[ad_1]
How our brains, A three-pound tissue is encased in a boney skull, and the generation of sensations from sensations is a long-standing mystery. A large amount of evidence and decades of continuous research have shown that the brain cannot simply combine sensory information to perceive the surrounding environment like a jigsaw puzzle. The brain can construct a scene based on the light that enters our eyes, even if the incoming information is noisy and fuzzy, it proves this.
Therefore, many neuroscientists have begun to regard the brain as a “prediction machine.”Through predictive processing, the brain uses its prior knowledge of the world Make inferences Or generate hypotheses about the reasons for the incoming sensory information. These assumptions—not sensory input itself—will produce perception in our minds. The more ambiguous the input, the greater the dependence on prior knowledge.
“The beauty of the predictive processing framework [is] It has a very large—sometimes critics may say too big—the ability to explain many different phenomena in many different systems,” said Floris de Lange, A neuroscientist at the Predictive Brain Laboratory at Radboud University in the Netherlands.
However, more and more neuroscience evidence supports this idea mainly indirectly, and other explanations are acceptable. “If you study human cognitive neuroscience and neuroimaging, [there’s] A lot of evidence-but super implicit, indirect evidence,” said Tim Kitzman Doctor of Radboud University, his research field is the interdisciplinary field of machine learning and neuroscience.
So the researchers are Turn to calculation model Understand and test the idea of predicting the brain. Computational neuroscientists have constructed artificial neural networks whose design is inspired by the behavior of biological neurons and can learn to predict incoming information. These models display some incredible abilities, and seem to mimic the abilities of real brains. Some experiments with these models even suggest that the brain must evolve into a predictive machine to meet energy constraints.
With the proliferation of computational models, neuroscientists who study living animals are increasingly believing that the brain learns to infer the causes of sensory inputs. Although the exact details of how the brain does this are still blurred, the extensive brushstrokes are becoming clearer.
Unconscious reasoning in perception
Predictive processing seems to be a counter-intuitive complex perception mechanism at first glance, but scientists have turned to it for a long time because it seems that other explanations are needed. Even a thousand years ago, the Muslim Arab astronomer and mathematician Hasan Ibn Al-Haytham emphasized a form of it in his work. Optical book Explain all aspects of vision. This idea gained power in the 1860s, when the German physicist and doctor Hermann von Helmholtz believed that the brain would infer the external causes of its incoming sensory inputs, rather than relying on them. Enter “bottom-up” to build its perception.
Helmholtz elaborated the concept of “unconscious reasoning” to explain bistable or multistable perception, in which images can be perceived in more than one way. For example, this happens in the well-known blurred image, which we can think of as a duck or a rabbit: our perception is constantly flipping between two animal images. In this case, Helmholtz asserted that since the image formed on the retina does not change, perception must be the result of an unconscious process of inferring the cause of the sensory data from the top down.
In the 20th century, cognitive psychologists continued to build cases where perception is an active construction process that uses bottom-up senses and top-down conceptual input. This effort culminated in an influential paper in 1980, “Perception as hypothesis,” You late Richard Langton Gregory, It believes that perceptual illusions are essentially the brain’s wrong guesses about the causes of sensory impressions. At the same time, computer vision scientists are working hard to use bottom-up reconstruction to enable computers to observe without internally “generated” models for reference.
[ad_2]
Source link
