Why do you look where you look?

i-eca0cf2af9fc3ac4445c7dff7d8aab70-research.gifWatch the quick video below. First you'll fixate on a small dot in the middle of the screen. Then you'll get a visual cue which serves to direct your attention to a particular location. Simultaneously, four letter Os, each colored red or green, will appear. Your task is to say, as quickly as possible, the COLOR of the letter in the direction indicated by the cue.

Now try this one, same task, but with a different cue.

Much research on visual attention during the past 30 years has focused on the difference between these two types of cues -- central arrows versus peripheral indicators such as the box you saw here. The idea was that responding to a cue in the center of your visual field was a different process from a cue placed precisely where the target object would appear. Responding to the first type of cue was voluntary, researchers argued, while the second type of cue was involuntary.

But there are many ways to direct a person's visual attention, and a new experiment by Bradley Gibson and Alan Kingstone suggests that the distinction between voluntary and involuntary orienting is oversimplifying the process. Consider this new video, with a different type of cue (again your task is to say, as quickly as possible, the color of the letter in the direction indicated by the cue):

Now how about this one?

In one sense, using a word as a cue is just like the arrow in the first movie: arguably, responding to the word is a voluntary action, just like responding to the arrow. But when Gibson and Kingstone showed similar movies to volunteers and recorded their response times, a different pattern of results emerged. They presented four different types of cues -- the words, arrows, and peripheral squares you've seen, as well as a central face which cued viewers by looking in the direction of the relevant letter. Here are their results:

i-06c9032c4e9b6f3c81922fa7d8d3c08a-gibson1.gif

Now the central / peripheral distinction doesn't appear to be as important as whether the cue is a word or a symbol. What's more, the type of word mattered, too. Viewers responded faster to the words "above" and "below" than "left" and "right." For the other cues, there was no difference between left/right or up/down response times. Gibson and Kingstone claim that these results offer support for a new theoretical understanding of how attention can be cued. Instead of classifying cues as only voluntary or involuntary, they also classify them as "projective" or "deictic." Words are projective cues because the observer must first establish a frame of reference before using the word to orient their attention. The arrows and other symbols are deictic -- they orient the viewer directly to the relevant object.

One aspect of Gibson and Kingstone's analysis doesn't quite satisfy me. Wouldn't it be simpler to argue that reading and understanding the words accounts for much, if not all of the difference in reaction time between word cues and symbolic cues? Is it possible to have a deictic word cue? Perhaps, if an experiment could be devised that compares projective and deictic uses of words, it could demonstrate whether the projective / deictic distinction is a useful one.

Gibson, B.S., & Kingstone, A. (2006). Visual attention and the semantics of space. Psychological Science, 17(7), 622-627.

More like this

I would be apt to think that response times within the word cue category would be highly variable due simply to differences in word frequency (left and right are more common than above and below). Doesn't seem like a great design, to me.

By MagicTony (not verified) on 04 Jan 2007 #permalink

The verbal processing dominates the response delay. As an illustration, imagine a red square around the target item, with written directions to please look at the item indicated by the red square. Most people would think the instructions idiotic because they looked first at the indicated item, then back at the instructions, reading them, then returning to the indicated item: they'd be critical of the poorly designed scheme.

I wouldn't knock the projective/deictic issue, especially with "left/right." I was completely stymied because I didn't know whether I should use my left/right, or the opposite. It's not as stupid as it first sounds, given that I'm used to anatomical illustrations (in which the "right" atrium is shown to the reader's left, etc). I was only able to answer at all because it didn't matter, both lateral letters being the same color.

Above/below gave me no problem, since there is no normal situation in which I'm confronted with vertically reversed images; I answered faster than I did with the small box. It just seems the frame of reference is insufficiently defined in the "left/right" question. I wonder if a "clockwise/counterclockwise" question could be designed, since that is unambiguously based on the viewer's frame of reference.

I only know experiments in which the cues are presented before the presentation of the to-be-attended targets. Thereby, the onset of the peripheral cues attracts attention involuntarily and the viewers are prepared to look at this location. In the videos presented above, all four stimuli appear at the same time as the peripheral cue, so all of them might attract attention involuntarily. I had the impression that I had to search the peripheral cue to find it. I would suspect that the peripheral cue would work much better if the cues were presented before the onset of the stimuli.

This is such a horrible study. First of all, it doesn't show anything new. I totally agree with you that all the difference is in the reading/semnatic processing. Furthermore, it totally makes sense to predict that the frequency of the central word will play a big part. I bet that if they separated left-right and above-below cases, the latter would take longer than the former. It's amazing what gets into PsychScience these days...