Perspectives On Medical Research
Volume 5, 1995
Aping Science
A Critical Analysis of Research at the Yerkes
Regional Primate Research Center
Primate Models of Human Traits
Richard C Lewontin
Museum of Comparative Zoology, Harvard University
Introduction
We would like to understand both the immense diversity of physiology and behavior among human beings and to learn what, if any, are the limits on health, longevity and behavior. Why are some people healthy and others sick, some shy and others outgoing, some pacific and others aggressive, some able to do mathematics and some not? Are there limits to the functions of individuals and our species as a whole, a "human nature" that cannot be eliminated or transgressed? What programs, social or individual, would have to be carried out to produce people with or without certain behavioral and physiological characteristics? In particular, are behaviors "hard-wired" at birth into the anatomy and physiology of people, in which case they could be influenced only by physical manipulation of the wiring pattern, or are the connections being remade and rebroken under the impact of external stimuli, in which case social organization is relevant? These are the questions that connect our political lives to our scientific work and they obviously go to the heart of what is seen as most important in the organization of society.
The problem is that we cannot manipulate human beings experimentally in the same way that most scientists feel free to manipulate non-human animals, so we have no methodology for dealing with these questions of human behavior directly.
The response of scientists has been to substitute primates as model organisms for humans, to manipulate those models behaviorally, anatomically, and physiologically, and then to assume that the differences between, say, chimpanzees, and humans is so small as to allow a direct translation from the models to us.
The scientific problem of using primate models to study human beings, ethical issues quite aside, is an example of both a general issue of how models are to be used to study physical systems, and how biological systems pose particular difficulties for the use of models. In what follows, we look first at the general problems of model making, then at the specific translation of these problems into the biological realm, and last at the unique problems of primate models of human traits.
General Problems of Models
We use models to study a system when the study of the thing itself is inconvenient or impossible. In the case of the study of human beings, especially of human behavior, social and ethical convention, as well as law, make it undesirable, repugnant, practically impossible or simply illegal, to manipulate human beings in ways that would be regarded as necessary for well controlled experiments. In place of actual human beings as the objects of experiment, use is made of formal models such as mathematical equations or computer programs, or of physical models embodied in other living organisms, or sometimes of acceptably derived bits and pieces of humans, as in cell culture studies taken from biopsy, or, more controversial, from aborted fetuses. The scientific issue in all such cases is the same: How do we use the properties of the model to infer the properties of the thing modeled?
A model is chosen because it is supposed to embody certain properties that are isomorphic with the system being modeled. That is, these properties of the model are in a one-to-one correspondence with a set of properties of the real system in such a way that when the model is perturbed in a certain direction, the result of the perturbation is in a known one-to-one correspondence with what would happen if we perturbed the real system of interest. It is important to note that the result in the model system does not have to be identical with what would happen in the real system, but only that we have some reliable rule of translation from the model result to the real object.
So, if I raise the temperature in which rats are living and measure the change in their heart beat, we do not require it to be the same change in heart rate as would occur in human beings, but only that it have some predictable relation to what would happen in humans. Indeed, in general, a model provides us with a set of correspondences between its own properties and those of the system being modeled. As we shall see, the establishment of these correspondences is one of the deep problems of model making.
We know, in advance, that these properties of the model objects are to be excluded from any reasoning about the objects being modeled. The general problem, however, is that we often do not know what properties the model has that are not isomorphic to our system, so the behavior of the model when perturbed may include all sorts of effects that we do not want to carry over. To try to avoid such extraneous effects two paths are taken. One is to make the model as different as possible from the modeled object except in those deliberately constructed properties that are meant to be used in prediction. That is the purpose of computer models. The program is formally isomorphic with deliberately chosen properties of our system, but we are not fooled by the hum and the lights. The alternative is to go in the opposite direction and try to find a model that is as totally similar as possible to the object being modeled, in the hope that whatever results are seen in the model can be carried over correctly. That is why primates are regarded as good models of human beings. But this latter tactic is a dangerous one. No model is identical with its object or it would be the object and not a model. Norbert Wiener once remarked that "The best model of a cat is another, or preferably, the same, cat." The moment we change cats we will carry over some non-isomorphic property and all we can hope is that it doesn't matter.
