Reading in the Brain

by Stanislaus Dehaene

This book is loaded with information but is poorly written. I found it so turgid that I had to take it in small chunks, and therefore it has taken me two months to plow through the thing. As a consequence, I have difficulty remembering specifics from the book. I shall therefore concentrate on those points that I found striking.

The first was the discussion of how letters are decoded. Mr. Dehaene observes that there appear to be three basic visual elements used by the brain to characterize any shape: a straight line segment (“I”), an elbow (“L”), and a conjunction of three lines (“Y”). These basic shapes appear to be burned into the brain, and it uses them as the basic components that it recognizes while reading. This constitutes the starting point of the mass of processes that recognize a word.

I was fascinated by the many images obtained by the new process of magnetoencephalography, which permits high-resolution movies of the brain while a person is thinking. Figure 2.17, for example, shows frames from two movies of a reading brain. Recognition of a spoken word begins in a small patch of the brain, showing up within 55 milliseconds of the completion of its articulation. From there, brain activity spreads through the side of the brain, encompassing about a third of the brain surface area within 420 milliseconds. Recognition of a spoken work, however, is carried out very differently at first. The activity begins in the occipital lobe, where vision is processed, and then spreads along the lower surface of the brain. Within 420 milliseconds, about half of the active area overlaps with the areas used for recognizing spoken words.

Another striking discovery was the fact that there is a single point in the brain, only a few millimeters in diameter, that seems to act as the vertex for reading. Mr. Dehaene calls it the “letterbox area”, and it is always active during reading, irrespective of the language being read. In other words, this area is universally used by all people for reading comprehension (although much more of the brain is involved, this one area is always active during reading.)

Mr. Dehaene regales the reader with countless examples of experiments into reading, many of which are extremely clever in design. For example, languages such as Italian have a direct relationship between spelling and pronunciation, and Italian readers show more activity in parts of the brain that differ from those used by English speakers.

Chinese writing presents us with a particularly interesting case: it is not alphabetic and uses a complex system that combines pure morphemic symbols (one symbol for one word) with markers that differentiate one use of the basic symbol from another. This in turn results in a completely different approach in Chinese readers. The reading process is more probabilistic than deductive; the reader must take context into account. I was also charmed by a long tale written in Chinese about a poet eating lions in a stone room. The story is about ten sentences long, and its written expression looks no different from other bits of Chinese writing. However, the entire poem is pronounced as a sequence of a single syllable: “shi”. There are actually two different pronunciations of “shi” using two different tones, so the poem, read aloud, might sound something like this: “SHI shi shi shi SHI SHI shi SHI SHI shi shi shi SHI...” In effect, every word in the tale rhymes with every other!

By the way, context-dependent interpretation is certainly not limited to Chinese: consider these newspaper headlines in English: “Red Tape Holds Up New Bridge”; “Two Convicts Evade Noose, Jury Hung”; “Milk Drinkers are Turning to Powder”.

There are lots of these fascinating details buried in the text, but you have to suffer through some weak writing to figure it out. The writing isn’t bad; the grammar and spelling are perfect and the prose does not suffer from use of an academic style. Still, the sentences sometimes don’t quite make sense, and have to be re-read to figure out the meaning.