The challenge of reading writing is evolutionary too young for specific brain areas to develop. But how do we manage to recognize regularities in letter combinations and derive meaning from them? A new study shows that the basis for this is an evolutionarily old mechanism, which is based on the fact that we recognize recurring patterns and perceive them as known. It made no difference in the experiment whether it was a question of letter-like characters, geometric structures or varying grid shapes.
Reading is a demanding task for the brain: It has to recognize shapes as letters that, in certain combinations, represent specific sounds and make sense. The first human written languages only developed around 5,000 years ago. In terms of evolutionary history, however, this period is too short for our brain to be able to adapt specifically to the new challenge. In contrast to, for example, touching or smelling, there is no specially developed reading center in the brain. So apparently it uses older mechanisms.
Words made from shapes
A team led by Yamil Vidal from the International School for Advanced Studies (SISSA) in Italy has now investigated the mechanisms involved. To do this, the researchers tested their test subjects to see how well they recognize recurring patterns in letter combinations – a skill that is considered to be essential when reading. Unlike in classic studies, Vidal and colleagues used not only letter-like characters as stimuli, but also structures that have little in common with letters. The assumption behind it: “If reading is based on general visual mechanisms, some effects that occur when we are confronted with orthographic characters should also occur when we are exposed to non-orthographic stimuli,” the researchers say. “And that is exactly what this study showed.”
For the investigation, the test subjects should first familiarize themselves with short “words”, each consisting of three letter-like characters. In order to prevent the participants from being influenced by their previous knowledge, the characters were similar to writing, but had no meaning. In the next step, the participants saw known and new combinations of these pseudo letters and were asked to identify which of the words were “correct” and which were “incorrect”. “We found that participants learned to recognize words in this made-up language by how often certain parts appeared together: words that consisted of more common pairs of pseudo-letters were more easily identified,” the authors report.
They repeated the same experiment with three-dimensional objects with three arms, the ends of which were each shaped differently – analogous to the three “letters” in the first experiment. In another test, the researchers used various grid shapes that differed in terms of the distance, thickness, contrast and inclination of the grid lines. From experiment to experiment, the stimuli became more abstract and dissimilar to real letters. Nevertheless, in these experiments, too, the test subjects were able to differentiate between suitable and inappropriate stimuli.
Regularities in words and faces
“What emerged from this study,” the authors explain, “not only supports our hypothesis, but also tells us a little more about the way in which we learn. It suggests that a fundamental part of this is the recognition of statistical regularities in the visual stimuli that we perceive around us ”. We observe what surrounds us, subconsciously break it down into elements and intuitively analyze their frequency. The left fusiform gyrus, part of the cerebral cortex, is responsible for this in the brain.
Previous studies have shown that this region is active both in reading and in recognizing objects, especially faces. According to the researchers, this evolutionarily old ability is “recycled” when people become able to read. In all cases, it is crucial to recognize regularities and give them meaning. “In short, there is an adaptive attitude to stimuli that occur regularly. This finding is important not only to understand how our brain works, but also to improve artificial intelligence systems that base their learning on the same statistical principles, ”said the researchers.
Source: Yamil Vidal (International School for Advanced Studies, Italy), Current Biology, doi: 10.1016 / j.cub.2020.12.017