Do Elephants, Chimpanzees, and Dolphins Think?

(Reflections on an Unfortunate Paradigm Shift in Ethology)

A Talk Given at the 30th International Ethological Conference, August 18, 2007, Halifax, Nova Scotia, Canada


Moti & Donna Nissani


Department of Interdisciplinary Studies, Wayne State University, Michigan, USA


Contact us:

Some of the articles—and many photos of chimpanzees, elephants, and dolphins—can be found in:


Abstract:  A common tripartite classification of the roots of animal behavior traces a particular behavior directly to one’s genes, to equally mindless trial-and-error learning, or to thinking.  Extant experimental lines of attack on the problem of demarcating thoughtful actions from other types of actions are briefly reviewed, paying special attention to widespread verificationist approaches and to the nearly-forgotten Fabre’s and Thorndike’s falsificationist methodologies.  We then review some of our own experimental applications of the falsificationist approach to elephants, chimpanzees, and bottlenose dolphins.  We propose similar, concrete, applications of falsificationist experimental designs to claims of thinking in corvids, apes, and other animals.  As long as we fail to apply falsificationist methodologies, Fabre’s claim that we all too often try to “exalt animals” instead of objectively studying them will remain as relevant today as it had been over a century ago.  Finally, even if all non-human animals lack consciousness, self-awareness, theory of mind, and thinking, we argue that ecology, aesthetics, ethics, kinship, human borderline consciousness, and limits to what science can know all suggest that animals deserve a far more prominent place in our moral compass than they enjoy now. 

The questions we wish to ask:  Can animals of some species think?  Solve problems in their head? (not just blindly try everything?)  Have a concept of self?  Are animals conscious?

These are probably interrelated aspects of the same underlying neural circuitry, but here we shall just focus on one aspect of this complex equation:  thinking.  We’ll make use of Marian Dawkins’ reflections (1993, Through Our Eyes Only, p. 97):


A possible definition of what we mean by thinking is not only having internal representation of the world but that it should be able to perform some sort of internal manipulation of that representation—working out what would happen if one element were changed, for instance—and behaving appropriately according to its changed presentation.  Thinking is therefore most likely to occur in animals where working out in advance what the best course of action might be is a great deal quicker and safer than trying each one in turn and seeing which one is best in practice.  If all an animal can do is to follow a set of rules, then there is no reason to suspect it has a mind—that it can “think.”


Two misconceptions must be first be dealt with:


Misconception #1: Thinking confers great evolutionary advantage by allowing an animal to read other minds and solve problems in one’s own mind without trying everything through blind trial and error.  But these advantages are perhaps more than counterbalanced by:


a.     The most successful species, e.g., E. coli, fruit flies, probably do not think. 

b.     The brain is expensive to develop and maintain (cf., for example, Paul R. Manger, 2006, Biol. Rev., 81: 293–338).  These costs alone might outweigh any potential benefit.

c.     Thinking often involves an appreciation for complexity and uncertainty, and hence may lead to fumbling and hesitation in critical situations (that is one reason why organizations like the Iranian Revolutionary Guards and the U.S. Marines work so hard to eradicate thinking). 

d.     The only species that is demonstrably capable of thinking and consciousness is very young (<2 millions years?).  Going by this one exemplar, it would appear that the kind of thinking that evolution seems capable of producing (feeble, intermittent, conformist, self-centered, narrow in scope) could well be a recipe for extinction.


Misconception #2:  Animals deserve our love, our compassion, our moral consideration, only if they think.  We emphatically reject this narrow view.  We cannot take this issue up here, and instead we’ll merely support our objections with one quotation (R. D. Rosen, 2007,  A Buffalo in the House, p. 200):


He had been around a long time and he hadn’t met  many men who could understand the great but fragile web we were all part of and wanted to help keep it together.  What Roger felt was that there had to be a better America than this one.  And maybe it was that hope that helped explain his love for the crippled beast, this symbolic survivor of nineteenth-century America, in the back of his horse trailer.  It was animals who kept men honest.  No amount of money or flattery—or even carrots—could get an animal to be untrue to itself.  The ground on which humans, which were part animal, met animals, more human than we know, was sacred.  Animals taught us to love even when we couldn’t know whether we are loved back.  It was there, in an animal’s heartbeat, that we could feel the pulse of something bigger than we were.  There, on that ground, we could feel the pulse of something bigger than we were.  There, on that ground, we could feel that we were a part of nature, not apart from it.  Roger flashed on the situation in Yellowstone again.  He knew that a man who didn’t treat an animal with respect not only had no respect for nature; he had no respect for himself.


