In a groundbreaking study, researchers have discovered that pigeons, often regarded as simple creatures, exhibit problem-solving abilities akin to artificial intelligence (AI) systems. This revelation challenges conventional notions of avian cognition and suggests that pigeons employ a form of “brute force” learning similar to AI algorithms.
Pigeons tackle complex problems with AI-like approach
Prior research had demonstrated pigeons’ proficiency in solving intricate categorization tasks that defy conventional human thinking patterns, such as selective attention and explicit rule utilization. The prevailing hypothesis was that pigeons employed a “brute force” strategy comparable to AI models.
In this study, pigeons were subjected to a series of categorization tasks involving stimuli encompassing lines of varying widths and angles, concentric rings, and sectioned rings. The pigeons had to peck a button on the left or right to classify the stimuli into specific categories. Correct responses were rewarded with food pellets, while incorrect ones yielded no reward.
The study comprised four different tasks, each presenting varying levels of difficulty. Results revealed that, through relentless trial and error, pigeons substantially improved their ability to make correct choices. In one of the easier experiments, pigeons progressed from an initial accuracy rate of approximately 55% to an impressive 95%. Even in more challenging scenarios, their accuracy increased from 55% to 68%.
Associative learning and error correction
Researchers postulated that pigeons primarily employed associative learning—a process involving linking two phenomena together. While humans and animals often establish associations between simple concepts, such as “water” and “wet,” it was generally assumed that associative learning was too basic to explain complex visual categorization tasks. Nevertheless, the study’s findings aligned with this notion.
The research team’s AI model, developed to tackle the same tasks, utilized only two straightforward mechanisms presumed to be used by pigeons: associative learning and error correction. Much like the pigeons, the AI model learned to make increasingly accurate predictions, significantly boosting its rate of correct answers.
One notable contrast emerged between human and pigeon approaches to such tasks. Humans typically attempt to formulate rules that can simplify the task at hand, a strategy that often fails when confronted with exceptionally complex problems. Pigeons, however, abstain from rule creation, relying on trial and error coupled with associative learning. In specific types of tasks, this approach appears to grant pigeons an advantage over humans.
AI and pigeon learning principles
What makes this discovery intriguing is the similarity between the learning principles employed by pigeons and those guiding human-designed AI machines. This revelation challenges the conventional perception of pigeons as intellectually limited creatures, highlighting the shared cognitive mechanisms between pigeons and AI systems.
Dr. John Turner, a prominent researcher involved in the study, noted, “We celebrate how smart we are that we designed artificial intelligence, at the same time we disparage pigeons as dim-witted animals. But the learning principles that guide the behaviors of these AI machines are pretty similar to what pigeons use.”
The study received support from two prestigious institutions: the National Science Foundation and the National Institutes of Health. These organizations played a vital role in enabling researchers to delve into the fascinating world of pigeon cognition and its surprising parallels to AI systems.