From dashing off an e-mail to writing War and Peace, communicating thoughts and meanings involves translating a complex idea or set of ideas into a one-dimensional string of words that another person can read in sequence and then recreate the meaning in his or her mind. Communication is at once one of the most basic and the most mysterious of our daily activities: Everyone does it; few are really good at it.
Is there a basic, underlying structure to the effective communication of complex ideas? A team of scientists that included physicists and language researchers at the Weizmann Institute of Science and elsewhere investigated this question by applying scientific methods to some of our culture’s most successful models for the skillful transfer of ideas: classic writings that, by common agreement, get their messages across well. In research published in the Proceedings of the National Academy of Sciences (PNAS), the scientists created mathematical tools that allowed them to trace the development of ideas throughout a book. The international team included Prof. Elisha Moses and postdoctoral fellow Dr. Enrique Alvarez Lacalle of the Weizmann Institute’s Physics of Complex Systems Department as well as Prof. Jean-Pierre Eckmann, a frequent visitor from the University of Geneva, and research student Beate Dorow from the University of Stuttgart.
Because strings of words are one-dimensional, they literally lack depth. Yet our minds and memories are able to recreate complex ideas from this string. Moses and his team hypothesized that this process may rely on hierarchical structures “encoded” in the narrative. (An obvious hierarchical structure in a text is chapter-paragraph-sentence.) The implication is that our minds decipher not only the individual words, but the encoded structure, enabling us to comprehend abstract concepts.
The scientists applied their mathematical tools to a number of books known for their ability to convey ideas. They included the writings of Albert Einstein, Mark Twain’s Tom Sawyer, Metamorphosis by Franz Kafka and other classics of different styles and periods, to see if it was possible to identify common structures. They defined “windows of attention” of around 200 words (about a paragraph), and within these windows, they identified pairs of words that frequently occurred near each other (after eliminating “meaningless” words such as pronouns). From the resulting word lists and the frequency with which the single words appeared in the text, the scientists’ mathematical analysis was used to construct a sort of network of “concept vectors” – linked words that convey the principal ideas of the text.
Mathematically, these concept vectors can go in many directions. When we read a text, we’re taking a tour along the paths that make up the resulting network. The multidimensional concept vectors seem to span a whole “web” of ideas. The scientists’ findings suggest that this network is based on a tree-like hierarchy, and that such a hierarchy may be a basic underpinning of all language. The reader or listener can reconstruct the hierarchical structure of a text and so enter the multidimensional space of ideas. Thus, from a flat page and a one-dimensional string of words, we are able to grasp the full complexity of “the author’s meaning.”
Moses: “Philosophers from Wittgenstein to Chomsky have taught us that language plays a central evolutionary role in shaping the human brain, and that revealing the structure of language is an essential step to comprehending brain structure. Our contribution to research in this basic field is in using mathematical tools to connect concepts or ideas with the words used to express them. The structure serves to transmit concepts and reconstruct them in the mind of the reader. A deep question that remains open is whether the correlations we uncovered are related to making a text aesthetic as well as comprehensible.”
Prof. Elisha Moses’s research is supported by the Clore Center for Biological Physics; the Center for Experimental Physics; and the Rosa and Emilio Segre Research Award.
The Physics of Finger Tapping
The volunteer sits with earphones on his head and an electrode taped to his finger, tapping to a beat. Suddenly, his forefinger swings out of rhythm. The researcher standing behind him, postdoctoral fellow Dr. Nestor Handzy, has aimed a painless, external magnetic pulse at one part of his brain, which causes the finger to twitch involuntarily. For that split second, explains Handzy, the finger’s actions were controlled by two opposing instructions. This sort of competition, says Prof. Moses, lends itself to treatment with concepts from physics.
Moses and Handzy have teamed up with Dr. Avi Peled of the Technion and Shaar Menashe Mental Health Center, a psychiatrist who believes that physics can help find better treatments for schizophrenia. In Peled’s words: “The brain is an extremely complex, non-linear network of neurons, and schizophrenia is probably a disorder of connectivity.” Moses: “As physicists, we framed the question and set up the experiment using a complex systems approach.”
With the simple finger-tapping test, the team has already found that schizophrenics’ brains show an atypical response pattern. What’s more, they’ve seen evidence that magnetic pulses might be able to straighten out skewed lines of communication between brain areas in schizophrenia patients. The scientists’ dream is to eventually create a “brain pacemaker” to “reset” these faulty communication signals.