Pigeons Use AI Like Problem Solving to Accomplish Some Complex Tasks

A Bird-Brained Plan to Alter the Course of History

The year is 1943. Missiles are flying back and forth across Europe, devastating but poorly targeted. A determined-looking scientist sits down and slides an envelope across the table at the National Research Defense Committee in Washington D.C. Inside the envelope is a plan to win the war, with the help of a secret weapon: PIGEONS.

Pigeons Take Flight as Missile Navigators

The scientist is B.F. Skinner. His plan is to pilot the missiles using a handful of pigeons. These pigeons will peck the screen when the missile approaches the target. The pigeons will have an image of the target, and they have been trained to peck the screen when they see a matching image. When the pigeons agree on the target, the missile gets launched.

The Pigeon-Guided Missile Never Gets off the Ground

The US Military funds Skinner’s PROJECT PIGEON (actual name), and he spends the next year luring wild pigeons in old grain mills, and designing tiny cables to attach to their heads to steer the missiles. Despite a successful demonstration of the pigeon’s piloting abilities, the project never gets the final go-ahead.

The Alleyway to Behaviorism: Pigeon Bowling to Psychological Paradigm

However, B.F. Skinner’s long days in the old grain mills are not wasted. He uses his spare time to build tiny bowling alleys and teaches the pigeons to become quite proficient little bowlers. These experiments eventually lead to discoveries that will help B.F. Skinner establish a whole new branch of psychology: behaviorism.

Will We Ever Stop Pigeon-Holing Avian Intelligence?

Nearly a century later, the humble pigeon still has a great deal to teach us about learning and intelligence. A study led by Brandon Turner at Ohio State University tested the pigeon’s ability to solve complex problems, and once again, we seem to have underestimated these birds. The study shows that pigeons use the same principles as AI to solve problems.

Associative Learning: From Slime Molds to ChatGPT and Beyond

All intelligent forms of life (even slime molds) can learn from a process called “associative learning.” This process depends on trial and error. It's often considered a primitive and rigid approach. This attitude is beginning to shift, thanks to the remarkable success of ChatGPT and other products of machine learning that depend on “associative learning.”

Does Language Enhance Human Intelligence or Limit It?

Humans learn through associative learning too, but we also use language to make sense of the world. Language allows us to come up with rules that can be interpreted. Our use of rules, created through language, allows for complex categorization and problem-solving. Unsurprisingly, humans have assumed that language gives us an intelligent edge as a species. Sometimes, however, we over-complicate things.

Pigeons' Secret Talent for Complex Categorization

Despite the fact that pigeons are poor conversationalists, this doesn’t stop them from being very, very good at categorizing visual stimuli. You might not be surprised to hear that they can match images of trees, flowers and dogs, but how on earth do they know how to match letters of the alphabet or medical images of human tissue?

Simplicity Triumphs in Visual Problem Solving

The pigeon’s incredible visual matching skills are based on their associative learning abilities. This can give them an edge over humans when solving visual puzzles. We humans use our wordy rule-based problem-solving strategies so much that our assumptions get in the way. In other words, we overcomplicate the issue. The simplicity of the pigeon’s strategy makes it an excellent problem-solver.

Pigeons and AI's Parallel Paths to Success

Despite the incredibly complex feats performed by AI these days, the basic learning process is the same as that of the pigeon. Machine learning is based on associative learning. The machine-generated responses are rated by trainers, creating a trial-and-error scenario. The algorithm discards the responses that lead to errors, and builds on responses that are deemed correct.

Simple, Effective Learning

The Pigeon Paradox is a stark reminder that humans often fail to understand what lies outside our comfort zones. Complex associative learning does not only belong to machines. It’s happening all around us, even under our window sills and on our sidewalks.

In an interview with Ohio State News, Turner explained how pigeons learn like AI, “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 like humans would.”