Brain biologist Joe Tsien got famous six years ago for creating a genetically engineered ''smart mouse" with an improved memory.
But for all his success, he felt troubled: He had altered memory, but he still did not know what exactly memory was. The dictionary definition did not help. He wanted to understand what exactly brain cells did when they engraved an event for future recall.
Now, said Tsien he thinks he may. And inspiration for this latest work came in part from The Tower of Terror, the runaway elevator ride at Disney World.
In research released online yesterday in the Proceedings of the National Academy of Sciences, Tsien, of Boston University, and colleagues recorded activity in mouse brains as the animals went through experiences as startling as the plummet in the Tower of Terror.
The mice felt the sudden cold gust of air that for them could mean an owl attack; they went through the shaking of an ersatz earthquake; and they felt the cookie jar holding them drop precipitously, like a plunging elevator.
Such shocks, Tsien theorized, would produce particularly robust memories -- just as the Tower of Terror did for him. And more powerful memories could mean more pronounced brain activity.
The patterns his team managed to record from scores of electrodes in the mice's brains seem to offer the beginnings of a code-key for how the collective firing of neurons translates into memories in mice and -- eventually, perhaps -- in humans, Tsien said.
If borne out, the work could be one step toward a major goal of modern neuroscience: understanding the magic of how chemical and electrical activity in the brain is transformed into memories -- or thoughts, or feelings, or images -- in our minds.
For more than a decade, brain researchers have used techniques similar to Tsien's, recording from electrodes as animals underwent experiences and laid down memories.
But his experiments used improved tools -- recording from more neurons, up to 260 at a time, and powerful forms of data analysis, he said. They were also the first to use ''startle" memories instead of usual memories.
Such experiments are extremely tricky to perform, and must be very cautiously interpreted, warned Matthew Wilson, a professor at the Massachusetts Institute of Technology's Picower Center for Learning and Memory.
''While this may not represent a universal key to memory," he said, ''it does point out the potential for these techniques to inform us about the structure of memory."
Tsien calls the patterning he found ''collective co-spiking of neural cliques." He translates: it is not that groups of neurons are marching in lockstep like a marching band. Rather, it is like when you go to a football stadium and can't really hear anything because everyone's talking. But when a group sings a song, ''all of a sudden you can hear that message."
Those findings fit in with a growing consensus among brain researchers that large groups of neurons interact when the brain takes in and processes information, said John P. Donoghue, a neuroscience professor at Brown University.
In Tsien's experiment, the researchers reported that some neurons in the hippocampus, an area of the brain crucial to memory, responded to all three kinds of startling events. Other groups of neurons responded to only the airblast, only the cookie-jar drop, or only the earthquake. Some responded by getting more active, some by slowing down.
Their analysis of the mountains of data they gathered allowed them to make predictions, as well: just by looking at a pattern of brain activity, Tsien said, they could predict with about 95-percent accuracy which type of experience they represented. They could also tell where it had happened, the equivalent of ''in Disney World or
''You can actually directly read the neuronal activity and say, 'Aha! He just had an airpuff stimulation," Tsien said.
And one interesting phenomenon: Those recognizable patterns would recur several times at various points after the mice had been startled -- perhaps the mouse equivalent of reliving an experience.
''We think it has something to do with the fixation of newly acquired memory," Tsien said. ''After you have a roller-coaster ride, you come down and walk away with friends and can't help talking about, 'Oh my God, that was scary.' "
With the idea that the ''neural clique" is a basic coding unit, Tsien and his colleagues used sophisticated mathematical manipulations to convert the patterns of activation into a binary code of 0's and 1's. That way, they reasoned, they could compare the patterns from brain to brain -- and eventually, they hope, from species to species (though the devices implanted in the mice are too invasive to be used in human experiments).
In time, Tsien plans to examine whether other kinds of thinking and learning use similar neural codes.
Remus Osan, a Boston University co-author of yesterday, paper, described another intriguing prospect: using the brain-monitoring techniques on mice that have been genetically engineered to have unusual types of memory, to see if that yields insights into how memory works.
In the long run, he said, improved understanding of memory could lead to a variety of benefits, from help for people with diseases that affect memory, to decisions about the best age to learn math.
For now, Tsien said, one obvious lesson emerges from the inspiration he gained at Disney World: ''Go out and have fun," he said. ''That's where your best ideas come from."
Carey Goldberg is reachable at Goldberg@globe.com.