Australasian Science Magazine

This pigeon, named Einstein, correctly pecks the lower number (8) of the pair 8 vs 9. When the pigeon pecks the image correctly a border is displayed and a short tone sounds.

By Damian Scarf

Research into the intellectual abilities of pigeons reveals that the brains of birds, while very different to our own brains, are capable of much more than they’re given credit for.

Many people think pigeons are not the sharpest crayons in the box, and I must admit that I shared this view at the beginning of my PhD. But I was wrong.

My research has shown that pigeons are intelligent and, remarkably, can handle numbers in sophisticated and abstract ways, just like humans and other primates. My findings add to a growing body of work showing that you do not need to have hands or a primate-like brain to be intelligent, and add to the evidence showing that there are a number of evolutionary paths that lead to intelligence.

I conducted my PhD under the supervision of Prof Michael Colombo, where pigeons were the standard experimental subjects in the lab. Given their bird-brain reputation, I did everything I could to avoid using pigeons. In fact, the very first experiment I conducted set out to show that chickens, due to the pecking orders they form, were more intelligent than pigeons.

To my surprise the chickens were no better than the pigeons. If anything it appeared that the pigeons had an edge in the intelligence game.

Still not happy with the idea of using pigeons, I turned to magpies. While the magpies proved impossible to train on the touch screen tasks I was using, the pigeons took to pecking the touch screens like a duck takes to water.

Being a scientist, I weighed up the empirical evidence and came to the conclusion that pigeons were the ideal experimental subjects for my research. Since then I have been constantly surprised by the many things these “bird-brained” animals are able to do.

The question I am most commonly asked is why I spend my time investigating pigeon intelligence. I have two answers to this question. First, finding out what pigeons are able to do, and thus the limits of their brains, is critical to developing a deeper understanding of the human brain. Indeed, like all animals, pigeons and humans share a common ancestor, and the abilities we identify in pigeons help to illuminate the evolutionary origins and precursors of our own intelligence.

Second, testing the limits of pigeon intelligence also allows us to answer questions about the structures and neural machinery required for complex behaviour. While the bird brain has a very different structure to our primate brain, functionally it is difficult to tell them apart. This tells us that the structures that characterise the primate brain, like the layered prefrontal cortex, are not a prerequisite for intelligent behaviour.

A recent example from my research will help to illustrate these points and will also highlight the impressive, and surprising, capabilities of the humble pigeon. The example I will use is from the domain of numerical intelligence.

Obviously, the numerical abilities of humans are unparalleled in the animal kingdom, and range from counting to solving complex mathematical formulas. One hallmark of our numerical intelligence is our ability to learn and employ abstract numerical rules. For example, once we have acquired the rules of long division (e.g. carry the remainder) we can apply those rules to novel problems with new numbers.

Can we find precursors to our abstract numerical abilities in non-human animals? To borrow a line from Bob the Builder’s theme song, “Yes we can!”

In a seminal study, Elizabeth Brannon and Herbert Terrace trained two rhesus monkeys to order images that contained up to four objects of various colours, sizes and shapes. The four images were presented on a touch screen, and the monkeys were rewarded with a banana pellet if they pressed the images in ascending order (i.e. press the image with one object, then the image with two objects and so on). The monkeys were trained on many of these lists and were then tested to see if they had acquired an abstract numerical rule or a simple nominal rule. If the monkeys learnt an abstract numerical rule (i.e. respond to the images from smallest to largest) they should be able to correctly order pairs of new numbers. In contrast, if the monkeys had learnt a nominal rule, they should have difficulty with numbers outside of the training range (i.e. numbers up to 4). Amazingly, when presented with pairs of images containing numbers of objects the monkeys had never seen before (e.g. 5 vs 9; 7 vs 8 etc.), the monkeys were able to correctly order them.

The fact that monkeys can learn abstract numerical rules suggests that this ability is not unique to humans. The obvious question that follows is whether it is unique to primates.

To answer this question we trained our pigeons just like the rhesus monkeys described above. Rather than training our pigeons to order up to four objects, we trained them to order images containing up to three objects. Just like the monkeys, we trained our pigeons on many lists and then presented them with pairs of images containing up to nine objects.

Impressively, the pigeons’ performance was almost identical to that of the monkeys. Indeed, on the pairs containing two completely novel numbers (e.g. 5 vs 6; 7 vs 9; 6 vs 8 etc.), the monkeys selected the lower number on 74.0% of the trials while the pigeons’ performance was almost identical at 73.6%.

In addition to their impressive performance with the novel number pairs, the pigeons displayed two response functions that characterise numerical ordering in humans. One of these is the symbolic distance effect – as the distance between two numbers decreases, performance also decreases. For example, the pigeons performed better on pairs with a distance of eight (i.e., 1 vs 9) than they did on pairs with a distance of seven (i.e., 1 vs 8, 2 vs 9), and better on pairs with a distance of seven than they did on pairs with a distance of six (i.e. 1 vs 7, 2 vs 8, 3 vs 9).

The pigeon’s accuracy was also constrained by Weber’s law, which describes the fact that pairs with the same distance are not created equally. For example, while the pairs 1 vs 2 and 8 vs 9 have a difference of one, they have drastically different ratios (calculated by dividing the smallest number by the larger one) of 0.50 and 0.89, respectively. Just like the monkeys, the pigeons’ accuracy decreased as the ratio approached one.

Together, the distance and ratio effects suggest that pigeons represent number in a similar way to humans.

What does this finding tell us? First, it adds another piece to the puzzle on the evolution of numerical intelligence, and suggests that these shared numerical abilities are either a homologous trait or the product of convergent evolution. If it is a homologous trait it means that the ability was present in the last common ancestor of primates and birds some 300 million years ago. Conversely, if it is the product of convergent evolution it suggests that birds and primates developed the ability independently, perhaps due to similar selective pressures related to foraging (e.g. being able to pick between two locations with different amounts of food).

Finally, showing that pigeons can acquire abstract numerical rules demonstrates that the ability is not unique to primates, or mammals, and that the bird brain, although structurally very different to the primate brain, contains all the neural machinery required for numerical intelligence.

My next project is to test two bird species – keas and African grey parrots – which many people believe are among the smartest of all birds. This project will provide us with information about how widespread numerical intelligence is among birds and will also highlight any differences that exist between bird species.

I am also looking to uncover how homing pigeons find their way home after being released hundreds of kilometres away from their loft. The pigeons’ amazing navigational abilities are perhaps the only ability that pigeons have over primates, including humans, and it will be interesting to unlock just how these bird brains accomplish such an amazing feat.