make the "old" configurations unique to each participant

Thank you for offering to help :)
 

Information from the paper:


Participants were seated in a dimly lit room in front of a 19″ CRT monitor (100 Hz, 1024 × 786 pixels resolution) at a distance of approximately 57 cm. In the visual search task, each display contained one target letter “T”—tilted 90 degrees either to the left or to the right—and 11 distractor letter “L”s—rotated randomly by 0, 90, 180, or 270° [see Figure 1(b)]. Each letter subtended 0.68° in size, and the vertical line of the L was shifted by 0.12° to the right in order to make it more similar to the target. Stimuli were placed within an invisible 8 by 6 matrix (cell size 3.40°) with positions randomly jittered by ±0.4°.

Participants performed the visual search task, which consisted of one practice and 16 experimental blocks, with 24 trials in each block (total 408 trials). The main independent variable was display type (old or new). In old displays, the same 12 randomly generated configurations were shown repeatedly across blocks. In these the target was presented in the same location within any particular configuration, and thus the target–distractor spatial layout was consistent (but not the target orientation). For new displays, 192 different configurations were randomly generated for 12 new configurations in every block. To rule out location probability effects, the target appeared equally often in each of 24 possible locations throughout the experiment (12 target locations used for old and new, respectively

1) How to create this grid and get the stimuli to appear in the right place

The grid is just a set of positions, i.e. horizontal and vertical coordinates. You either assign positions to objects or objects to positions. See e.g. https://www.millisecond.com/forums/Topic5501.aspx for a simple example.

2) how to make the stimuli - the targets are T's but the distractors are L-like letters (to make them harder to distinguish from T's)

Since the stimuli are supposed to be rotated / tilted to varying degrees, you need to use <picture> elements. The items of those pictures are the variously rotated L's (for the distractors) and T's (for the target). The example in this thread may be useful to look at in this regard: https://www.millisecond.com/forums/Topic7372.aspx

3) how to make the "old" configurations unique to each participant

Most straightforwardly, you'd generate those sets configurations beforehand, hardcode them into the script and then treat each set (one per participant) as a between-subjects condition, i.e., select one such "old" set (a different one per participant).

Thank you for sharing those helpful posts! The first one is quite similar to what I need because the example you gave had type one stimuli (targets in my case) and lots of type two stimuli (distractors in my case). However, I was wondering if you could suggest how to work out the positions for the stimuli if they are to be displayed in an invisible 8 x 6 grid = therefore 48 possible positions?

It works exactly the same way as in the example. Your invisible 8x6 grid is a set of 48 x and y position pairs. At a minimum, set up two <list> elements -- one holding the 24 x coordinates, the other holding the corresponding 24 y coordinates. Pair the lists and sample your required N positions from them, i.e., one x/y pair per object.

As for working out what exactly those 48 positions should be, the study you are basing this on only specifies visual angles. Given the physical dimensions of the display you are using / the stimuli as well as the distance of the observer from the display, you need to calculate those things: http://www.yorku.ca/eye/visangle.htm

Hope this helps.