There are two experiments that you can participate in online, here at cogsci.nl. Both are cognitive psychology classics. Try it, it's kinda fun (for a little while)!

I just analyzed the data from the feature integration experiment, based on the classic study by Treisman and Gelade. In this experiment, you have to indicate whether a red “T” is present in the display. (Btw, you can still participate in the experiment.)

In the “color popout” condition the red “T” is presented among green “T”-s. In the “shape popout” condition, it is presented among red “X”-es. These conditions are relative easy, because the target differs from the other letters (the distractor stimuli) on a single feature. That is, the target is either the only “T” (easy) or the only red object (even easier). According to Treisman and Gelade, you do not need to search the display for the target in these trials—the red "T" just pops out! Therefore, it doesn't really matter how many distractors there are. In the graph, you can see that, indeed, the “search slopes” for the popout conditions are relative flat: People do not respond much slower or make many more errors, if the red “T” is presented among many green “T”-s (large display size), relative to only a few green “T”-s (and analogously for the shape popout).

The situation is radically different for the conjunction condition. In the conjunction condition, the red “T” was presented among green “T”-s, green “X”-s and red “X”-s. In other words, the target was only unique on a combination of features, namely in being both red and a “T”. According to Treisman and Gelade, attention “glues” the colors and the shapes together. Because this is a painstaking process, which sort of happens one object at a time, there is a very steep search slope for the conjunction condition. Indeed, as you can see from the graph, the number of distractors (display size) has a huge impact on the time it takes you to find the target in the conjunction condition.

I think it's quite nice that this classic result is so easily replicated in an online experiment!

P.S. You may notice that I conveniently collapsed across target present and absent trials, which is another story in itself.

References