Study shows pigeons match wits with artificial intelligence
Washington, US: A new study showed that pigeons approach some challenges in the same way that artificial intelligence would, allowing them to perform challenging tasks that humans would struggle with.
Previous studies demonstrated that pigeons learned how to tackle complicated categorization problems that human ways of thinking, such as selective attention and explicit rule application, would be ineffective in addressing.
According to Brandon Turner, the main author of the new study and professor of psychology at Ohio State University, researchers hypothesised that pigeons utilised a "brute force" style of problem-solving similar to what is used in AI models.
However, this study may have confirmed it, Turner and a colleague evaluated a rudimentary AI model to see if it could solve issues in the same way that pigeons did - and it did.
"We found really strong evidence that the mechanisms guiding pigeon learning are remarkably similar to the same principles that guide modern machine learning and AI techniques," Turner said.
"Our findings suggest that in the pigeon, nature may have found a way to make an incredibly efficient learner that has no ability to generalize or extrapolate as humans would."
Turner conducted the study with Edward Wasserman, a professor of psychology at the University of Iowa. Their results were published in the journal iScience.
In the study, pigeons were shown a stimulus, which could include lines of various widths and angles, concentric rings and sectioned rings. They had to click a button on the right or left to indicate to which category it belonged. If they got it correct, they received a food pellet - if they were wrong, they received nothing.
There were four different tasks in the study, some harder than the others. Results showed that, through trial and error, the pigeons improved their ability to make the correct choices in one of the easier experiments from about 55 per cent to 95 per cent of the time. Even in a more difficult scenario, their correct responses improved from 55 per cent to 68 per cent.
Researchers believed the pigeons used what is called associative learning, which is linking two phenomena with each other. For example, it is easy to understand the link between "water" and "wet." People teach their dogs to link sitting when they are commanded to receive a treat.
"Associative learning is frequently presumed to be far too primitive and rigid to explain complex visual categorization like what we saw the pigeons do," Turner said.
The researchers' AI model tackled the same tasks using just the two simple mechanisms that pigeons were presumed to use: associative learning and error correction. And, like the pigeons, the AI model learned to make the right predictions to significantly increase the number of correct answers.
For humans, the challenge when given tasks like those given to pigeons is that they would try to come up with a rule or rules that could make the task easier.
"But in this case, there were no rules that could help make this any easier. That really frustrates humans and they often give up on tasks like this," he said.
"Pigeons don't try to make rules. They just use this brute force way of trial and error and associative learning and in some specific types of tasks that helps them perform better than humans."
"We celebrate how smart we are that we designed artificial intelligence, at the same time we disparage pigeons as dim-witted animals," he said.
"But the learning principles that guide the behaviours of these AI machines are pretty similar to what pigeons use."
