Bird Brainpower: Investigating the Remarkable Minds of Crows
They watch us. They learn about us and our habits. We are a big part of the environmental conditions to which many of them have adapted.
They’re like us. They hang around in groups. Individuals have different personalities. Pairs bond together for years at a time, maybe lifetimes, and they take good care of their kids. They’re loud, opportunistic, mischievous, and messy. And they’re smart.
Meet Crows
Members of the genus Corvus—crows—include birds with “crow” in their common names as well as ravens, rooks, and jackdaws. There are 45–50 named species of Corvus at the moment (the naming of species is a dynamic field), though that range will change and increase as more information from populations in unstudied areas becomes available. They are medium- to large-sized birds with big heads relative to body size and, usually, large to massive bills. They live all over the world except South America and Antarctica, in the varied habitats that exist across continents and on islands, from southern to northern high latitudes.
American Crows live in populations having different social organizations and dynamics in different regions of North America. The members of populations that breed in the north migrate in spring and fall of each year, with known one-way travel distances of up to 1,740 miles (2,800 km). Each year, they spend months commuting, and they live in one place during spring and summer and another in fall and winter. During migration, they spend nights at the same giant communal roosts along their routes, much like humans returning to the same campgrounds on annual road trips.
Ravens are the largest crows and occur all around the northern hemisphere. Most don’t breed until they’re 4 years old, and some not until they’re 10. Once they become breeders, ravens tend to live in pairs. In the years between leaving natal areas and breeding, individuals join other nonbreeders in small groups and larger flocks as they jointly acquire the skills needed for successful breeding, including being able to reliably find food (carcasses that are unpredictable in when and where they’ll become available, across huge landscapes).
New Caledonian Crows occur on only two of New Caledonia’s islands. Since they inhabit primary forests, observing them in the field is challenging. They are not very social, although kids tend to stick close to parents for extended periods, up to 2 years. New Caledonian Crows are the only nonhuman animals known to make tools from materials with which they have no experience1 and the only ones known to make cumulative improvements to tool design over time.2
I have had the opportunity to observe and study many crows myself, and to learn about the behavior and cognitive capabilities of other species through experimental research and fieldwork published by other scientists. There is a tendency to want to compare the results of studies of cognition in nonhuman animals to humans. How do they measure up? How about compared to apes? Such comparisons are already not straightforward when comparing other mammals, and even less so when comparing crows. They are very different types of organisms that live in three-dimensional worlds, without hands, and with brains, eyes and ears that are different from those of mammals. And yet in test after test, crows perform equal to or even better than apes, and are on par with human children or occasionally even exceed adult human capabilities!
American Crows
I began studying American Crows in the early 1980s, on a golf course in Encino, CA. For purposes of my research, I needed to be able to tell individuals apart, and so I had to catch them, to be able to mark them. They were very difficult to capture! With the use of traps and nets3, 4 and climbing to nests to temporarily obtain and mark late-stage nestlings,5 I got a bunch marked and was able to peer into their worlds. They are one of the most civilized species of which I am aware.
The crows I studied in California were year-round residents that nested colonially (that is, having lots of nests in the same general area) and defended only the small areas of space immediately surrounding their nests—if they defended any space at all. Neighbors often foraged together, members of breeding groups were regularly observed in others’ core areas, and breeders rarely prevented others from entering their nest tree or landing on or near their nests. Most individuals did not breed until they were at least 3 or 4 years old, and many nonbreeders remained in natal areas associating with parents or joined the resident non-breeding flock.
I had one of my favorite fieldwork experiences, ever, on that golf course: Because population members had come to associate me with things that caused them distress (e.g., climbing to their nests), they transferred that to other situations and would yell at me, when I arrived, after something bad had happened. One day I drove around on golf cart paths looking for the cause of their yelling, and on the ground found a female with an injured wing. She could not fly but she could run, and the crows dive-bombed and yelled at me as I chased her down. I had her examined by a vet and taped up, and she spent 8 weeks in a cage in my bedroom as her wing healed. In the field, her mate and 1-year old daughter continued to care for the four nestlings in her nest. Three weeks after I took her to my place, a strong storm blew her nest out of its tree and all of the nestlings died. Her mate and daughter hung around for another two weeks, but then were not seen very often. After two months her bone had healed, but her flight muscles had atrophied. I moved her to a flight cage and put her through regular daily exercises. Finally, eleven weeks after her removal, I brought her to the golf course.
Ravens form alliances, reconcile conflicts, and console distressed partners.