The little-recognized Henri Fabre (a working class Frenchman with too many principles and too few connections, but nonetheless, in a fairer world, a serious contender to the title of “father of ethology”, e.g., <1915, The Hunting Wasps.  Read Fabre’s works, then read Lorenz’s, and judge for yourself!) provided beautiful, solid, numerous observational refutations of the then near-universal belief that insects think.  In one case, Fabre concluded “to understand that she [digger wasp] can take a leg instead of an antenna is utterly beyond her powers.”  The actions of insects. 

are like a series of echoes each awakening the next in a settled order, which allows none to sound until the previous one has sounded.  What a gulf separate intelligence and instinct!


A few decades later, a similarly powerful inference about the lack of the “power of rationality” in animals emerged from E. L. Thorndike’s (1911, Animal Intelligence) observations on chicks, cats, and dogs. For instance, cats learn to escape from a puzzle box gradually, suggesting a stamping in of the association between the stimulus and successful response, not causal reasoning.  On the whole, Thorndike’s general conclusions about the mentality of chicks, cats, and dogs seem to have stood the test of time.  For instance, in a recent study, dogs learned to pull on a string to obtain food, but they did not seem to understand the means-end connections of string-pulling tasks (Osthaus, Lea, & Slater, 2004, Animal Cognition, 8:37-47).  Thorndike said:


 Most of the books [on animal psychology] do not give us a psychology, but rather a eulogy of animals.  They have all been about animal intelligence, never about animal stupidity.” Human are eager to find intelligence in animals.


Our plea is twofold:

 1. We need, indeed, to allow publication of a catalog of animal stupidities.  Ethologists, and the scientific community as a whole, should have in mind both seemingly intelligent—and seemingly unintelligent—exemplars before deciding the issue of animal cognition.  We can find such examples everywhere, once we start looking for them. 

2. Before attributing intelligence to a particular behavior, we must try harder to rule out the possibility that this behavior is solely traceable to genetics, trial-and-error learning, and the interaction between them (this is a variation of Morgan’s rule:  the point here not to automatically take a minimalist approach, but to try to experimentally disprove thinking).  

àWe shall give a few examples of this latter approach in the rest of this presentation.


String-pulling in ravens and elephants


 For instance, before concluding that ravens figured out in their head the string-pulling task, one might want to look for counterexamples, e.g., tying the string to a second perch, and seeing if the raven can still solve the problem.  If it understands the original setup, it should immediately grab the string that is dangling down from the second perch.  If it doesn’t understand the setup, it will first grab the one it grabbed before.

We have done this with elephants (Nissani, 2004, in:  Comparative Vertebrate Cognition.  Rogers, LJ, Kaplan G, editors, pp 227-261), using a bungee cord.  Like the ravens, elephants have mastered the task well: video.  At this point, we (and everyone else that saw this) felt that we have proven that elephants are capable of insightful behavior.

We then applied Fabre’s approach.  In one case, we have tied the end of the cord to a post (instead of placing it on the ground).  Conceptually, the two tasks are identical.  But with this minor discrepancy in the setup, our insightful elephants suddenly turned into a fumbling elephant!   We concluded that in the elephant’s case—and perhaps also (an experiment worth doing!) in Heinrich’s more famous case involving ravens—no insight was involved  (Video)



Wanda, formerly of the Detroit Zoo, readily mastered the string-pulling task (through a coordinated action of trunk and foot or trunk and mouth)—until the string was tied to a post.





àHere is a testable prediction:  Heinrich's ravens, like our elephants, will NOT pass a similar transfer task (and if so, their string-pulling is not insightful)

Do animals know that people see?