The second, related, problem of model systems is how we are actually to establish the rules of correspondence between model and object. The only fool-proof way is to understand all the properties of both, but if we could do that we would not need the model in the first place. Even a computer model of a system, one that has been deliberately constructed with certain formal properties, does not solve this problem, because, although we may understand perfectly the properties of the computer program, we do not know whether is has captured all the relevant properties of the object being modeled. The only way to do that is to know all the answers to begin with. Computer models are nothing but logical apparatuses based on certain assumptions. The question is about the assumptions. The alternative is to make a physical model and count on some general properties of physical systems that will carry over from one to another.
The rules that govern the relationships between parts in the model, if they are a manifestation of some general physical properties of nature, can be regarded as preserved in the object of real interest. The more overall physical similarity the model and object have, the greater confidence the model maker has that the relations between parts carry over in detail. Once again, similarity does not have to mean identity. Many models are scaled down in size from the real objects, but if volume scales as the cube and surface area as the square of linear dimensions, then appropriate adjustments can be made for a change in size. Of course, the question is whether a mouse is, for a certain purpose, really nothing but a small rat.
Application to Biology
Biologists have had their consciousness of their subject formed from the example (model!) of physics. Newton's Laws, the regularities of the solar system, the laws of combining proportion, are all held out as the proper model of a real science, and it is felt that biology, being after all nothing but the study of particular physical systems, can be built along the lines of physics, with a few general laws and principles that will then apply across all living organisms. But this has not happened. There are no “laws” of biology like Newton’s Laws. Only one of Mendel’s “Laws’ has any wide application at all, the law of segregation, and there are many known cases where it does not work. Even the ”law” of all life from life, is historically contingent, since there must have been the abiogenic origin of life at sometime in the past. It is simply that conditions have changed. Among other things, the living organisms that now exist simply consume any complex molecules that might serve as a new origin of life. The reason that there are no general “laws” in biology is that organisms are physical objects with two properties not seen in those physical systems to which the great general laws of physics apply. First, they are intermediate in size. Both gravitational and electromagnetic forces are weak and are effective only between objects that are either extremely massive like planets, or extremely close like atoms. Physics does a bad job of describing what actually happens when an intermediate size inelastic object, say a rock, hits the ground. It is true that organisms are composed of molecules, but so is a stream of water, yet there are no general equations of hydrodynamics that will predict the behavior of a stream of water as it leaves the tap. Hydrodynamics, which requires complex computer programs to describe the behavior of its objects of interest, is a much better model for biology than nuclear or classical physics.
Second, biological objects are internally heterogeneous in a way that is functionally relevant. While the earth has some variation in its internal composition, the behavior of the earth as a solar system object is unaffected by that heterogeneity. Organisms, on the other hand, are affected at every level of their functioning by their internal heterogeneity. (They can even change the way they fall “freely” through space as gravitational objects, as every sky diver knows.)
The consequence of intermediate size and internally heterogeneity is that organisms are not governed by one or two major physical forces or conditions, but are the nexus of a very large number of interacting forces, none of which is individually powerful or determinative. The force of gravitation and the velocity of the planets determines heir orbits. No such simple dependency exists for organisms. They are extremely complex machines made of a large numbers of different bits and pieces that are partly in close interaction, but also partly segregated into subsystems within which interactions are strong but between which there only weak dependencies. Moreover many of the molecular species that are critical to the functioning of these machines are in very low concentration, sometimes only a single molecule per cell, so that the laws of mass action cannot be counted on to produce conventional chemical equilibria.
The result of these properties of organisms is that no two are exactly alike, not even members of a clone within a species, because accidents of development and physiology make even genetically identical organisms different. From standpoint of using one organism as a model for another, the question is, how different and in what way? One the one hand everyone is conscious in an intuitive way of the immense diversity of living forms. Yet there are functional universals for organisms, the requirements of energy metabolism, protein synthesis, DNA copying, which are presumed to enforce certain similarities on all living forms.
There is, moreover, the historical element that produces similarities. All organisms are more or less distantly related and have acquired their various characteristics by descent from common ancestors. So we expect that two organisms that have had a recent common ancestor will be rather alike in their properties. Two mammals will be more similar than a mammal and a bird, and a member of the human species ought to be more similar in general to a chimpanzee than it is to a rat.