I wafted her into her nest tree and threw a bunch of peanuts on the ground. Crows began to fly to the peanuts, and she joined them. Almost immediately, I saw her mate headed right for her from across a busy 4-lane road. He landed beside her and both of them bowed low to each other and produced a slow, melodic, low frequency vocalization that I had never heard before. The pair then proceeded to walk around the group of peanut eating crows and stopped to bow and vocalize to each other three more times. I was crying. The pair was reunited.
The crows I studied in Oklahoma were year-round residents in small-to-large territories that were only sometimes defended against neighbor intrusion. Most delayed breeding until at least 3–4 years old, and many remained “at home” with parents until they bred. Many also left home and moved in with other groups within the population before becoming breeders. Individuals had friends in groups other than their own, and some that had moved out of the population returned occasionally to natal territories and spent time with their parents. Some visited their siblings in other groups, and some moved in with their siblings’ families. Several males established territories adjacent to their parents, and extended families of at least three generations would spend time together.6
One day, I sat in my car watching a group in a residential backyard. One of the crows walked along a wooden fence railing to the end post and attempted to get at something in the interior of the hole supporting the railing. Unsuccessful with its bill, it pecked at the wood surrounding the hole and loosened a section at the top, pulling on it until a triangular piece of wood broke off. The crow placed the piece of wood under its feet, with the wide end closest to its body, and hammered several times at the tapered end. It then picked up the piece of wood by the wide end and probed the hole with the tapered end for about 20 seconds. Another crow in the group called from some distance away, and the toolmaker placed the probe into the hole and took off. I went to the hole, saw only the remains of a spider’s web, and retrieved the probe. It did not match the gap from which it had been pulled—the tapered end had clearly been narrowed.7 A few days later when I approached the post, a large spider dashed out of the hole.
Also in Oklahoma, I watched the mother of the nestlings in the nest to which my co-worker was headed hammer repeatedly at a branch of a nearby pine tree. At first, I thought she was exhibiting displacement behavior but then a pinecone at her feet loosened (she had been hammering at its connection), and she carried it to above my co-worker and dropped it right on his head! She repeated this behavior three more times, hitting him on 3 out of the 4 tries.
So that I could observe crows behaving naturally when I wasn’t trying to capture them or get to nestlings, I donned a disguise on the latter occasions. Years later, it was officially demonstrated that American Crows can remember “dangerous” human faces for at least 2.7 years,8 and they can even learn whom to worry about from others!9
Ravens
Ravens are scavengers and regularly store (cache) away surplus food obtained at carcasses, and they rely on their caches for sustenance. They are not known to use tools much in the wild.
Caching behavior has been the focus of many studies10, 11 and ravens are skilled strategists. If they know another raven is watching them, they will go to a location out of the observer’s view before caching. Cachers behave differently in the presence of competitors who have or have not seen the caching event: if they have been watched while storing food, cachers move their caches when knowledgeable competitors get close.
Competitors behave differently depending on the situation. If they know where the cache is but the raven they’re paired with doesn’t know that they know, they run right over and retrieve it. If they’re paired with the cacher and the cacher knows they’re wise to the location, they act as if they don’t know and dawdle and fiddle around, seemingly hoping to take advantage of any lapse in focus by the cacher. This level of understanding of what others know rivals that demonstrated in chimpanzees.12
When given an opportunity to pay attention to another cacher while caching, ravens performed better than humans when asked to retrieve both their own and the other cacher’s caches.13 And when paired with a partner who kept taking advantage of the situation, a raven employed a human-like solution: deception. Ravens were trained to find and retrieve food hidden in a maze of film canisters and one raven was better at it than a dominant male. At some point, the dominant raven quit playing and would just wait for the other one to choose a canister and begin to open the lid, then fly over and steal the food. The raven “being taken advantage of” then changed tactics and initially went to go to a canister it knew to be empty and pretended to try to open it. When the dominant bird flew over and was distracted for a few seconds expecting to get the food, the other flew to a canister it knew to be filled, and got the food!14
Ravens successfully solved the problem of obtaining meat dangling from a branch by pulling up sections of the string and stepping on them to keep them from falling back down. And then they were successful with the non-intuitive task of pulling down on the string to bring the meat up.15 They did as well as apes in tests of choosing appropriate tools (despite not being tools used in the wild) and they did better than orangutans, chimps, and bonobos when asked to choose the correct currency for bartering for food.16 Ravens were able to select the right tool in environments different from where they learned to use it; even in the face of 17-hour (overnight) delays between having to select the appropriate tool and being able to use it, providing evidence of their forward-planning abilities. They did better than 4-year-old children in the first-trial performances at the tool- and currency-choice experiments,17 and they are perceptive enough to follow the gaze of a human to a location out of view and hop over to see what’s up.18
