We can judge the emotional content of pictures in as little as 13 milliseconds (with more polls!)

Two days ago, we discussed research suggesting that the emotional qualities of images affect short-term memory for those images. When you watch a movie depicting a rapid sequence of images, you're more likely to remember images that have strong emotional content -- especially images that are arousing, whether they depict positive or negative things. Most fascinating of all: when we see images for a very short time, we're more likely to remember positive images, but when we view them for a longer period, we're more likely to remember negative ones.

But our discussion didn't address some of the possible objections to the work. It's possible, for example, that something else about emotionally-laden pictures is different: maybe they're brighter than other pictures. Maljkovic and Martini analyzed the colors in the photos they used in their study and found that there was indeed a small correlation between brightness and valence: positive pictures tended to be slightly brighter than negative pictures. What's more, brighter pictures were better remembered than darker pictures. However, there was no indication, as in the emotion data, that there was a difference in how the pictures were remembered based on how long the pictures were viewed. Brightness alone couldn't explain their results.

Another possible objection might be that the test wasn't really about memory at all. Perhaps it simply takes longer to process an image as "negative," so for short viewing times these pictures are never perceived as "negative" at all. To address this concern, Maljkovic and Martini designed a new experiment. I'll show a couple examples of their displays below.

This movie (QuickTime required) will start with a grey screen. Then an image will quickly flash, and then get replaced by two random patterns. Your job is simply to classify the the emotion depicted in the image as positive, negative, or neutral.

Movie 1:

Now try another one:

Movie 2:

The researchers showed hundreds of such movies to volunteers and had them rate them on a 9-point scale where 5 was neutral, and found that they were able to reliably rate them, even when the exposure time to the image was just 13 milliseconds (our example above flashes the image for 1/30 of a second, or 33 milliseconds). Here's a graph of their results:

So clearly we are capable of classifying images as positive or negative even after seeing them very briefly. This leaves only one plausible explanation of the results: our memory, not our perception of images, is what varies depending on the emotional content of those pictures.

You can test our demo by manually scrolling through the movies to see what was portrayed. I think there'll be little dispute that the first image is negative, while the second one is positive.

Maljkovic, V., & Martini, P. (2005). Short-term memory for scenes with affective content. Journal of Vision, 5, 215-229.