American psychologists from Ohio State University discovered similarities between ordinary pigeons and artificial intelligence (AI) models. It turns out that both birds and neural networks rely on similar principles to solve problems. The research was published in a scientific journal iScience.
Previous research has shown that pigeons learn to cope with complex tasks for which human thought is not useful, with selective attention and the ability to follow certain patterns.
In the current study, pigeons were shown a stimulus that could consist of lines of different widths and angles, as well as concentric and cross-sectional rings. The birds had to peck the button on the right or left to indicate which category a particular shape belonged to. If the birds answered correctly, they received a food pellet.
The experiment involved four different tasks of increasing difficulty. The results showed that through trial and error, the pigeons increased their ability to make the correct choice from about 55% to 95% in one of the simplest experiments. Even in the more difficult scenario, the rate of correct answers increased from 55% to 68%.
“We found really convincing evidence that the mechanisms governing pigeon learning are quite similar to the principles underlying modern machine learning and artificial intelligence techniques. Our results show that nature has found a way to make the pigeon an incredibly efficient learner who lacks the ability to generalize or predict like humans do.” shows that it can happen,” said lead author Brandon Turner.
Biologists before I learnedIt turned out that kingfishers can dive into water without damaging their brains due to genetic mutations.