dirty rag

Ravens, Frans de Waal and Morals - Dan Dreifort

Corvus corax’ exemplary cognitive umbra vis-à-vis de Waal’s Pongidae ignoratio elenchi

(Examples of the common raven’s amazing cognitive abilities opposed to de Waal’s fallacy of offering proof and methodology irrelevant to the moral apes question)


Though they’re without a cerebral cortex (NATURE, 2002), the part of the brain usually associated with higher functions like learning, memory and expression of individuality, common ravens (Corvus corax) possess several requisite faculties or building blocks for morals. They share food, understand and plan for a future, recognize others and others' knowledge, and exhibit other startling cognitive and social abilities hinting at potential moral capacity. They achieve this cognitive freakshow without a cerebral cortex thanks to a part of the forebrain “called the hyperstraiatum that allows them to perform synonymous functions” (NATURE, 2002).

Ultimatum Game
So did Baltimore choose the raven because it was so smart? Will the Baltimore Ravens sue me for using this logo?.

Primatologist Frans De Waal's assertion that “it is only when we make general, impartial judgments that we can really begin to speak of moral approval and disapproval” (de Waal, 2006, p. xvi). necessarily implies that morals are limited to humans who are inevitably the problematic judges of any “impartiality.” de Waal defines a moral vacuum of sorts, reeking of human bias where we’re plaintiffs, judges and jury in a kangaroo court protecting the human-centric moral chip on our shoulders. Furthermore, Peter Singer points out that even de Waal permits that humans, “often fail to put the impartial perspective into practice” (de Waal, 2006, p. 144).

That an agent understands the concept of right and wrong and reacts to it emotionally is hard enough to prove for humans and especially for non-humans because we can't ask a raven how it feels and if those feelings induce certain actions toward or among others. Furthermore we can't determine why or how those feelings developed. With de Waal's criteria we'll likely never find a non-human moral agent. The criteria are flawed in that they're human-centric or at least require language use that is understandable and believable by humans. Even so, most humans fail the morality test.

Raven brains are different; so are their societies, and well, they fly! I will not argue that ravens possess moral faculty in de Waal’s sense, nor will I present an alternative, relative definition for morals with ravens championing them (though I believe it might be possible.) Instead I will show that de Waal wears his pro chimp/bonobo/human bias on his sleeve at the expense of recognizing ravens’ significant aptitude and their potential membership in the morals fraternity.  Recent experiments by Bernd Heinrich, Thomas Bugnyar and others highlight cognitive and social abilities of ravens attributed exclusively to primates by de Waal.

This article presupposes that morals exist and their origins are rooted in evolution.

Ravens, de Waal and Morals

In several of his books, Frans de Waal eloquently describes the incident of Kuni the bonobo picking up an injured starling and tossing it toward the barrier of her enclosure. de Waal's version has the bonobo comforting the bird and attempting to help it fly again. How do we know she wasn't being playful? Maybe she wanted to know what would happen if she hurled the starling into the air. Maybe Kuni was overtaken by some almost scientific curiosity[1]. Regardless, it's easy to imagine different and perhaps less complicated “whys” without involving nebulous notions of interspecies empathy.

Whereas de Waal occasionally (if not frequently) resorts to moral anthropomorphism, simple explanations are Heinrich and Bugnyar's bread and butter. It drives their research and explanations of their results. Do they think ravens possess the building blocks of morals? They're not saying. Morals seem too cloudy a concept for their fact-driven research.

Heinrich and Bugnyar's experiments involving the common raven suggest the conniving birds might surpass even great apes in cognitive abilities. The authors start strong, mentioning seemingly amazing stories of raven accomplishments in the wild. A raven warns a man of a hungry cougar's threatening presence by cawing loudly as it approaches his blind side. Another raven feigns illness to ward off a predator or perhaps to scare other ravens away from a potential food bounty.  Though perhaps anthropomorphic, until recently these descriptions were accepted as reasonable explanations.

Less discerning readers are instantly sold on the birds' genius until presented with more simple explanations for the astonishing behaviors. Though it’s comforting to think its shrieks were protective warnings, Ravens are now known to lead carnivores to prey they can't overpower themselves. And though it might seem obvious at first glance that the other raven was playing possum or feigning poisoning to keep potential food competitors away, such an explanation seems more convoluted once we know that ravens are well known in the bird world for their playful behavior such as randomly rolling around on their backs. However, we're left with enough ammunition to warrant further inquiry into the raven mind. De Waal stands to learn something from this trustworthy anti-pitch. The modesty of this approach is ultimately inviting.

