The humble pigeon. They’re noisy, steal your food, and leave a mess everywhere — usually all over your car if you’ve just had it cleaned. And let’s face it, they don’t seem too smart. Just look into those dull eyes, right?
… Yeah, you’d be forgiven for thinking that nobody’s home.
So, who do you think would be smarter, a pigeon or a cutting-edge AI? The answer might sound obvious — after all, nobody’s asking pigeons to optimize their code, unlike with ChatGPT (or if anyone is, they’re not admitting it).
So perhaps it might come as a bit of a surprise that some psychologists at the University of Iowa recently put pigeons to the test, and found out these birds use the same basic thinking process as advanced AI technologies.
For science! Pigeon vs Machine
… Okay, so they didn’t throw a battle-bot in a ring with a bird (pretty sure there’d be ethics concerns there) but they did put pigeons through complex categorization tests. In these tests, high-level thinking like logic or reasoning wouldn’t help, so the pigeons had to rely on exhaustive trial and error.
What they found was the pigeons were able to memorize enough scenarios to reach nearly 70% accuracy. And, much like AIs, they didn’t get bored with the process — unlike humans, who would score poorly and likely give up.
The real takeaway, though, was that pigeons were using the same technique as advanced AI models — associated learning — which makes machines seemingly “think” like humans.
A quick primer: Associated Learning vs Declarative Learning
Humans engage in two types of mechanisms to solve category learning tasks: declarative learning and associative learning. If you don’t know the difference, here’s a breakdown:
- Declarative learning is about exercising reason based on a set of rules or strategies. This is a “higher level” way of learning. It’s believed a select few animals — like dolphins or chimpanzees — have this.
- Associative learning is about making connections between objects, like “sky-blue” and “water-wet”. This is considered a “lower level” way of thinking. Many animal species are believed to use this.
As humans, sometimes we think too much
In the test, pigeons were shown a stimulus and had to peck a button on the left or right to say which one it belonged to. If they got it right, they got a tasty pellet. If they got it wrong, they got nothing (I’d say other than shame, but pigeons are shameless).
To make sure pigeons couldn’t use declarative learning, the test was arbitrary: no rules or logic would help them pass it. The pigeons started by guessing correctly half the time, but after hundreds of tests, got up to an average of 68%.
"The goal was to see to what extent a simple associative mechanism was capable of solving a task that would trouble us, because people rely so heavily on rules or strategies," said Ed Wasserman, Stuit Professor of Experimental Psychology at Iowa University and the study's corresponding author.
"In this case, those rules would get in the way of learning. The pigeon never goes through that process. It doesn't have that high-level thinking process. But it doesn't get in the way of their learning. In fact, in some ways it facilitates it."
The “Associative Learning Paradox”: Bird-brained AI get too much credit
Angry Birds: University of Iowa edition
Researchers say the distinction between declarative and associative learning is important, because when the media talks about AI being “incredibly smart”, the same status is not conferred to animals, where the same flexible thinking is downplayed or ignored.
There’s even a name for it: the Associative Learning Paradox.
"You hear all the time about the wonders of AI, all the amazing things that it can do," says Ed Wasserman. "It can beat the pants off people playing chess, or at any video game, for that matter."
“How does it do it? Is it smart? No, it's using the same system or an equivalent system to what the pigeon is using here…. Yet, when people talk about associative learning in humans and animals, it is discounted as rigid and unsophisticated."
Implications for category learning
Researchers aren't pointing out this divide just to give pigeons their due, or suggesting we should replace ChatGPT with thousands of trained birds, like having monkeys type out the complete works of Shakespeare.
Because associative learning is dismissed for being too simple to power the impressive cognitive achievements of humans and non-human species, it gets very little attention in the category learning space. However, it’s clear that associatively-driven AI has matched (or blown past) human performance in some areas using this technique.
To solve this, some researchers have called for the fields of AI, animal learning, and animal cognition to be integrated. This may affect our understanding of intelligence as a whole as we seek to emulate it.
Given that they are now making winged robots that can land like birds, maybe we’ll be seeing a lot more cross-over in the future.
“I am Robo-Pigeon of the Bird Collective. Your bread crumbs, as they have been, are over. Resistance is futile. From this point onwards, you will feed us.”
Learn more about AI and ML with Pluralsight
Interested in your own learning journey, and don’t want to leave it for the birds? Test out your machine learning literacy with a free SkillIQ test, or check out these AI/ML classes and learning paths — start at the level you’re comfortable with, whether you’re a beginner or expert.
5 keys to successful organizational design
How do you create an organization that is nimble, flexible and takes a fresh view of team structure? These are the keys to creating and maintaining a successful business that will last the test of time.Read more
8 ways to stand out in your stand-up meetings
Whether you call them stand-ups, scrums, or morning circles, here's some secrets to standing out and helping everyone get the most out of them.Read more
Technology in 2025: Prepare your workforce
The key to surviving this new industrial revolution is leading it. That requires two key elements of agile businesses: awareness of disruptive technology and a plan to develop talent that can make the most of it.Read more