Amazing demo of what we remember visually, and why

i-eca0cf2af9fc3ac4445c7dff7d8aab70-research.gifWhat types of images are you more likely to remember over the short-term? Pleasant? Bright? Arousing? Disturbing? One method of testing short-term visual memory is to flash a rapid series of pictures, then ask viewers which ones they saw. The pictures are shown so quickly that it's impossible to remember them all. Click on the image below to view a movie (Quicktime required). Wait for the entire movie to load, then watch it. You'll see 12 photos rapidly flash by. Play it twice, but no more.


Other than the first and the last one, did you remember any? I'll put a quick test below (fair warning -- there's a bit of nudity in the test images).

[make sure you play the movie before you scroll down!]

Picture 1:

Picture 2:

Picture 3:

Picture 4:

So which pictures really were in the movie? I'll give the answers below. The larger question, of course, is which type type of pictures viewers are more likely to remember, and why. Vera Maljkovic and Paolo Martini have designed a much more systematic test.

They showed viewers hundreds of pictures in rapid serial visual presentation (yes, that's abbreviated RSVP) just as I did above, in sets of 10. The time each image was presented varied from about 1/100 of a second to four full seconds. Of course, the most important factor in memory for a picture was how long it had been viewed, but does the type of picture matter too?

You might guess that viewers are more likely to remember a nude woman than a minivan, but what about a firing gun, or a corpse? Each of these pictures offers viewers a different emotional experience, and previous research on emotion has consistently found that this experience varies in two different ways. Photos (and other stimuli) can be more or less arousing, and they have a positive or negative valence. These dimensions of emotion are independent, as this chart shows:


Photos can be positive and arousing (sex), or negative and arousing (death). Moderately arousing positive pictures include children and flowers, while moderately arousing negative pictures include smokestacks and graveyards. Innocuous objects like irons and file cabinets aren't arousing at all, and offer neither a positive or negative valence.

So, which pictures are remembered best? As it turns out, it depends both on the picture type and how long the picture was viewed. When just arousal is considered, the data maps out how you might expect:


More arousing pictures are remembered better, no matter how long the pictures were viewed. But when valence is subjected to a similar analysis, a different picture emerges:


For short viewing times, positive photos are remembered better, but for longer viewing times, negative photos are remembered better.

So just as arousal and valence have different effects on emotional experience, they also have different, independent effects on visual memory. For short viewing times we remember positive, arousing images best, but for longer viewing times we remember negative, arousing images best.

Was our demonstration movie able to replicate this effect? The last three images tested were all present in our movie. Maljkovic and Martini's results would suggest that the reclining nude Venus would be remembered best (positive, highly arousing), followed by the firing gun (negative, highly arousing), and the minivan (neutral, not arousing). How did your results compare?

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

More like this

Very strange... I take it there's something sadly, sadly wrong with me as the only picture I remembered seeing was the minivan.

By Anonymous (not verified) on 06 Mar 2007 #permalink

I also only remembered the minivan. And I incorrectly said the first one was also in the movie.


Either we've got one sexy, intimidating minivan or something else is going on here! But reclining Venus is still the early leader.

I remembered the minivan and the cupboard :/

Ian / Cristi:

Some of the distractor images were displayed for longer periods of time. The man's face was shown for the longest. The bridge and book shelf ("cupboard") were displayed twice as long as the images we tested.

me too the minivan, does it suggest some kind of trouble as it is not parallel to the road?

By Massimiliano Di Luca (not verified) on 06 Mar 2007 #permalink

strange, even after watching the sequence over and over, I can BARELY see the gun. I guess the test simply cannot be reproduced online because of processor speed and so on...


Good point; I hadn't thought of that (actually it's backing out of our snow-covered driveway).


It might reflect something about your display. The movie is not compressed, but if your monitor doesn't have a high enough refresh rate, then it could be missed. But it's actually being displayed for just as long as the other photos.

I think the issue is more likely due to the fact that it's the second frame of the series. We didn't have the opportunity to randomize where the photos appeared.

Actually, I went through the video more carefully and it is quite uneven. The images you chose to be the subject or question were only on the screen for one frame, whereas other pictures were on the screen for 3 or more frames. So, images that you didn't query us about had a longer duration on the screen (an artifact that I also noticed while watching the video the first and 2nd times).

It might be a result of a video editing procedure or a result of how it came through on the web, but that would seem to skew your results pretty far as our minds would focus on more lasting images that can be interpreted accurately over images that are fleeting.

Just a thought.

Actually, I went through the video more carefully and it is quite uneven. The images you chose to be the subject or question were only on the screen for one frame, whereas other pictures were on the screen for 3 or more frames. So, images that you didn't query us about had a longer duration on the screen (an artifact that I also noticed while watching the video the first and 2nd times).