Raven research “published in 1943 by Otto Koehler of the former Zoological Institute of Konigsberg… demonstrated that his 10-year-old pet raven, Jakob, could count up to seven by training it to retrieve food from under one of several vessels with various numbers of spots on the lid” (Heinrich & Bugnyar, 2007), implying  the birds' ability to use logic to solve problems. Though there was a gap of nearly half a century in research implying logical abilities of ravens (Heinrich & Bugnyar, 2007), the birds’ ability to memorize and recall myriad geographic locations over time has long been suspected from witnessing cache retrieval in both natural and controlled environments.

The authors note other common, clever raven behaviors but remark that observations thereof do not rule out simple explanations e.g. “instinct or learning specific actions by rote” (Heinrich & Bugnyar, 2007). In a sense de Waal’s morals have an instinctual background in that some building blocks thereof are not developed, but rather are inborn and seemingly inevitable, at least in most humans and some apes. Additionally the social aspects of de Waal’s morals involve a continuing process of acquiring behaviors, though not necessarily by rote, certainly by force of habitual exposure to societal norms.   Our raven authors discount instinct in that they’re testing for more advanced and rare cognitive abilities to explain remarkable raven actions. Nonetheless, a creature's ability to learn or even to share knowledge or behaviors is clearly of import to both research parties' motivations.

Cognitive Pecking Order – Problem or Useful Contrivance?

Heinrich and Bugnyar establish a hierarchy of sorts to weigh the three possible explanations for examples of ravens’ cognitive accomplishments. Innate abilities (akin to a spider’s ability to spin a perfect web fresh out of the egg,) learning (e.g. trial and error,) and what we’ll call advanced cognitive abilities (like assessing and planning before leaping,) are the three candidates. In their hierarchy thereof, which I call the cognitive pecking order (CPO) they imply that higher abilities (like evaluation combined with real-time reasoning and metacognition,) are superior to learning/memory which in turn trumps innate programming.

Their criteria for the CPO rankings aren’t totally absent, but neither are they explicitly stated. They come closest to justifying or explaining their CPO rankings when they define the top tier reasoning as “choosing from alternatives that are represented in the mind and evaluated” (Heinrich & Bugnyar, 2007), implying a requirement for meta-cognitive abilities not present in most lower animals, though recent research shows even the common rat exhibits metacognition (Choi, 2007). Additionally, they contend that if the ability to identify individuals is commonplace it is then the main defining feature of that society and therefore the ability to predict responses of others becomes more valuable than innate programming (Heinrich & Bugnyar, 2007).

After implying CPO they go on to identify their experiments' goals as, “to try to distinguish the role and the relative importance of [potential CPO explanations of raven feats.]” They largely succeed at the former (the role of CPO explanations) while the latter remains necessarily unproven. I.e. without defining clear criteria for determining relative import, it's a subjective endeavor at best. Furthermore I’m left wondering why they need a hierarchy as its presence clouds what is otherwise an eloquent presentation. Perhaps they propose CPO because they suggest ravens’ abilities are, “on par with, or exceed, those of the great apes” and because their research inevitably concludes raven behavior often climbs to the top of the CPO, perhaps a rare feat in non-humans. That ravens perch atop the highest rank of the CPO, a hierarchy our authors would have us believe is somehow important, might help to sway fence-sitters at the end of the day. Regardless, CPO is the only aspect of their write-ups I find awkward as it’s largely unsupported and therefore of questionable merit[2].

Raven Experiments

In the first experiment, Heinrich and Bugnyar (2005) devised a system to determine what sort of cognitive abilities ravens aged ten months could muster to deal with new problems. Unlike anything they might encounter in nature, food is tied to a perch with a long string. The raven must not only pull the string to raise the food, but is also required to hold the already-pulled string with its foot and then repeat several times to obtain its reward. Amazingly, the raven is able to quickly obtain the prize. In the mass media write-up of their experiments, they do not advertising how many of the birds successfully completed the tasks (Heinrich & Bugnyar, 2007), but neither do we get the complete picture when we read of special accomplishments of de Waal’s exceptional ape specimens. De Waal's arguments lose weight when we ditch the rarest of his extraordinary media-darling observations, e.g. the ape Binti Jua rescuing a baby human, Kuni’s bird, etc. It's stories like these that ring the loudest and longest for de Waal as they tug at our moral heart strings. Of note, upon further investigation into scientific journals we find that only one raven was unable to complete this test (Heinrich & Bugnyar, 2005). This first experiment shows us that ravens likely understand causal relationships between objects. This is not to be confused with similar tool use in chimpanzees in that their analogous behavior “involves much learning and is not necessarily initially because of causal understanding” (Heinrich & Bugnyar, 2005).

