American and French scientists were able to train a linear classifier of letters and words on the activity of the inferior temporal region of the brain rhesus monkeys: this Department is accurate to 70 percent recognize the individual letters and even distinguishes real words from pseudoself. Due to this, scientists concluded that the same region of the brain of the people responsible for the recognition of letters and words, developed not exclusively for the purpose, but simply adopted to implement the required functions. Article published in the journal Nature Communications.
The advent of writing as a natural stage in the evolution of language: over time, information becomes more and more, they are complicated and the carriers no longer enough oral language to give each other the necessary information. The writing is optional and not the primary condition for the existence of the language. First, the appearance of an ancient ancestor of modern languages , date to the period 100-200 thousand years ago, and the first written language, apparently, came much later — about three thousand years BC. Secondly, speakers of many minority languages without writing in General: the transmission of information in a closed commune rarely requires the ability to read and write.
The role of letters for the language and its speakers, so there is some contradiction. On the one hand, the understanding of symbolic systems of language precedes the analysis of the phonological and semantic components (first the reader recognizes written letters, and then adds in the word and understand the meaning). On the other hand, writing, apparently, is not to be considered mandatory — if you take into account the fact that some language may not be.
To explore this contradiction is interesting and from the point of view of brain development: there is the theory that parts of visual cortex responsible for the recognition of linguistic symbols, did not develop separately, but only took over the necessary functions. According to this theory, some makings of orthographic processing (i.e. processing of letters and other written signs) must be from other primates in addition to human — and it just needs to answer the area of the brain that then reprogrammability on this skill in humans.
To test this theory experimentally decided by scientists under the leadership of James DiCarlo (DiCarlo James) from the Massachusetts Institute of technology. They focused on the lower temporal region — part of the ventral path of information processing, which accepts signals from all sections of the visual cortex responsible for the recognition of objects and images. Additionally, scientists also considered the area V4 of the visual cortex, the main source of information for inferior temporal region: this Department, however, used as a control condition.
In the experiment, scientists took part four rhesus macaque in the lower temporal region, and V4 which implanted microelectrodes for recording electrical activity of individual neurons. None of the macaques have not previously participated in similar experiments, seen the letters was for her a completely new one.
As the task for the macaques, showed the words (like true and pseudoloma) of four letters and individual letters of the English alphabet. While viewing scientists recorded the activity of individual sections 510 of the lower temporal region and 277 sites V4 region: all regions during the execution of the task poppies were active.
After this activity, scientists used to train a simple linear classifier, which activity or inferior temporal region, or V4 detect what word or letter the monkey sees in front of him. After this, the monkeys were given the job — showed them the words written in different fonts, words with the same letter or behramaj or words mixed with pseudoclosure and classifier attempted to determine whether macaque to distinguish the stimuli.
It turned out that untrained macaques, in fact, see the difference between different letter stimuli: the activity of the lower temporal region it was possible to distinguish words from pseudoself, words written in different fonts, and words that use the same letters, up to 70 percent (depending on the task and the number used for recognition sites). In V4 this task were not as effective.