Ravens form alliances, reconcile conflicts, and console distressed partners.19, 20, 21 They remember former group members and their relationships with them after long (years) periods of separation.22 When disappointed or frustrated, for example by being offered less preferred food, they respond in a way that other ravens observing them can identify, and the observers themselves are then negatively affected.23 In measures of value, compatibility, and security, the quality of raven social relationships was said to be analogous to those of chimpanzees.24
New Caledonian Crows
In 1996, a paper published in Nature changed everything: New Caledonian Crows were manufacturing tools, in the wild, at a level of complexity not ever seen among nonhuman animals before.25 To extract prey from burrows and natural crevices, they make hooks and probes from twigs and pieces cut from leaf, some of which require sophisticated manipulation and modification skills. No other nonhuman animals do anything like it.26, 27
Betty was the name of a New Caledonian Crow caught in the wild and taken (with several others) to the University of Oxford for testing.28, 29 She was partnered in a cage with a male given the name Able. In an early test, Betty and Able were allowed into a room with a table that had a clear plastic vertical tube, secured in a plastic pan, containing a basket-shaped container of meat at the bottom. There were two wires on the table; one had already been bent so there was a hook at one end. The researchers wondered if one of the crows would use the hook to grab the basket handle, and Betty at first picked it up, but Able took it from her and flew away with it. Betty wanted the meat. She picked up the straight wire (in her bill) and inserted it into the tube but, of course, it was useless in its straight form. And so with force, she jammed the wire into a corner of the pan several times and bent it into a hook! She then used the hook to get the basket. Her behavior made clear that she had a mental representation in her mind of the problem and the solution, and therefore of the instrument she needed to make.
Crows perform equal to or even better than apes, and are on par with human children or occasionally even exceed adult human capabilities.
New Caledonian Crows were able to spontaneously solve a “metatool” task (using a short tool to obtain a longer one needed for food extraction),30 and they were able to keep in mind the out-of-sight tools they had available (and where they stored them), while performing sequences of tool tasks, providing strong evidence that they can plan several moves ahead.31
From field and lab studies of tool behavior, scientists have also learned that New Caledonian Crows:
- Choose sticks of appropriate length to match the distance to food.32
- Choose sticks of thickness that track the diameter of holes.33
- Choose the right tool for a task that will only occur in the future.34
- Unbundle a preferred stick rather than use a nonpreferred one.35
Such selectivity suggests these crows have representations of the situations in their minds and so can select the appropriate tools. They also tend to keep their preferred tools safe, under their feet and in holes.36
Individual New Caledonian Crows tend to be lateralized in their tool use (i.e., right- or left-billed): they usually hold probe tools in their bills, with the nonworking ends pressed against one side of their heads37 and individuals prefer one side over the other for different tasks.38 Lateralization is thought to be associated with complex tasks and mental demands (i.e., as tasks increase in difficulty, “control areas” in brains become specialized/localized),39, 40 suggesting that, as in humans, species-level lateralization is an adaptation for efficient neural processing of complex tasks.41
Evidence from more than 5,500 tools suggests that narrow and stepped variations were likely improvements on the wide-tool design.
Tools made by New Caledonian Crows from Pandanus leaves come in three types, all with a barbed edge: unstepped narrow and wider probes, and “stepped” tools,42 the latter being made through a sequenced process involving a series of distinct snips and tears along the barbed edge of a leaf to produce a probe that increasingly narrows toward the tip (the “steps”) and has barbs along one edge.43 Evidence from more than 5,500 tools suggests that narrow and stepped variations were likely improvements on the wide-tool design.44
And so, these crows have evolved minds powerful enough to develop and improve upon tool design, something thought possible only by humans (technological progress is thought to be one of our hallmark characteristics). That there is geographic variation in tool manufacture and the innovations are passed from generation to generation45, 46, 47 suggests there may be cultural mechanisms at work.
Experiments with captive-bred, hand-raised New Caledonian Crows have demonstrated a strong genetic component to tool interest, manufacture, and use—young crows start playing with twigs, leaves, and crevices on their own, suggesting the phenomenon is an evolved adaptation.
More Crows
In preparation for studies of cognition and neurophysiology, crows have been trained to monitor screens in experimental setups and respond to visual and auditory signals so that they can be trained to do all kinds of other things. Carrion Crows, for example, have been trained to identify complex pictures despite distractions,48, 49 express their understanding of the concept of greater than/less than50 and to respond to the switching between “same” and “different” rules provided both visually and auditorily.51 They’ve been trained to discriminate quantity, ranging between 1–30 dots on a screen52, 53 and they have been trained to peck different numbers of times and to indicate “I’m done” when they’re finished with their answer.54, 55 That these birds can understand the training protocols is almost as impressive as the results of the studies!