To be sure of their findings, they modified the experiment. In the second experiment, after wrangling the food all the way up to the perch, ravens are shooed from the perch to find out if they understood what they'd done. If they did not understand the intricacies of the procedure, the birds would hold on to the food as they flew away, inevitably causing some slapstick moments as the line goes taught and the food is violently yanked from the raven’s beak. Alas, this did not happen. Ravens quickly dropped the prize as they were shooed away. Furthermore, ravens flew off with string-tied food when it was not attached to a perch. This great use of the scientific method to determine the true depth of previous results is again something de Waal could use to advance his fascinating albeit largely untested observational body of work.

Sticking to the scientific method like an ants to a picnic, Heinrich and Bugnyar devised yet another test to solidify previous results. To determine if indeed ravens were exhibiting top CPO abilities they surmised that, “another group of string-pull naïve ravens should face difficulties in solving a mechanically similar but non-intuitive problem: pulling down to get meat to come up. However, birds that were already proficient at the pull-up task might transfer their knowledge to string-on-food provided in other ways” (Heinrich & Bugnyar, 2005).  They modified the string/perch setup so that the string traveled upwards from the perch through a wire mesh cage. The subjects could see but not grab the food directly.  Subjects had to pull down on the string to gradually bring the food within striking range. None of the new subjects were able to obtain the food. All but one of the birds experienced with meat on a string, quickly solved the pull-down task by reaching up, pulling a loop of string down, stepping onto the string and repeating the sequence until the meat could be reached” (Heinrich & Bugnyar, 2005). (The same poor sap failed both experiments.)

The authors explain their results:

As we carefully controlled for factors such as age at testing, conditions of keeping, experience with humans and exposure to string, we suppose that differences in performance reflect the ravens’ sensitivity to means-end properties of the tasks. Assuming means-ends comprehension, the direction of pulling would be essential for acquiring the skill because pulling down to get something [sic.] come up is counterintuitive (Heinrich & Bugnyar, 2005).

Though I agree this third experiment proves, “logic and not the rewarding of randomness” (Heinrich & Bugnyar, 2007) solved the prior experiments, part of me is conflicted over the intuitive/counterintuitive dichotomy they imply with the two different versions of the string test. Pulling down to make something rise reminds me of a pulley which at first blush seems a purely logical instrument. However, my vague memories of grade school pulley and lever lessons inevitably reinforce what our experimenters, ravens and John Kerry already seem to know: flip-flopping is indeed counterintuitive. Young humans don’t understand pulleys until they’re taught or otherwise immersed in a similar situation.

The three experiments demonstrate that, “ravens have the ability to test actions in their minds and project the outcomes of those actions” (Heinrich & Bugnyar, 2007). The authors suggest “a good adaptive reason” (Heinrich & Bugnyar, 2007) such behavior is remarkable and absent in most animals. “Thinking and logic can be notoriously unreliable and can lead to much mayhem...” (Heinrich & Bugnyar, 2007) which sounds reasonable on one hand, but is an unsupported offhand remark on another. Maybe it's an overly generalized tenet of biology and/or philosophy with which I'm unfamiliar.

So why then, (and they too ask this,) are some creatures not genetically programmed “to do it right” (Heinrich & Bugnyar, 2007)? If the powers that be see fit to endow a wasp's flyspeck brain with innate magical engineering and behavioral prowess, why then do humans toil endlessly for analogous/homologous accomplishments? The authors imply that our genes aren't coded for complexities because of the unpredictability of our complex social environment. Those same conditions enable us to think, learn and share. Perhaps there are similarities in the raven milieu.