We applied Povinelli’s ingenious falsficationist paradigm to 6 chimpanzees and some 20+ elephants, with numerous variations and rigorous controls.   We faced a methodological problem involving the poor visual acuity of elephants.  Most likely though, we find that Povinelli’s claim of 50% rate is almost certainly mistaken in some of his own variations, and that in both species, in some tasks, performance is roughly at the 70% level.  This is statistically significant but, in our view, Povinelli’s main conclusion still stands (Nissani, 2004, in:  Comparative Vertebrate Cognition.  Rogers, LJ, Kaplan G, editors, pp 227-261).  The data are most consistent with an acquired trial-and-error preference to interact with a human who faces you as opposed to a human who doesn’t.  One has to see a chimpanzee, or an elephant, hesitating, looking intently at a person who has her head covered with a bucket and a person with a bucket on her shoulder, and then begging from the person who cannot see them, to come around to this view.  (That was the point when one of us changed his mind about animal cognition, seeing our brightest chimpanzee doing this). (Video)




Do elephants know that people see?  In the buckets task, our chimpanzees performed at chance level, while elephants chose the seeing person in 58%-78% of the cases.  In this trial, the elephant is begging from the person who could not see her.





Lifting a lid off a bucket and retrieving a reward

Our next example comes from the simple maneuver of lifting a lid off a bucket to get a reward inside the bucket.  Now, an elephant can be taught to do this well in some 30-60 minutes.  It looks like this: video

But what happens if you now place the treat inside the bucket, as before, and the lid alongside the bucket, on the ground?  Will the elephant behave as Fabre’s digging wasps and pine moths?  It sure does (Nissani, 2006, J .Exp Psych: Anim Behav Proc, 31:91-96):  video



After an elephant learned to lift a lid to retrieve food from a bucket, the lid was placed alongside the bucket while the food was simultaneously placed inside the bucket.  All elephants continued to toss the lid before retrieving the reward, raising the possibility that they have no understanding of this simple causal relationship








The above experiments:  String-pulling paradigm with variations, Povinelli’s paradigm with variations, and lid-lifting paradigm with variations, led us to suspect that chimpanzees and elephants do not think.


The “throw the net” signal of the dolphins of Laguna

We are now in the process of studying another candidate for thinking in the animal kingdom: the bottlenose dolphin.  To do this, we reluctantly let go of the experimental option and conducted preliminary observations of a striking (and beautiful) behavioral sequence:  the human-dolphin fishing cooperative in Laguna, Santa Catarina, Brazil (Pryor et al., Marine Mammal Science 6:77–82; Simões-Lopes, 1991, Biotemas, 4: 83-94). 



A Dolphin in Regular Motion A Dolphin Asking for a Net
A Dolphin in Regular Motion




The Signal and subsequent dive


As can be seen in the following video, the dolphins of the Southern coast of Brazil herd fish towards a line of throw-net fishermen.  At a certain point, the dolphin signals to the fishermen to cast their net.  To my knowledge, the signal is unique, not seen elsewhere in the behavioral repertoire of this animal. 

A detailed motion analysis suggests that the signal is short (mean time that any body part is seen above water while signaling: =1.4 seconds) and fairly uniform across individuals. (videos).

Our question, as before:  Do the dolphins understand what they are doing or was the behavior acquired through mindless trial-and-error learning?  We have here, undoubtedly, a culture, if we accept Whitehead and Rendell’s definition of culture, but to us, the critical question is not the existence of culture in this limited sense.  Rather, it is:  Is this a mindless or a thinking culture?  We hope to provide a more definite answer to this question when we finish analyzing our extensive video collection.


In this presentation, we’ve tried to contrast two ways of seeing the world, two paradigms, which, in their turn, create strikingly different ways of looking at our animal companions on this planet, something like the famous Young Lady and the Hag. 



An illustration of a Gestalt shift (A=hag; B=young lady; C=young lady and the hag)







These contrasting paradigms are not mere semantic quibbles, but they entail profound effects on our way of practicing ethology.  Here we shall only provide four illustrations of these effects:

Training of animals:  The two models—thinking vs. trial-and-error—lead to radically different training regimes.  (We may note in passing that all animal trainers we know operate as if the trial-and-error model is the correct one).

