Attentional blink is a fascinating phenomenon that occurs when we see images, words, or numbers presented in a rapid sequence. As images flash by at about one every tenth of a second, you're asked to look for two in particular. If you were looking for numbers in a sequence of letters, the sequence might be SDLX3DJ9WVNBDR. The number 3 would be easy to spot, but 9, which follows 3/10 of a second later, is spotted much less frequently.
The effect works for images as well. You might be asked to look for flowers in a sequence of furniture pictures. Again, flowers that follow between 2/10 and 4/10 of a second after the first flower image are spotted much less frequently. But there's an interesting exception to the effect: we seem to be immune to it when looking for faces. This short video offers two examples, with a bit of a twist. You'll see a random stream of pictures of office equipment, flashing by one every tenth of a second. Embedded in each stream are two pictures: First, a fruit, and then either a face or a watch. You'll be instructed whether to look for a face or a watch, and what to notice about it, before each stream.
People tend to miss the watch more frequently than the faces. The effect can be difficult to replicate online because of the differences in computers, and because you're only seeing one pair of movies instead of hundreds of them, but it's been reliably duplicated in several studies.
Ayelet Landau and Shlomo Bentin showed student volunteers 900 such movies, using over a hundred different faces and watches. Instead of office supplies, the other images in the movies were pieces of furniture. In every movie, respondents were first asked to look for a flower (a sunflower or tulip), and then a face or a watch (they saw 450 "face" movies and then 450 "watch" movies, or vice versa). The faces/watches were separated from the flowers by 1/10, 3/10, or 7/10 of a second. Half the time a face or watch was present, and half the time it was not. First the students had to say what type of flower they saw, and then whether or not they saw the face/watch. This second question was the key: how accurate were the students at detecting the faces and watches when they were present in the videos? Here are the results:
The students were nearly perfect at detecting faces, with almost 100 percent accuracy no matter how long the faces appeared after the flower. But watches showed a classic attentional blink, with significantly lower accuracy 3/10 of a second after seeing the flower picture. So why do faces seem to be immune from the attentional blink?
One possibility is that faces are processed completely separately from all other visual images -- that there's a completely different perceptual process for handling faces. But if that's true, then we should never see a similar sort of immunity for non-face objects. Landau and Bentin created a new set of movies using furniture in the background as before, but this time the students would be looking for cars and watches, with no faces or flowers involved at all. The students were divided into two groups. Group 1 saw cars near the beginning of the movie and had to determine whether the cars were facing to the left or the right. Group 2 saw watches near the beginning and again made the analog-digital judgment. Both groups saw blocks of movies with cars and watches in the second position, following the first image by either 3/10 or 7/10 of a second, and as before were only required to say whether the second car or watch actually appeared in the movie. Here are the results:
So when the second object is from the same category as the first object, attentional blink is significantly worse. This suggests that there may be separate processing going on for various different types of objects, not just faces. But there was still an attentional blink even when a watch followed a car, or vice versa, so this situation is different from the total immunity that appears to exist for faces.
In another experiment, the researchers showed that we can even process details of faces (the way the eyes are looking) better during the attentional blink than details of other objects (orientation of hands of watches). So while there does appear to be something special about faces, faces may not be processed completely differently from other images. Landau and Bentin say that the only special thing about faces is their perceptual salience -- because we're good at recognizing faces, faces skirt some, but not all, of the limitations of the attentional blink.
Landau, A., & Bentin, S. (2008). Attentional and perceptual factors affecting the attentional blink for faces and objects. Journal of Experimental Psychology: Human Perception and Performance, 34 (4), 818-830 DOI: 10.1037/0096-15188.8.131.528
"People tend to miss the face more frequently than the watches."
The rest of the text seems to indicate this statement is backwards.
"The students were nearly perfect at detecting faces, with almost 100 percent accuracy no matter how long the faces appeared after the flower."
The quicktime video worked great for me - much easier to notice the face than the watch.
Jay: You're right. I've fixed it now -- thanks!
So is attentional blink due to conflict due to processing the first item just when the second item pops up? And faces use a different pathway during that period?
"So when the second object is different from the first object, attentional blink is significantly worse."
Your graphs and the following text would lead one to the opposite conclusion - you probably mean the other way around, right?
Eamon, thanks for the correction. I fixed it in the article. Together, we'll get this thing right!
But watches have faces too. I think what these experiments show is things with wristbands magnify the blink effect. ;-)
I only noticed the watch the second time, but the face the first time. But the second time I made myself stare :/