05 May 2005

(76) Raven Consciousness - Heinrich


Raven Consciousness
Bernd Heinrich

In my most recent research I tried to figure out if ravens can think, that is, if they have the ability to execute the best solution to a simple, but at least novel problem, without first being programmed to do it (such as by purely hard-wired responses or by trial-and-error learning). Before starting this project, I had not given much thought to the idea of trying to get data on what may or may not be occurring in an animal’s mind, largely because I was skeptical of being able to get results. My intent here is to provide an overview of a research trajectory that spans a range of taxa with whom I’ve had experience, and to provide my assumptions and approaches. The results, conclusions, and steps in the logic have been published elsewhere.

Beginning With The Bees

Starting with insects in the 1960s, I tried to solve questions that involved primarily physiology and evolution, such as: is body temperature regulated, and if so, how and why? Relatively clear answers could be found through long-standard methods of measuring body temperature, blood flow, energy expenditure, heart and breathing rates, heating and cooling rates and so forth, in the context of comparative physiology. However, when trying to solve puzzles of evolution and adaptation, the ultimate reference is the field where there is no clear boundary between physiology and behavior. The lab situation, because it is controlled and thus contrived, allows discreet answers to the most basic, fundamental of mechanisms that, like "bricks", build the whole animal. Thus, a bumblebee might at one kind of flower, in one kind of weather, under one condition of the colony, precisely regulate a thoracic temperature within a degree of 42C and have a variable abdominal temperature of 25 - 30C. Change any of the above and thoracic temperature might be 30C and abdominal temperature 10C, or both temperatures might be regulated near 35 - 40C (Heinrich 1979b). In another taxon the data would likely be radically different, despite similar underlying generalities that apply to all. Details matter profoundly. The complexity that was revealed in insects hinted at sophistication that seemed unanticipated and surprising, but it ultimately "made sense" after all when seen in terms of the larger picture of adaptation (Heinrich 1993).

Not every potentially-relevant factor could be measured. For example, it seemed that a bee exhibited something akin to "excitement" when it found flowers with a high nectar content: its breathing rate and body temperature shot up immediately, it flew much faster, its flight tone went from a hum to a buzz, it became more selective in flower choice, and it made more frequent foraging trips. The change of behavior clearly and unambiguously registered that the animal could measure food quality, but whether it might know this consciously, as opposed to reflexively, was of no relevance to the questions I asked or felt I could ask. The behavior could be accounted for in terms of rote learning superimposed on innate programming (Heinrich 1976, 1979b, Heinrich et al. 1977). Bumblebees have a relatively open program concerning which flowers to visit and how to manipulate them to most quickly extract either pollen or nectar (Heinrich 1979a), but within a few flower visits they learn to heed specific flower signals and adjust their foraging routes and flower-handling skills accordingly.

The bees’ behavior was, after all, predictable, and much like their physiology the responses served specific functions either in the context of predictable environment or predictable changes of the environment. They were ideal organisms for demonstrating often highly intricate evolved responses, including specific learning tendencies, to all sorts of environmental contingencies. Although I saw no evidence that their sometimes complex responses could not be accounted for by programming alone, there was, of course, no objective reason to either exclude or accept the possibility that they consciously "knew" what they were doing after they were doing it.

In the whole animal the various responses are integrated and "make sense" in terms of a larger program. Thus, the energetics of thermoregulation is a component of foraging behavior, because thermoregulation is primarily used for foraging (Heinrich 1979b). In bees, furthermore, the foraging responses of individuals tie in with the colony economy and cannot be fully understood except through the perspective of the colony response in the context of specific environment. For example, honeybee workers communicate location and quality of potential food sources to hive-mates. Bumblebees who are "equally" social, do not. The difference is that honeybees, originating in the tropics, are adapted for harvesting from clumped resources, such as flowering trees. Bumblebees, on the other hand, are tundra- or taiga-adapted animals who forage from widely-dispersed flowers where communication is of less importance to the hive economy (i.e. the queen’s reproductive output).

