Both methods, which involve making connections between objects or patterns, allow pigeons and artificial intelligence to excel at certain tasks, challenging the belief that this method is a rigid and unsophisticated form of learning.
In a new study, psychologists at the University of Iowa investigated the functioning of the pigeon’s brain and how the bird’s learning is similar to artificial intelligence. The researchers gave the pigeons complex tests by classification that high-level thinking, such as the use of logic or reasoning, does not help to solve. Instead, due to a lot of trial and error, the pigeons were eventually able to remember enough scenarios in the test to reach nearly 70 percent accuracy.
Researchers equate the pigeons’ iterative and trial-and-error approach with artificial intelligence. And they also believe that computers use the same basic method and are taught how to recognize patterns and objects that are easily recognized by humans. Of course, computers are becoming more powerful due to their enormous memory and storage power and the ever-increasing growth in these areas, and will go far beyond anything a pigeon’s brain can comprehend.
However, the basic process of associative learning, considered a low-level thinking technique, is the same between test pigeons and the latest advances in artificial intelligence.
The researchers sought to discover two types of learning: one, declarative learning, which is based on applying reason based on a set of rules or strategies, and is a so-called higher level of learning that is more attributed to individuals. Another is associative learning, which focuses on recognizing and making connections between objects or patterns, such as “sky blue” and “wet water”.
Many animal species use associative learning, but only a few, such as dolphins and chimpanzees, are thought to be capable of declarative learning.
However, AI has expanded to include computers, robots, surveillance systems, and many other technologies that apparently think like humans.
Each experimental pigeon was shown a stimulus and had to decide which category the stimulus belonged to by pecking a button on the right or left side. The categories included line width, line angle, concentric rings and cut rings. A correct answer earned a tasty seed, and an incorrect answer earned nothing. What makes the test so difficult, Wasserman says, is that it’s arbitrary: There are no rules or logic to help decipher the task.
Wasserman also says: Pigeons are like masters of artificial intelligence. They use a biological algorithm, one given to them by nature, while computers use an artificial algorithm given to them by man.
The common denominator is that AI and pigeons both use associative learning, yet this basic level thinking is what allows pigeons to ultimately succeed. If people take the same test, they score poorly and are more likely to give up, Wasserman says.
The study, titled Resolving the paradox of associative learning based on category learning in pigeons, was published online February 7 in the journal Current Biology.
Leave a Reply