The Birds of Thought and Memory
CIN member Andreas Nieder and his team investigate the cognitive abilities of birds of the Corvus genus (crows, ravens, rooks, and jackdaws). These birds’ behaviour shows capabilities in diverse tasks of memory and problem-solving second to none in the animal kingdom, except for those of the primates (monkeys, apes and humans). The researchers study in the brain structures underlying the birds’ performance.
Despite the common misconception, birds’ brains are very advanced, having evolved to perform necessary and complex functions with a minimum of weight. Recent studies have shown that birds’ brains actually contain as many or more neurons than similarly highly evolved primates’ brains (1). However, they have evolved on a radically different path – birds are the descendants of the reptilic dinosaurs – and often show neuronal representation much different from that in the endbrains of mammals.
By studying corvid brain functions and comparing them to those of primates, Nieder and his team therefore aim to arrive at more universal models for the neural basis of behavioural functions: memory, strategic planning, and abstraction, among others.
The genus Corvus have certainly earned their place among the most symbol-laden animals in human culture. The obvious association for these carrion-eating birds has often been with death and warfare. But over the millennia, humans have noted another quality common to all members of the Corvus genus to an even larger degree, and made it a part of myth: their brains. In Native American mythology, Raven is revered as the god or spirit who created the world, or as a Promethean trickster figure who brought fire and light, and thus, tool use and the spark of intelligence. To the Australian Aborigines, Crow is a similar figure. In Norse mythology, it is the twin ravens Hugin and Munin (whose names quite literally mean „thought“ and „memory“), who provide Odin with news and intelligence of all that happens in the world. In many cultural contexts, ravens or crows bring news or otherwise act as messengers.
This mythologisation of ravens and crows has its origin in the astounding intelligence that these birds show in their behaviour. From abstract association, to numerical processing, to memory, to strategic planning, corvids seem equipped to think their way out of a large number of advanced problems, leaving the average Collie, Shepherd or Golden Retriever outmatched.
Nevertheless, not much about corvid brains has been known, owing to the fact that birds lack a neocortex. This most recently evolved structure of the mammalian brain forms the basis for primate intelligence and has therefore – understandably – been the focus of cognitive neuroscience for a long time. Birds, with their brain structures evolving over many more millions of years than mammals, have developed very different neural ‚solutions’ for similar ‚problems’, though.
Andreas Nieder’s research group was the first internationally to dig into the specifically avian representation of what is often subsumed under the heading of ‚higher brain functions’.
In a first study published in Nature Communications in 2013 (2), Nieder and his student Lena Veit looked into the neuronal substrate of rule-based decision making via single-unit recording in behaving carrion crows (Corvus corone). They trained crows to carry out memory tests on a computer. The crows were shown an image and had to remember it. Shortly afterwards, they had to select one of two test images on a touchscreen with their beaks based on a switching behavioral rules. One of the test images was identical to the first image, the other different. Sometimes the rule of the game was to select the same image, and sometimes it was to select the different one. The crows were able to carry out both tasks and to switch between them as appropriate. That demonstrates a high level of concentration and mental flexibility which few animal species can manage – and which is an effort even for humans.
The crows were quickly able to carry out these tasks even when given new sets of images. The researchers observed neuronal activity in the nidopallium caudolaterale, a brain region associated with the highest levels of cognition in birds. One group of nerve cells responded exclusively when the crows had to choose the same image – while another group of cells always responded when they were operating on the ‚different image’ rule. By observing this cell activity, the researchers were often able to predict which rule the crow was following even before it made its choice.
Interestingly, even though the brains of birds and mammals have different anatomies, the ‚solutions’ employed for decision-making is quite similar on a cellular level. They represent a general principle which has re-emerged throughout the history of evolution. There are many examples of body parts in unrelated species evolving on different paths to fulfil analogous functions, e.g. the fins of whales, sea tortoises, and penguins. This is called convergent evolution.
Veit & Nieder’s results are evidence that the evolution of decision-making neurons in birds’ nidopallium caudolaterale and mammals’ prefrontal cortex may be another example of this phenomenon. For an early treatment of this idea in more depth, one may read the interesting review article by Emery and Clayton published in Science in 2004 (3).