And Going To The Birds

This is where the ravens came in. Ravens are well-known to be solitary and territorial breeders (Boerman and Heinrich 1999). As such, they should have no apparent advantage, like honeybees, to communicate locations of food bonanzas. However, since I was myself attracted to a ravens’ feast due to the birds’ loud activity, I was impelled to test whether their vocalizations attracted other ravens. Indeed they did. That is, other birds came to vocal playbacks who then also fed; strictly and objectively defined, the food was being shared. To me, whether the food was being shared "willingly" in the sense of "deliberate" recruitment, or whether recruitment resulted "inadvertently" or from the fact that they behaved mindlessly (without knowledge of consequences) but as in the bees in a way that was adaptive, was at that point not a relevant question. Others had to be answered first. 1) Does their vocal activity draw in others? 2) Do those that are drawn in get to feed? 3) Is there an advantage for the ones whose vocal activity attracts the others to have the others come? The psychological underpinnings to their behaviors were surely interesting. But they were out of my realm as a behavioral ecologist. As in the bees, sharing behavior among ravens could evolve by natural selection. For example, there would be some advantage for ravens to share very rare super-bonanzas if they all do it. The biggest theoretical hurdle to the above sharing-the-risk idea was that there seemed to be no mechanism for ensuring "honesty" in what would proximally involve altruistic behavior, given that the raven crowds were not likely to be groups of kin nor closed flocks of individuals who knew each other and would, furthermore, remember favors and therefore be able to play tit-for-tat.

The research that ensued to try to decipher the ravens’ sharing behavior was physically demanding, but perhaps the intellectually most rewarding for me so far. I knew that within the birds’ overt behavior lay a huge enigma (Heinrich 1989). At the heart of this puzzle was the question of how or why sharing among strangers, or near strangers, could occur on the basis of self-interest? There had to be an immediate advantage for attracting others to the feast. It turned out, of course, that there was: The sharers were juveniles who got access to new, untested and hence feared food and/or food defended by more dominant adults (Heinrich 1988; Heinrich and Marzluff 1995). Given this advantage, the other and perhaps later even main advantages (such as sharing the risk of not finding food) could be easily added on as "riders". Recruitment and sharing occurred (Heinrich and Marzluff 1991) even in the proximate unlikelihood of any psychological willingness to share (Marzluff and Heinrich 1991; Heinrich et al. 1993) and it occurred with non-kin (Parker et al. 1994), i.e. without kin selection. These data thus closed the loop on the problem I set out to solve.

"Cognition," used in the sense of at least some conscious knowing with resultant purposive actions, then seemed like a possibility to think about. I had not credited bees with knowing or being conscious of the consequences of their waggle dances, and thus performing them because they anticipate the positive consequences (i.e. not doing them if the situation were manipulated to cause negative consequences). Why? Largely because this scenario of corrective action presupposes they get not only satisfaction from dancing, as such, but that they also get a reward from the consequences of their dance, i.e. seeing others rush out of the hive to forage at the food indicated. Not crediting bees with such to them probably superfluous powers, I would therefore not risk valuable research time hoping for positive results in trying to test such a scenario. With ravens, on the other hand, there is a difference -- a huge difference. Closeness to various pet birds since my childhood has acquainted me with their emotional nature, a nature that is presumably adaptive(by rewarding fitness-enhancing behavior). Might not a raven be emotionally rewarded if it sees others come that will now make it easier to feed? And might it therefore also not be motivated to recruit because it anticipates the same psychic and hence later material rewards?

I could not and have not eliminated certain aspects of cognition from the mechanism that we have elucidated in ravens whereby strangers recruit to food bonanzas and share. I do not know what they intend or are consciously aware of and what behaviors are proximate reactions to stimuli. However, I am thrilled that sharing can be explained without invoking any motive of sharing, because that makes it all the more remarkable and rational. It is much more convincing, and elegant, to find a mechanism where cooperation occurs as each individual attends to their immediate interests without having to invoke purposive logic (which is all too easily faulty in the long-term since it is subject to derailment by faulty or incomplete information to affect consequences). Nevertheless, that in no way precludes conscious involvement, despite it often being a detriment to efficient or rational responses (such as in gambling, for example).

The logic (i.e. seeing what is out of sight) is always timebound: there are instant or immediate consequences, consequences hours or days later, and potential breeding consequences. Awareness in terms of consciousness, if present, could be applied for deliberate planning to almost any time span. But the first and basic premise of logic is that steps can be tried out in the mind (Heinrich 1996, 1999) and mistakes corrected (see Allen and Bekoff 1997) for an anticipated outcome. How far into time consciousness may extend the reality perceived by any one animal is, however, less relevant to me than experimentally answering whether awareness, as opposed to programming, plays any role for any time span at all, in any animals other than ourselves.

