How does one remember thee? Can one count the ways?

In "Too Hard For Science?" I interview scientists about ideas they would love to explore that they don't think could be investigated. For instance, they might involve machines beyond the realm of possibility, such as devices as big as galaxies, or they might be completely unethical, such as experimenting on children like lab rats. This feature aims to look at the impossible dreams, the seemingly intractable problems in science. However, the question mark at the end of "Too Hard For Science?" suggests that nothing might be impossible.

The scientist: Robert Stickgold, director of the Center for Sleep and Cognition at Harvard Medical School.

The idea: How many memories does a person create in one day? Assumptions regarding this number are at the foundation of many studies of the brain. "For instance, if one proposes that a function of sleep is that it helps you get rid of unnecessary memories so that you don't have too much activity building in your head, are we losing, say, 90 percent of them by the next day?" Stickgold asks. "Or if we have, say, 5 million discrete memories, knowing that might affect studies exploring how fast we access our memories."

One could put recording equipment on volunteers and compare what they experienced to what they actually remembered about events, Stickgold suggests. This is something he and his colleagues actually attempted, using cameras they hung around the necks of volunteers that took pictures every 15 seconds.

The problem: A subtle difficulty when it comes to exploring this idea has to do with how one remembers memories. "If you show people a list of 100 words, giving them five seconds to memorize each word, a task about eight minutes long, and if you ask them to recall what they saw, if they can remember 20 afterward, they're doing pretty well," Stickgold says. "However, if instead you show them a list made up of 100 words they saw plus another 100 they didn't see and ask them which they recognize, people will be about 90 percent accurate. So the memories are all there."

It's the difference between recall and recognition," Stickgold explains. "In recall, you're given a context and ask to remember the details, and in recognition, you're given the details and asked to remember the context. So if we're asking people about their memories, we have to consider if we're asking what they recall or what they recognize."

Also, just because a person does not claim to recall or recognize a memory does not mean he or she might not unconsciously react to something from a past experience without any memory of it coming to mind, a phenomenon known as priming.

Show a subject a list of 250 words, then ask her, 'What's the first word that comes into your mind that starts with 'TH,'' and she responds 'thoroughbred.' 'Was that one of the words you saw on the list?' you ask her, and she says, 'No,' but in fact it was, and there's like zero chance that she would have come up with that if she didn't 'remember' it from the list," Stickgold says. "Don't take my word for it — repeat the test with a thousand 'Intro Psych' students."

Another problem with this idea regards how one counts memories. "If someone says they have a memory about going grocery shopping, and you ask them more about it, they might tell you what they bought, and how they chose it, seeing it, putting it in the cart, going to the register — so is that one memory or five memories?" Stickgold asks. "Do I have hundreds like these from this one event in my day?" This quality of memories suggests they have a somewhat fractal nature, he suggests — the deeper a person looks into one memory, the more details one can unearth.

The solution? Scientists have spent a lot of time and effort attempting to measure recall, recognition and priming via behavioral responses, Stickgold says. However, when it comes to how one counts memories, "There probably really isn't anything in the brain that's a discrete memory," he notes.

Brain activity involves neurons wired together with connections known as synapses, each perhaps of different strengths and influenced by other synapses and circulating molecules. "Under certain circumstances, particular networks of cells are co-activated, restoring a pattern that has occurred in the past and which, in some conditions, we become conscious of," Stickgold says. "But those same neurons, even those same synapses, might be involved in hundreds of memories as parts of other, overlapping networks."

And, of course, the whole brain is just one huge vastly interconnected network, so how much you glom together a call a memory — well, it's not a question that's meaningful at the level of the brain," he adds. So maybe the solution is to redefine the question in a more meaningful way.


If you have a scientist you would like to recommend I question, or you are a scientist with an idea you think might be too hard for science, email me at