The scope of the brain’s activities presents scientists with a seemingly impossible challenge: Each of its 100 billion nerve cells – neurons – is in contact with about 10,000 others. They communicate through a series of electrical signals that get translated, at the synapses between them, to chemical signals. To interpret these “conversations,” we would need a sort of “bar code scanner” that could read the electrical signal patterns. But even for a relatively small group of cells – just 100 – the number of possible combinations of signals is 1030 (that’s one followed by 30 zeroes) – more than all the stars in the universe.
Clearly, mapping the brain neuron by neuron is an impossible feat; thus, many scientists choose to look at the small picture – interactions between small groups of neurons. But Dr. Elad Schneidman of the Institute’s Neurobiology Department adopts the lessons from small networks and develops new methods for combining them to decipher the workings of larger ones. He takes as his starting point groups of tens or hundreds of interconnected neurons, working upward from the observed relations between pairs and triplets of cells. “The most interesting activities involve hundreds or thousands of neurons. With today’s technology, we can investigate groups of 100 to 200 – enough to allow us to explore the fundamental design principles of brain networks,” he says. And his results are providing insight that couldn’t be obtained with individual-neuron methods. In his postdoctoral research, for instance, he showed that, even though the typical correlation between pairs of neurons is weak, the combined effect of these correlations in the group is strong. “It’s like peer pressure: One opinion doesn’t carry much sway, but the joint effect of many weak signals together can control the group as a whole.”
Using advanced computational tools borrowed from computer science and physics, Schneidman analyzes brain networks, breaking them down into their components and looking for patterns that can clarify the rules and regulations of collective behavior for large groups of nerve cells. For example, in a series of experiments conducted with Dr. Ronen Segev of Ben-Gurion University of the Negev, the electrical responses of a network of nerve cells in the retina of a salamander were recorded. The scientists showed long movie clips of the salamander’s natural environment to a group of 100 of these cells and measured their reaction, obtaining a bar code readout of patterns of electrical signals. After uncovering some of the basic rules of population activity, they then demonstrated that this allows for direct “reading” of the neuron code: Schneidman and his collaborators reconstructed the instantaneous gray level of a small part of the simple movie directly from the activity patterns of the neurons – a feat that is unattainable with single-cell methods.
The collective-as-a-complex-sum-of-its-parts approach may help to explain a puzzling phenomenon: that of the brain’s ability to deal with its own internal “noise.” Not only are single nerve cells notoriously inconsistent in their reactions to the same stimulus, but communication between cells is surprisingly unreliable, as well.
Schneidman believes that nature didn’t make the effort to create precise, noise-free nerve cells because getting them to act as a group is a “cheaper,” more efficient solution, and one that has other benefits: Such redundant activity, in addition to neutralizing noise, creates a sort of “backup” that keeps the activity going even when individual brain cells die. Furthermore, such noise may facilitate the use of trial-and-error tactics for learning.
The principles Schneidman is uncovering for the group activities of nerve cells may be valid for other systems. “These mathematical rules and models cross the borders between scientific fields,” he says. “Different physical and biological phenomena – from the collective actions of atoms in a magnetic field to cooperative food searching strategies of groups of animals – share similar organizing principles.” For example, he is working on mathematical models that elucidate the group behavior of shoals of fish seeking food or flies zeroing in on a food source. These models, he hopes, may provide insight into more complex group behavior: how individuals influence each other, the role of “free will” within the group and its limits, group learning, how groups deal with communal crises and more. Schneidman hopes that uncovering principles of group behavior will help shed light on little-understood aspects of social conduct in many kinds of creatures, including those that may possibly be the hardest to pin down – human beings.
Dr. Elad Schneidman’s research is supported by the Nella and Leon Benoziyo Center for Neurological Diseases; the Clore Center for Biological Physics; the Carl and Micaela Einhorn-Dominic Brain Research Institute; the Eisenberg-Keefer Fund for New Scientists; and the Peter and Patricia Gruber Award.
From Computers to Brains
Dr. Elad Schneidman was born in Jerusalem in 1969. He studied for a B.Sc. in physics and computer science through the Hebrew University’s “Amirim” program for outstanding students before serving in the IDF as a research and development officer. He then worked for a number of high-tech companies and, in 2001, completed his doctorate in the Hebrew University’s Interdisciplinary Center for Neural Computation. After five years of postdoctoral research at Princeton University, Schneidman returned to Israel and joined the Weizmann Institute’s Neurobiology Department. He is married to Hadas Mechoulam, an ophthalmologist, and they have three children.