There are then two general principles that the biological model maker uses. First, it is assumed that the detailed similarity between organisms increases as one goes to more and more basic cellular processes within organisms. So, a rat and a human may not look alike, but they are assumed to have similar general structures for their nervous systems, extremely similar chemistry for the actual firing of individual nerve cells, and identical chemistry for the copying of the genes that code for the production of these chemicals. So, the model maker feels comfortable carrying over the results from rats to people if it is a question of basic cellular processes. That is why rats are used for drug trials, at least to screen for harmful effects of new drugs. Second, it is assumed that whatever differences may exist at any level, these get smaller and smaller as one compares organisms that are more and more closely related in evolution. Thus, the model maker feels confident about carrying over nearly anything seen in chimpanzees to people.
The problem is that neither of these rules turns out to be reliable. There is an extraordinary diversity among organisms and between different aspects of the same organism even for the basic cellular processes. When the first work was done on the molecular mechanism that turn genes on and off during the metabolic activity of an organism, it was a assumed that there was one general model of so called “gene regulation.” That is, it was assumed that there was a single form of signaling between the organism and its genes, to indicate when a particular product was required. It is now known, 40 years later, that there are dozens of different forms of molecular instructions and paths of communication between an organism’s state and its apparatus of production of proteins from genetic information, and that different genes in the same species are radically different in their control circuits. Indeed, a given gene may be more similar in its control across species differences than two different genes in the same species. It all depends on cases. At the level of cell metabolism, there is even less carryover from species to species. An interesting case is the toxicity of erucic acid, a component of rapeseed oil. Rapeseed oil (canola oil) is an extremely important agricultural product in Canada, and there was considerable alarm when it was discovered that the oil contained a compound, erucic acid, that was quite toxic to laboratory rats and mice. Millions of dollars were spent breeding rapeseed to get rid of this compound, but it turns out that humans are not affected by erucic acid, nor indeed are laboratory animals like guinea pigs or rabbits.The sensitivity is a peculiarity of rat and mouse metabolism. Of course, it is better to err on the side of caution, but suppose that the cases had been reversed and that it was humans who had the unique sensitivity. No animal model would have detected it. Until one has a detailed understanding of the physiology and metabolism of a species, it is not clear when one animal can be a model for another, even at the level of basic metabolism.
When we pass from basic metabolism to behavior, for example, all bets are off. The behavior of different species, even closely related species, is the outcome of the evolution and day to day reaction of organisms to their particular mode of existence, and this varies markedly even between close relatives. For example, the two species of baboon, the gelata and the hamadryid of east Africa, that are favorite subjects of observation by primatologists, have utterly different social and individual behavior patterns, including the size of their troops, how they apportion their time during the day, their mating patterns, their family structures, etc. These differences are probably the result of the very different availability and pattern of food resources in Kenya and Ethiopia in which the two species live. Another example is to acquisition of song patterns in birds. Perching (passerine) birds have songs that are characteristic of their species and a great deal of effort has been expended by behavioral biologists in studying how perching birds acquire their song. The answer turns out to be that it all depends on the species. For some, they will sing their characteristic song, even if they have never heard it. For other species, they must learn it from their parents. For some, hearing only a roughly similar song will trigger the development of their characteristic song. For others they must hear exactly the right one. Some will learn any song at all if they hear it early enough, others cannot sing anything except their species song. A bird like the cathird has no characteristic song, but constantly varies what it emits, with no repetition. There is no general rule, and one cannot judge the mechanisms of song acquisition of one species of perching bird from studying others.
These studies of baboons and birds illustrate the problem with the second “rule” of using model organisms: that closely related organisms will have the same mechanisms for the same function. What is involved is the widely misunderstood concepts of “analogous” and “homologous” characteristics. Two organisms have analogous traits if the traits have some recognizable similarity, but arose completely independently in evolution, while traits are said to be homologous if their possession by two different organisms is the result of their acquisition from a common ancestor. [See figure 1] To take a simple example from outside biology, two persons with the last name “Smith” may come from two completely independent family trees, both named “Smith” because the original founders of their family lines were actually makers of iron objects. These “Smiths” have only analogous names. On the other hand two people named “Smith” may have had the same great-grandfather named “Smith”, and so their last names are homologous with each other. A case in biology is the origin of wings. Bats have wings and birds have wings, but their wings are only analogous, because both birds and bats originated independently from separate wingless ancestors. On the other hand, the wings of sparrows and the flippers of penguins are homologous, even though one is used for flying through the air and the other is used for flying through water, because penguin flippers are modified wings that they acquired from a winged bird ancestor. The more recently two species have a common ancestor, the more likely in general it is that two similar characters they possess are homologous. So, the argument goes, if we use a very closely related species as a biological model for another species, we can make inferences about what is going on in the species of interest. That is why chimpanzees are used as models of human beings.