Bees cannot rely on conscious planning for the future in storing pollen and honey, etc. (What if some forget?) anymore than any animal can safely rely on having sex only and solely for the rationally purposive reason of having offspring. The ultimate rewards must be subservient to stronger, proximal rewards, when the interviewing steps are long, arduous and complex. On the other hand, it may be quite difficult to preprogram a squirrel to choose the best route through unpredictable mazes of branches to a nut, conscious planning by mentally trying out a number of possible routes would likely be simpler and more reliable. Even jumping spiders appear to be capable of pursuing prey out of view from sight, while using indirect routes and changing their tactics as required, while invading the webs of other spiders upon which they prey (Jackson and Wilcox 1993, 1998).

My first intimation that ravens have some sort of awareness of immediate consequences, necessary for conscious planning, that would then guide their behavior concerned their food-caching behavior (Heinrich and Pepper 1998). Having numerous birds together in a large outdoor aviary, it was an education for me to observe their interactions. Bees could, through programming, execute impressive behaviors. But the ravens acted as though they could gauge the results of their actions even before they executed them; they altered their responses, moment-to-moment,contingent on what was happening. When they went out of their way to bury excess food, others tried to follow even though the food was being carried out of sight in the gular pouch. The followers (if subordinate individuals) acted surreptitiously, and they did not venture near the others’ hidden food until they had left the area. The cachers (if dominant) in turn either attacked the cache raiders (but not others) when they came near their caches, or else they relocated their caches after having been watched. Nevertheless, as much as all of this behavior looked like each bird knew what the other was going to do, it was still possible that the birds did not "know" in the sense of anticipating others’ actions until after they had taught themselves or had learned from experience. Of course, as in our own learning behavior, the birds may become conscious, "knowing" after learning the consequences, so the conservative criterion of a test (of knowing without prior learning) was not met. Nonetheless, given my day-to-day observations of the ravens, I eventually wondered if they might know something even though they had not learned it or had not been genetically programmed to know. In short, I wondered if they could go through behavioral "steps" in the mind, without also committing the body to the same steps first. If so, they could do the equivalent of "trial-and-error" leaning in the head, thereby omitting many errors from being committed by the body.

It would not have occurred to me to present naïve birds with a test involving food dangled on an almost meter-long string from a perch, if it were not for my close observations of ravens caching, that sometimes suggested deliberateness and hence potential awareness. The food on a string puzzle (Heinrich 1996, 1999, 2000) was ultimately presented to a series of different ravens that had been reared from nestlings. Prior to the test these birds had never experienced food or some other objects dangled by a string, so that I could examine the details of their behavior on their very first exposure. Could they perform a series of dozens of consecutive steps that had to be executed in a very specific sequence? Could they reach down from their perch, grasp the string, pull it up, lay it onto the perch, then step onto the string before releasing it with the bill, then apply pressure to keep holding the string fast to the perch (with variable pressure depending on the load), while reaching down again and repeating the exact steps several more times in succession? To complete the whole task would require that they get a psychic reward not only from anticipation and/or eating the food, but also from completing each of the proper intermediary steps in a sequence that made their ultimate eating of food more likely. In short, no satisfaction from proximally-unrewarding steps could be gained unless they realized (i.e. understood) that what they were doing contributed to their objective (Craig 1918; Timberlake and Silva 1995). Furthermore, given the test, I could "cross-examine" what they knew or didn’t know by prior training, say to red string, and seeing if they are then conditioned to red string or would preferentially pull up food but now provided for the first time on a green string. The strings could be crossed, to see if their concept of reaching the food is to pull up string "above food" or "attached to food." I could arrange the string so that they had to pull down on a string to have food come up. I could determine if they knew the food was attached to string, by forcibly chasing them off the perch after they had pulled food up to see if they would fly off with food that was tied on. In short, the string test provided opportunity for a wealth of information where the relative contributions of innate behavior, learning, and cognition could all at least be partially teased apart. Obviously any one behavior contains some aspects of all three, but my main point was to be as conservative as possible, to see if one could rigorously prove that at least some cognition involving consciousness was involved. The results (Heinrich 1996, 1999, 2000) could not plausibly be explained by the alternatives (random chance, rote learning or innate programming) as sole explanations for the problem-solving behavior.