Raven Culture and Cognition

Of course Heinrich and Bugnyar follow the same path and decide it's time to examine raven society to perhaps explain more of the birds' behavior. Though solitary creatures for much of their lives, ravens engage in food sharing and aural/oral communication. Research shows their varied calls range from appeasement and intimidation to discerning the type and quality of food available (Bugnyar, Kijne, & Kotrschal, 2001). Determined to find out why ravens amazingly call to each other when they discover food, in 1984 Heinrich's research showed that only certain younger birds act this way (Heinrich, Ravens in Winter, 1989).  Though their motivation is apparently nothing as noble as altruism, Heinrich's findings that this behavior is in fact intended to attract others in an effort to overwhelm resident birds who would otherwise defend their cache, hints at the possibility of friendship, alliances and empathy.   Bugnyar takes it a step further when he posits that ravens,  “have developed quite sophisticated sharing behaviour: they do this year round, include food and non-food items ('toys'), and direct it towards specific individuals;  the function is most likely to form affiliate relationships (friends)” (Bugnyar, 2007). De Waal discounts the evidence of helping responses in non-primates as “largely descriptive” (de Waal, 2006, p. 33). and says no animals other than primates, dolphins and elephants exhibit similar responses. Heinrich and Bugnyar clearly elucidate his error of omission with this raven evidence.

While Heinrich’s findings were the (perhaps now dated) fruits of meticulous data collection and then seemingly endless months of following many false hypotheses while testing, I was unable to locate experimental data to support or undermine Bugnyar’s claims. Regardless, de Waal could again learn something from our authors’ approach. E.g. when de Waal claims to witness targeted helping as one chimp seems to assist another’s attempts to obtain water from an obstructed tire (de Waal, 2006 p. 31), he could devise a set of experiments to validate his theories instead of resorting to anthropomorphic descriptions of  his untested observations.

So though the food sharing ravens aren't necessarily showing gratitude (how would we know?) or an example of retributive kindly behavior, (though it might be that and more,) such behavior and Bugnyar’s observations (results?) are nonetheless likely not good examples of Westermark’s “disinterested” or truly moral/human fairness, though I encourage our researchers to test that possibility. To be fair, even de Waal indicates his observations of non-human primate food sharing is likely egocentric and therefore not grounds for proving a concept of fairness in primates (de Waal, 2006, p. 48).

To understand Heinrich and Bugnyar’s next social experiments, they provide an overview of some raven behavior and suggest reasons for it. Ravens are scavengers regularly dealing with the scraps of dangerous carnivores. Innate programmed responses to dangerous carnivores could often prove fatal to the ravens so maybe there’s an indirect reason ravens play, e.g. with and near other animals. They suggest that ravens play with carnivores when they're young, not in order to learn, but because knowledge of the reactions and abilities of their play targets comes in handy later in life as they're competing for food (Heinrich & Bugnyar, 2007).  Regardless, ravens eventually learn the limits (or die in the process) and then must hide food away for later days. Ravens cover food with items and earth and then memorize the locations of many food stores, even those of other ravens. In short ravens’ opportunistic feeding, seemingly random play and their sly predators all intertwined to cook an evolutionary cognitive soup.

To test whether playing helped young ravens “gain the ability to adjust their behavior flexibly” (Heinrich & Bugnyar, 2007), the authors designed a large aviary with near-natural conditions. Their elaborate experiment in which the authors acted as surrogate parents showed that young ravens learn to cache objects by playing. Within a month or two, though still dependent on surrogate parents for food, the ravens are caching edible and inedible objects and burying or otherwise hiding them, all the while observing their siblings. Did this playful caching help predict others' behavior so that they might hide food more successfully as adults?

To find out, one researcher (a) began to always steal the test subjects’ caches while the other (b) would examine the caches but leave them intact. When presented with food caches (no longer dealing with non-food objects,) the now older ravens delayed caching around the thief and would immediately recover any cache the thief approached. Conversely, the ravens cached immediately when the non-thief was present and did not react when he approached existing caches. The ravens not only learned from their play, they recognize individuals across species.

They then removed humans from the equation; one bird (The knower) is able to watch subject store food. A third non-knowing raven is then placed in the aviary. The subject ignores this non-knower even as it approaches a cache. However, when the knower approaches the cache, the subject either scares the knower away or removes the food immediately. Ravens are able to discriminate upon the knowledge of other ravens.

