07/11/2007
Capturing the Moment
I haven’t yet received my copy of Science that contains the article on which so much media attention was lavished last week. I’m talking, of course, about the work purporting to show that men talk just as much as women. While I haven’t seen the article, media accounts indicate that the study involved college students as the subjects. If this is correct, it seems to me there is much more work that needs to be done. The subjects were young. What about more mature men and women? Does marriage affect the talkativeness of either sex? As memory fades, is there more or less talking?
A few weeks ago, I attended a lecture on memory and Alzheimer’s disease. Slowly, the questions as to how Alzheimer’s shuts down memory and how the memory loss can be slowed down or prevented are being answered. On the other hand, an article by Joe Tsien in the July Scientific American looks at how memories are stored. Following up on some of this work, I visited a CalTech Web site on memory that cites a quote from Tennessee Williams’ novel “The Milk Train Doesn’t Stop Here Anymore”. “Has it ever struck you, Connie, life is all memory except for the one present moment that goes by you so quick you hardly catch it going?” This seems to me a truly profound insight by Williams. As you read this word, it would make no sense without your memory of the words that preceded it. Life makes no sense without memory.
Tsien and his group at Princeton University made headlines back in 1999 with their genetic tinkering that yielded a strain of unusually smart mice. The strain was termed Doogie, after the exceptionally bright, youthful doctor in the TV series “Doogie Howser, M.D.” The Doogie mice were quicker learners and could remember tasks longer than their normal colleagues. In his article titled “The Memory Code”, Tsien, now at Boston University, describes work on such mice to further our understanding of how memories are stored in the brain.
The trick is to monitor the activities of hundreds of neurons of mice that are awake and unfettered, a feat that has been accomplished in monkeys. The problem with mice is that their brains are much smaller, about the size of a peanut. Tsien says that previously only 20-30 neurons had been monitored simultaneously in mice. However, he and a postdoc, Longnian Lin, came up with a device allowing the recording of as many as 260 individual cells in the so-called CA1 region of the hippocampus in the mouse brain. This CA1 region is heavily involved in the formation of memory in us humans as well as in mice.
Having developed a means of recording these neurons, how to measure memory formation and what do you want the mice to remember? Well, what sort of memories stick with us for the rest of our lives? Tsien cites 9/11 as one example; those of us in the metropolitan New York area will certainly never forget the searing events of that day. Tsien apparently had a personal memorable experience falling 13 stories in Disney’s Tower of Terror. He and his group decided that the mouse equivalent of such “startling events” would be of the type that a mouse would likely remember.
Accordingly, Tsien and his team simulated (a) an earthquake by shaking a container holding the mouse, (b) an owl attack from the sky by a sudden puff of air on the mouse’s back and (c) the Tower of Terror by a short free fall in an “elevator”, which initially was simply a cookie jar. The workers subjected the mice to each of these startling events while monitoring the CA1 cells. The recording device registered the firing or lack of it in each of the 200 or more cells during the event and when the mice were at rest.
After repeated runs involving rest periods and the startling events, Tsien and another postdoc, Remus Osan, analyzed the data using software designed to recognize patterns in data. Tsien doesn’t go into details of the data analysis but does say that they could reduce all that data from the 260 or so neurons to some sort of 3-dimensional x-y-z plot. In the 3-D plot, the data cluster in 4 distinctly separate “bubbles” of activity for rest, air puff (owl), earthquake (shaking) and drop (Tower of Terror).
If they follow the patterns in time the workers can see the pattern from the rest bubble spread out to, for example, the puff bubble and then return as the mouse rests. They also see the startling event bubble patterns return on occasion in the mouse brain some time after experiencing the events. This is taken as indicating that the mouse is recalling the event. A startling event is indeed something to be remembered.
While the “bubbles” that result from the data analysis show distinct patterns for rest and the events, a more interesting thing to me was the finding that the neurons form what Tsien calls “neural cliques”. The researchers found that certain sets of neurons respond to particular characteristics. Tsien concludes that some neurons respond to the general category of a “startling event”, an event that’s unexpected and sufficiently unnerving to warrant filing it for future reference. Another clique responds to a “disturbing motion”; this clique of neurons would respond to the shaking and dropping but not to an air puff. Other cliques respond to the more specific motions of shaking and of dropping and to an air puff.
Tsien likens the building of a memory to a pyramid. Take, for example, the shaking. The base of the pyramid is the general feature (here the startling event). Building on the base is the more specific disturbing motion. On top of that is the more specific type of motion - shaking. On top of that, at the peak of the pyramid, there might be a clique that responds to location of the shaking, which for the mouse might be in a black box. Each memory is wired into the brain by the firing of the cliques associated with the various features of the event. This pyramid with a very general property at the base and becoming more and more specific as the pyramid builds to its peak is a useful concept to represent how a memory is built.
Obviously, Tsien and his team have only monitored a small number of neurons out of immense numbers in the brain, even of a mouse. There probably are other cliques that fill in other features of the startling events such as the intensity of the shaking. The number of neurons in the cliques is also up in the air. Tsien cites the work of Itzhak Fried of UCLA. Fried was studying neurons in the hippocampus of epilepsy patients and found that in one patient a particular neuron fired only in response to actress Halle Berry. We know celebrities have their groupies. Tsien suggests Halle may have a clique of neurons!
Tsien carries the concept even further by asking if the memory can be digitized. That is, can it be converted to the 1’s and 0’s of the typical computer code? Let’s digitize our mini earthquake. It is a startling event (1), a disturbing motion (1), not a drop or free fall (0), not an air puff (0) but is a shaking (1). The digital code for our mini earthquake would therefore be 11001. Tsien and his group have digitized their results from four different mice and can with almost 100% accuracy predict the event a mouse has experienced and its location from the digital code they derive from their data on the firing of the cliques of neurons.
Tsien’s final statement in the article is truly scary. If memory can be digitized, Tsien speculates that perhaps 5,000 years from now we might be able to download the contents of our minds onto computers and live forever on the network! Tennessee Williams probably never considered that type of life as a possibility! That’s more an Orwell kind of thing.
Allen F. Bortrum
|