Future work should include other birds, especially other corvids. Other birds with similar body construction should be physically just as capable as ravens are in performing the same task. Can they solve the same puzzle? If not, then why not? Future work will also test whether ravens can keep track of objects that are out of sight, a prerequisite for conscious planning. We already know ravens routinely keep track of food that others (other ravens and humans) hide. But can they project the trajectory of a moving object that is out of sight (such as a rodent moving through an opaque tube)? By these and a variety of other tests, conducted in a variety of taxa, we may hope to reveal one of the perhaps most variable phenomena in the animal kingdom, the ability to solve problems by the application of consciousness, as has so eloquently been suggested by numerous animal studies as summarized by Griffin (1998).


Allen, C. and Bekoff, M. (1997). Species of Mind. Cambridge, MA: MIT Press.

Boerman, W.I., and Heinrich, B. (1999). The common raven. (The Birds of North America, series edited by A. Poole.) Washington, DC: Acad. Nat. Sciences.

Craig, W. (1918). Appetites and aversions as constituents of instincts. Biological Bulletin 34: 91-107.

Griffin, D.R. (1998). From cognition to consciousness. Animal Cognition 1: 3-16.

Heinrich, B. (1976). Foraging specializations of individual bumblebees. Ecological Monographs 46: 129-133.

Heinrich, B. (1979a). "Majoring" and "Minoring" by foraging bumblebees, Bombus vagans: An experimental analysis. Ecology 60: 245-255.

Heinrich, B. (1979b). Bumblebee Economics. Cambridge, MA: Harvard University Press.

Heinrich, B. (1988). Winter foraging at carcasses by three sympatric corvids, with emphasis on recruitment by the raven, Corvus corax. Behavioral Ecology and Sociobiology 23: 141-156.

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

Heinrich, B. (1993). The Hot-Blooded Insects: Mechanisms and Evolution of Thermoregulation. Cambridge, MA: Harvard University Press.

Heinrich, B. (1996). An experimental investigation of insight in common ravens, Corvus corax. The Auk 112: 994-1003.

Heinrich, B. (1999). Mind of the Raven: Investigations and Adventures with Wolf Birds. New York: Harper Collins.

Heinrich, B. (2000). Testing insight in Ravens In The Evolution of Cognition, ed. C. Heyes and L. Huber, pp. 289-305. Cambridge, MA: MIT Press.

Heinrich, B. and Marzluff, J.M. (1991). Do common ravens yell because they want to attract others? Behavioral Ecology and Sociobiology 28: 13-21.

Heinrich,B., and Marzluff, J.M. (1995). How ravens share. American Scientist 83: 342-349.

Heinrich, B., Marzluff, J.M. and Marzluff, C.S. (1993). Ravens are attracted to the appeasement calls of discoverers when they are attacked at defended food. The Auk 110: 247-254.

Heinrich, B., Mudge, P., and Deringis, P. (1977). A laboratory analysis of flower constancy in foraging bumblebees: B. ternarius and B. terricola. Behavioral Ecology and Sociobiology 2: 247-266.

Heinrich,B., and Pepper, J. (1998). Influence of competitors on caching behavior in the Common Raven, Corvus corax. Animal Behaviour 56: 1083-1090.

Heinrich, B. and Smolker, R. (1998). Play of common ravens (Corvus corax) . In Animal Play, ed. M. Bekoff and J. Byers, pp. 27-44. Cambridge: Cambridge University Press.

Jackson, R.R. and Wilcox, .R.S. (1993). Observations in nature of detouring behaviour by Portia fimbriatat, a web-invading aggressive mimic jumping spider from Queensland. Journal of Ecology, London 230: 135-139.

Jackson, R.R. and Wilcox, R.S. (1998). Spider-eating spiders. American Scientist 86: 350-357.

Marzluff, J.B. and Heinrich, B. (1991). Foraging by common ravens in the presence and absence of territory holders: an experimental analysis of social foraging. Animal Behaviour 42: 755-770.

Parker, P.G., Waite, T.A., Heinrich, B., and Marzluff, J.M. (1994). Do common ravens share food bonanzas with kin? DNA fingerprinting evidence. Animal Behaviour 48: 1085 1093.

Timberlake, W. and Silva, K. (1995). Appetitive behavior in ethology, psychology and behavior systems. In Perspectives in Ethology, 11, Behavioral design, ed. N.S. Thompson, pp. 211-253. New York: Plenum.