That ravens are able to make and execute plans based on external meta information hints at knowledge of both past and future and possibly self-consciousness. Ravens have not yet been subjected to the mirror self-recognition (MSR) test; a test de Waal admits likely indicates high levels of self-awareness and cognitive empathy (de Waal, 2006, p. 36). Bugnyar indicates that, “We haven't tested [ravens] yet. But this is definitely a thing to do in the near future.” I have faith ravens will fly through this test easily. Even the lowly pigeon has passed MSR tests multiple times (Allan, DeLabar, & Cardinal, 1999; Epstein, Lanza, & Skinner, 1981). I was able to learn of one person who placed mirrors on the ground in an attempt to scare ravens from an area. The ravens defecated on the mirrors, covering them in excrement. I surmise the wise ravens acted as they did out of economy: eliminating their bowels eliminated the nuisance of the mirrors.

Conclusions and parting shots

I’ve shown that ravens have a theory of mind. Unlike chimpanzees, their interactions with their environment are rooted in understanding causal relationships. Ravens use logic to assess and solve problems and they have a concept of past and future. They can mull over what they know and apply it to new situations. Ravens communicate relative concepts and emotions and share food. They recognize individuals across species and recognize and react to knowledge of others. They do all of this without a cerebral cortex.

Though not discussed in this paper, other research shows ravens, like some primates, can follow and react to the gaze of another agent, even a human's... using it as a behavioral queue thus empathizing with another. Other research into raven communication indicates some calls provide information about the caller. I.e. some calls are functionally referential in that they elicit the expected responses when played back. I cannot argue that ravens are more deserving of the moral agent tag than primates, but I’ve presented a wealth of recent evidence to suggest that a moral raven is not entirely impossible and certainly not significantly less likely than a moral chimp.

I largely wasted some ink discussing the CPO, trying to define its criteria and overall import.  Here’s my closing statement on that: Heinrich and Bugnyar tell us that sometimes innate responses can be problematic but they admit as much about higher reasoning too. I think it’s unimportant to rank the explanations for behavior. Explaining them is enough if not even superfluous to these raven experiments.

I’ve mentioned several shortcomings in Frans de Waal’s observational methodology as compared to Heinrich and Bugnyar’s diligent predictions, controls and most importantly falsifiability. I find it noteworthy and disappointing that de Waal does not stick to the scientific method and instead often relies upon emotional fallacies to appeal to the mass market. That ravens are the most widespread naturally occurring bird in the world (Heinrich & Boarman, 1999) surely tells us something about their cognitive prowess. After all, what species is the most widespread primate? Frans de Waal’s fascination with primates is his Achilles heel in that it leads him to make bold unconfirmed statements.

Works Cited

Allan, R. W., DeLabar, J. S., & Cardinal, C. D. (1999). Self-awareness in the pigeon: A replication and controls. Association for Behavior Analysis Abstracts. Chicago.

Bugnyar, T. (2007, April 30). Can I pretty-please ask you a couple of questions about ravens? (D. Dreifort, Interviewer)

Bugnyar, T., Kijne, M., & Kotrschal, K. (2001). Food calling in ravens: are yells referential signals? Animal Behaviour (61), 949-958.

Choi, C. Q. (2007, May). Rat-Think. Scientific American , p. 36.

de Waal, F. (2006). Primates and Philosophers. Princeton, New Jersey: Princeton University Press.

Epstein, R., Lanza, R. P., & Skinner, B. F. (1981). "Self-Awareness" in the Pigeon. Science , 212 (4495), 695-696.

Heinrich, B. (1989). Ravens in Winter. New York: Simon and Schuster.

Heinrich, B., & Boarman, W. (1999). Corvus corax: Common Raven. The Birds of North America (476), 1-32.

Heinrich, B., & Bugnyar, T. (2007, April). Just How Smart are Ravens? Scientific American , 296 (4), pp. 64-71.

Heinrich, B., & Bugnyar, T. (2005). Testing Problem Solving in Ravens. Ethology , 111 (10), 962-976.

NATURE. (2002). Retrieved May 13, 2007, from PBS: http://www.pbs.org/wnet/nature/ravens/print/ravens.html


[1] Would a chimp's scientific curiosity be any less impressive or empathetic?

[2] Bugnyar mentions Premack & Woodruff’s cognitive levels hierarchy in ”Leading a conspecific…”.