Actually, that was intentional, and replicated the procedure Maljkovic and Martini used in their study (except they included an even wider variation of display times). But all the test images were displayed for the same amount of time. I think the fact that the minivan was second to last and the gun was second from first has an important effect on the result.

I did remember only the guy in blue (perhaps because he is identical to my father, and because of the long showing time), and the parking lot at the end. Answered "no" to all the pictures on the poll.

I think there's some unintentional variation in the frame rate affecting the results. I watched it twice and saw different images after the first few (which were slower).

I'm guessing from your results that the appearance of the gun image very very early in the sequence has probably had a large effect on the results. This could be a technology refresh rate or whatever issue or it might be a result of not being prepared visually for such a barrage.

It is kinda of startling to see all that wipe by so fast even if when I know it would. I think it took a half a second to recover from this and even attempt to remember any of it.

I went back to the movie and it took me a few tries to even see it. After that I was able to see it fairly easily which was probably a result of learned recognition of the actual images and learned ability to deal with the image barrage.

It would be interesting to do this test with a few initial tries first using different images. This way people would be used to the barrage and you'd overcome that initial startle reaction.

I participated but my choice of player has voided my results. Using VLC to play the movie, I see different images each time I play the video. I don't mean to say I notice different images, the player is actually omitting certain of the frames and in particular is not opening the display window until a point after which the video has begun. Watching in Quicktime Player, the movie is the same every time and I can actually see all the frames. I said no to all the questions, fyi.

By false_cause (not verified) on 06 Mar 2007 #permalink

The one image I really remembered was a reclining man and his arms stretched up -- he was dressed in bright blue.

Surely color plays a vital role in memory recall?

I remember the blue man and the bookshelf. I caught the snow part of the minivan photo, but really did not notice the vehicle. Certain colors have always caught my attention, though, including that pleasant sort of bright blue.

I saw the bear, a man in red with dark hair with his smiling mouth open, the minivan, and the buffalo at the end. The whole scenario played out so fast, I wasn't sure what I was seeing all "mashed" together! Interesting way to test for results!

I saw only the bear, the smiling man (in blue?), the empty (a negative) bookcase, the bookstore and the buffalo.

Even on the third viewing.

The first couple times I played the video, there were only 3 pictures that showed up: the bear, the gun, and the buffalo. I'm not sure if this has affected other people, but it might skew the results a bit.

Great article by the way

The only one I really remember is the man. It was easy to remember his blue coloured shirt and the fact that he was smiling. The smiling part in fact was the most useful part for me to remember.

I saw the man and the minivan and the bookshelf. I wonder if the sex (and orientation) of the participants matters -- maybe if you'd had a naked man in there, I would have remembered ;)

I also only saw the minivan. I think the order in which the pictures are shown is likely to be relevant. I assume everyone sees the same movie? I think it has something to do with noting the color shifts. I noted the blue man's shirt, the stark white minivan and the empty/grain wood tone.

I don't recommend watching it multiple times-- h e a d a c h e. Perhaps it's just me procrastinating from finishing a midterm essay exam. What? Doesn't everyone take memory tests at 2:30 a.m.? Funny how Midterm Week and National Sleep Awareness Week overlap...

That was great. I actually bought a book from Libreria Fahrenheit 451 on the Campo di Fiori while I was in Rome last summer. Great bookshop to check out if you are ever there.

I also only saw the cupboard and the minivan, and the first and last pics stood out the most, I don't remember the other two tested photos, so doc what does that say about me?

I scrolled back and forth through the video, it appears that some pictures were displayed much longer than others which would explain why so many of us remembered the cupboard, the van, and they guy in the blue shirt. The Naked Venus was only displayed for a tiny fraction of what other images were.

My browser displayed the opposite - the gun and the reclining venus, and ocasionally some of the others. I think browsers aren't suitable for this kind of test at the moment - shame.

I think browsers aren't suitable for this kind of test at the moment - shame.

I'm not sure I agree. Some people's computers aren't quite up to the task, but we've actually come fairly close to replicating the study's results. I think the minivan problem is more likely related to the fact that it's near the end of the movie.

If you look at the data from the study, very few people remember the images, and we've matched that quite well, including an impressive "positive arousing" effect.

Now what's interesting to ask -- does the mental state of observers prior to viewing the reel influence which pictures stay in the memory? I bet to some extent it does. E.g., people in a bad mood might remember the picture of the gun over sweet Venus, or maybe Venus over the gun...

By Bobby Hill (not verified) on 09 Mar 2007 #permalink

Besides the first and last picture I only remembered the smiling man in the blue shirt, and the empty bookshelf