In another study published in 2015 (4), Nieder and another of his students, Helen Ditz, investigated numerical ability. The surprising results: in corvids, counting – an activity quite different from decision-making – likewise seems to employ neural mechanisms similar to those found in primates. Ditz and Nieder trained crows to discriminate groups of dots.
During performance, the team recorded the responses of individual neurons in an integrative area of the crow endbrain. This area also receives inputs from the visual system. The neurons Ditz and Nieder investigated are apparently tuned specifically for numerosity: they ignore the dots’ size, shape and arrangement and only extract their number. What is more, each cell’s response peaks at its respective preferred number.
This study provides valuable insights into the biological roots of counting capabilities, corroborating an old legend told among hunters: when three hunters went into a blind situated near a field where crows often came to feed, they waited in vain. The crows refused to move into shooting range. One hunter left the blind, but the crows would not appear. When the second hunter left the blind, the crows still would not budge. Only when the third hunter finally left, the crows realised that the coast was clear and resumed their normal feeding activity. Apparently, when a crow looks at three dots, grains or hunters, single neurons recognize the groups’ ‚threeness’.
This discovery shows that the ability to deal with abstract numerical concepts can be traced back to individual nerve cells in corvids.
Even in tasks of learning and abstraction, crows not only show similar abilities to primates, but also perform these tasks using an analogous neural substrate. For a study published by Veit, Pidpruzhnykova and Nieder in 2015 (5), crows learned to sort images by associating them with one of two abstract shapes: The crow was shown an image, and after a slight delay, was presented with a red triangle and a blue cross. Tapping the correct shape with its beak resulted in the crow being rewarded. Within just a few iterations, the two crows in the experiment learned the process that was required of them. After only minutes, they were able to successfully sort the images that were already known to them. Then they were shown images that they were seeing for the first time, and had to learn the correct associations for these, as well.
At the same time, the researchers measured the birds’ brain activity – namely, the activity of more than 300 neurons in the nidopallium caudolaterale – to see what was happening during the learning process. Individual neurons responded to different pictures in different ways. The activity of some cells sorted the images according to the answer required: One of the cells responded strongly to all the ‚blue’-associated images, while another responded to the ‚red’ ones – both regardless of the images’ differing motifs.
The researchers found the quick learning observed in the crows’ behaviour mirrored in the neural activity: while the crows were guessing, many cells barely respond to an unfamiliar image. But after a few tries, as soon as the bird learned the correct association, the cells showed the right answer for the same image. Instead of putting the whole image in what one might call the ‚working memory’, the crows’ brain simply stored the image’s correct association – red triangle or blue cross – there. This kind of storage in the working memory makes sense. The birds don’t have to remember as many details, and they are prepared for the correct answer straight away.
Again, this kind of processing has only been seen in primates – prior to Veit, Pidpruzhnykova and Nieder’s study, that is. The much differently structured endbrains of birds and mammals employ very similar learning strategies.
Small differences in the learning process remain, however. They lead to the next big question that is now under investigation at Nieder’s lab: What does the different structure of the brain mean for the interaction of different brain regions during the learning process? Much work yet remains to be done to truly understand these birds of thought and memory.
- Olkowicz S., Kocourek M., Lučan R. K., Porteš M., Fitch W. T., Herculano-Houzel S., Němec P. (2016): Birds Have Primate-Like Numbers of Neurons in the Forebrain. PNAS 113(26): pp. 7255–7260.
- Veit L., Nieder A. (2013): Abstract Rule Neurons in the Endbrain Support Intelligent Behavior in Corvid Songbirds. Nature Communications 4: 2878.
- Emery N. J., Clayton N. S. (2004): The Mentality of Crows: Convergent Evolution of Intelligence in Corvids and Apes. Science 306(5703): pp. 1903–1907.
- Ditz H. M., Nieder A. (2015): Neurons Selective to the Number of Visual Items in the Corvid Songbird Endbrain. PNAS 112(25): pp. 7827–7832.
- Veit L., Pidpruzhnykova G., Nieder A. (2015): Associative Learning Rapidly Establishes Neuronal Representations of Upcoming Behavioral Choices in Crows. PNAS 112(49): pp. 15208–15213.