Jackdaws performed on par with apes in a test of self-control over motor impulses and did better than bonobos and gorillas despite brains that are 70–94 times smaller.56 Unlike chimpanzees, and similarly to observations about ravens described earlier, jackdaws respond to human gaze and nonverbal cues like pointing.57
Crows play, have friends, and mourn the death of friends and family members.
Hooded Crows have been shown to be capable of analytical reasoning. In tests called “match-to-sample,” where subjects are presented with paired stimuli that are the same or different (e.g., in size or shape) and then asked to match the concepts of “same” or “different” to brand new items, crows spontaneously perceived and understood the relationships without any specific training in categories of size, shape, and color.58 Such analytical thinking is thought to be foundational for “categorization, creative problem solving, and scientific discovery,” and was thought to be uniquely human.59
Carrion Crows were able to learn the Arabic numerals 1–4 and then produce matching numbers of vocalizations (e.g., “caw caw caw” for 3) when prompted by the visual image or an auditory cue.60 The modality of the cue did not affect their performance, indicating that their vocal production was guided by an abstract numerical concept. Evidence also indicates that the crows were planning the total number of vocalizations before they started vocalizing and that when errors were made—too few or too many—the crows had started out correctly but “lost track” along the way.
Carrion Crows have also been shown to be capable of recursion: the cognitive ability to process paired elements embedded within a larger sequence.61 For example, a “center-embedded” sequence would appear [{}] and is analogous to “the crow the experimenter chose passed the test,” with {} corresponding to “the experimenter chose.” An ability to use recursion might potentially, possibly infinitely, expand the range of ideas and concepts that can be communicated. Carrion Crows outperformed macaques and performed on par with human children in tests of recursive abilities; another characteristic thought to be unique to humans.
Rooks are not known to use tools in the wild, but they figured out that by plugging specific holes in the floor of an aviary (including tapping in the plugs!), they could create pools of water in which individuals could drink and bathe.62 Rooks also learned to get food in a trap-tube task (inserting a probe into one end of a tube with holes in it, in order to push out a food reward) and transferred what they learned to a new task on their first try,63 rivaling the physical intelligence of chimpanzees.64 One rook transferred concepts to two additional tasks, indicating she understood the physical aspects of the challenges (including gravity) and was able to “abstract rules” and form mental representations.65
In another set of experiments,66 rooks pushed stones into tubes to collapse a platform to obtain a worm and immediately transferred the concept and picked up stones to drop them in tubes. They chose the correctly sized stones when tube diameters were changed; when no stones were provided, they left the testing room to go outside to collect stones before returning to the testing apparatus! When conditions changed, they immediately used (provided) sticks in lieu of stones; heavy sticks were dropped in and light ones were shoved, suggesting goal-directed thinking. They solved a metatool task, were able to modify branches into functional tools, understood how a hook functioned and used one to retrieve a basket of food at the bottom of a tube, and bent straight wires into hooks, thereby rivaling the abilities of tool-using New Caledonian Crows. All of these findings provide evidence for insight being involved in the problem-solving abilities of rooks.
Final Thoughts
The “marshmallow test” is one of the most well-known and compelling demonstrations of the human ability to delay gratification. Videos showing children struggling to not eat the marshmallow after the experimenter and parent leave the room, so that they may claim more marshmallows later, are both endearing and powerful demonstrations of the heretofore-thought-to-be uniquely human experiences of impatience, frustration, self-control, reward and gratification, and the ability to plan ahead. Ravens, Carrion Crows, and New Caledonian Crows all aced versions of the marshmallow test, thereby breaching another hallmark.67, 68
♦ ♦ ♦
Crows play, have friends, and mourn the death of friends and family members.69, 70 It’s said that as more and more similarities in the cognitive capabilities, biases, and types of errors are exposed, the more likely it is that crows think like we do. And although their brains are built differently and most testing so far has been originally mammal-oriented, the list of cognitive capabilities crows share with us is already pretty impressive: abstract rules and analytical reasoning, consolation and reconciliation, mental representations and goal-directed behavior, innovation and insight, technological advances, transfer of concepts, knowing what others know, lateralization, tool manufacture and use, metatool use, comprehending quantity and numbers, planning for the future, recursion, motor and vocal control, tactical deceit, and even tracking humans, remembering our faces, and deciphering our intentions.
I wonder what else crows might show us if we knew what and how to ask. We are similar in that we are diurnal and we rely mostly on vision and hearing to perceive and respond to our surroundings, but our umwelts (the term coined by the biologist von Uexküll for the different perceptual worlds of different organisms) differ in myriad other ways. Right? They pick through poop to find bugs! They stand on ice in bare feet! They fly!
I wish we could know how they think, and that maybe in contexts such as greed, selfishness, cruelty, and war, that we could think more like they do.