Animal “cultures”: Culture claims include Japanese macaques, bottlenose dolphins (Shark Bait sponges, signal to fishermen in Laguna), Whitehead’s sperm whales, chimpanzees nut cracking, and others.  The transmission of behavioral patterns across generations is undeniable, but, if only trial-and-error learning underlies this, if it is mindless, it is perhaps unwise to talk about “culture” in such cases.  At least until the issue is resolved, we should let go of the word “culture” and talk instead about something like “social learning.”

The ongoing debate on whether primates are cognitively special.  There is a growing body of experimental evidence from cetaceans, elephants, dogs, corvids, parrots—perhaps even geese, fish, bees, and ants—that they too deserve a place in the rising cognitive sun.  Again, the two paradigms lead to radically new ways of interpreting such claims.  According to the thinking paradigm, these animals are the cognitive equivalent of primates; that is, parrots may be as smart as chimpanzees.  According to the trial-and-error paradigm, the claim of equality still holds, but for different reasons:  If no animal thinks, then all animals are equally (un)intelligent.

Limited expression of new behavioral traits.  The coordinator of my workshop, Dr. Zhanna Reznikova, has just noted that culture and social learning are often restricted to a few members of a population, e.g., not all members of the snow monkey population wash or salt potatoes, not all bottlenose dolphins in Shark Bay use sponges; not all bottlenose dolphins in Laguna cooperate with fishermen.  Such observations are scarcely reconcilable with the thinking hypothesis, but they are implied by the non-thinking hypothesis.

Everyday Encounters: e.g., an eating dog attacking its scratching paw (video); the familiar scene of a dog on a leash wrapped around a tree and unable to extricate itself; chimpanzees losing their hands in the wild because they are unable to figure out an exceedingly simple trap mechanism.  For cognitive ethologists and lay people, these are disturbing puzzles.  By contrast, the non-thinking camp expects blind genes and mindless learning to produce at times such behavioral cul-de-sacs.   


In our view, the question of animal intelligence is an open one.  Science is not religion, and we should not adopt a particular viewpoint simply because we find it agreeable or intuitively compelling.  To find out whether a construct like “cognitive ethology” is valid in the real world, we should refrain from exalting and eulogizing animals, from directing all our energies to one side of this question, from resenting the view and blocking the publications of those who remind us that the animals may not think.  the moment, ethology needs a more cautious attitude to the question of animal intelligence, and it needs to develop a catalog of animal stupidities.  Whenever possible, ethologists need to apply Fabre’s falsificationist approach to any given situation, before attributing thinking, insight, theory of mind, or culture to animals. 



1.     After millennia of interactions with animals, we do not have a single, clearcut example of thinking or understanding in animals.  This by itself should give pause to the cognitive ethology avalanche.

2.     The larger brain of some animals serves many useful functions, e.g., improved learning ability, but not necessarily thinking or understanding.  There is, nowhere, neurological evidence that animals think.

3.     The founder of ethology, a great observer and a brilliant experimenter, and still the greatest ethologist of them all, J. Henri Fabre, is, at times, not even mentioned in animal behavior textbooks.  Possible reasons:  A Frenchman (the contemporary world of science is dominated by the politically ascendant English-speakers), low-class origins (likewise, we still talk about Darwinism, but the real priority to the idea of natural selection belongs to the working class Wallace), an unfortunate disposition in the social sciences to take premature theories too seriously and to look askance at great experimenters, beautiful writing style, mocking critiques of the venerable Erasmus Darwin and others who jumped to the conclusion that insects and other animals think.  His experimental manipulations still present us with the best hope for resolving the question of animal thinking.

4.     The subject of animal thinking is open, and the paradigm shift in ethology towards thinking is unwarranted yet:  We must think it possible that animals do not think.

5.     We need a catalog of animal stupidities (e.g., similar to an extant catalog on deceptions).

6.     In any given experiment, before concluding that thinking, or planning, or anticipation, or culture, or anything like that are involved, remove your caterpillar from the tunnel (or . . . tie your elephant to a post, place your lid on the ground, place a bucket on your head).

eleline.gif (2247 bytes)

Elephant Corner                Moti Nissani's Homepage