One of the most obvious facts about memory is that it is not nearly so good as we would like. This definitely seems true in day-to-day life, and one focus of my research during the last couple years has been why our ability to remember what we see over even very short time periods is so very limited.
So memory is crap, right?
It may be hard to remember a visual scene over a very short time period, but new evidence suggests that it is remarkably easy to remember a visual image over a longer period of time (several hours).
Researchers at MIT showed participants nearly 3,000 visual images (see some of them here) over the course of 5 1/2 hours. Afterwards, their memory was tested. They were able to discriminate the pictures they actually saw from slightly altered versions of the same picture nearly 90% of the time.
This is frankly incredible. When I show participants much, much simpler images and then ask them to recognize the same images just 1 second later, accuracy is closer to 80%!
These results are going to necessitate some re-thinking the literature. It suggests that our brains are storing a lot more information than many of us thought just a little while ago. It also suggests a very strange interaction between time and memory strength that will need to be better understood.
So this is a surprise?
Yes, and no. The results are surprising, but their publication last week in the Proceedings of the National Academy of Sciences was not, at least not for me. I had the opportunity to first be stunned by these data nearly a year ago, when the third author gave a talk at Harvard. It came up again when he gave another talk during the winter. (Oh, and I’ve known the first author since we sat next to each other during a graduate school application event at MIT, and we still regularly talk about visual memory).
So I and many others have had the opportunity to think through the implications for a long time now, which means it is very possible that there are labs which have already completed follow-up studies.
While this has nothing to do with the big story itself — the sheer massiveness of visual memory elicited in this study — I bring it up as an example of my point from last week: the fact that America is (for now) the center of the scientific world gives us tremendous institutional advantages, the least of which is that it is much easier to stay fully up-to-date. If that mantle passes to another country, we will be the ones reading about old news only when it finally comes out in press.
If you yourself have done research like this, the first thing you probably wondered was where they got 3,000 carefully controlled visual images, not to mention all the test images.
Google Images, baby, Google Images. It still took a great deal of time, but as I hear the story told, the ability to download huge numbers of web images via Google was immensely helpful. This is just one more example of Web search as a tool for science.