While this distinction between analogy and homology needs to be borne in mind when using model animals, it really misses the point. When we study some characteristic in a model animal, all we really care about is that there really be a very close similarity between the traits at the level of anatomy, physiology and function that are of concern to us. This similarity at the appropriate level is not the same as homology. A penguin’s flipper has the same bones as a bird’s wing, and when we study the internal anatomy of these two structures, we really do see that they are derived from the same ancestral bone arrangement. But if we want to study the dynamics of aerial flight, the penguin’s flipper is a terrible model for a bird’s wing. It has the wrong shape, and it has no feathers which are essential in bird flight. If all I knew about birds wings was what I learned from studying penguin’s flippers, I would conclude that birds can’t fly. Conversely, I may learn a great deal about a structure or function by studying another structure that is only analogous. An octopus has a “camera” eye, as we do. That is, the eye has a focussing lens and a sensitive “photographic plate”, the retina, on which images are formed and are translated into nerve impulses that are sent to the brain. The octopus eye originated in evolution completely independently of the vertebrate camera eye, yet for some purposes of studying how lens defects may cause distortions on the retina it would be a perfectly adequate model system. All other things being equal, closely related species may indeed be more likely to have homologous traits, and homologous traits are more likely to be similar in various aspects than analogous traits, but there are no guarantees and some bad mistakes can be made. There is no substitute for detailed knowledge of the traits.
One of the greatest sources of error in counting on closely related species to have the “same” trait is the phenomenon of “recruitment”, the origin of novel traits by taking over some already existing organ or mechanism and turning it to new functions. That is what birds did in developing wings. They recruited what were originally front legs into flying appendages, and in the process gave up some of the functions of those legs, like running and grasping. Penguins then recruited their ancestors wings to make flippers and had to give up flying. But recruitment does not always involve losing functions, because there can be a chain of recruitment so that all functions are still maintained although carried on by other parts of the animal. The bones of our inner ear, which are used for transmitting sound, were, in our reptilian ancestors, bones of the hinge between the lower and upper jaw. Mammals recruited them for hearing and replaced the hinge bones with an elongation of the jaw bones themselves. So we can hear things that reptiles cannot, and we can still open and close our jaws. It is precisely the recruitment of parts to serve new functions, while still preserving old functions that leads to the greatest errors in using animal models for other species. In particular, as we will now discuss, it is the fallacy of primate models of human behavior.
Primate Models of Human Biology
The argument for using other primates as models for studying human behavior rests on both ordinary observation and scientific principle. First, the observation of chimpanzees and gorillas suggests to us in an intuitive way, that they are “just like us.” So much of what they do appeals to us in an every day sense as very “human.” Even the most skeptical observer is captivated by the apparent humanity of much of ape behavior and gesture. There is a gorilla in the National Zoological Park who, on being fed grapes, pieces of orange and banana, will pick up the pieces, arrange them separately on a cross bar of his cage, lie down on his back on his wooden bench, and then reach back coolly over his head and take one piece of food at a time and eat it. It is hard not to describe this behavior as “laid back.” The intuitive impression of similarity is greatly enhanced by those primatologists who spend a lifetime in the field observing and interacting with primates, who give them names, describe their “person”alities and have obvious affection for their objects of study. The intuitive impression is also reinforced by investigators who are teaching primates to communicate by means of pressing buttons on computers. Their descriptions and interpretations of the outcome of these experiments are always cast in terms of primates being able to communicate grammatical structures, of primates “saying” things in the same sense that humans say them. There is, of course, a strong prejudice in these investigators toward finding speech ability.
The scientific principle on which the primate models are based is that very closely related species will share characters. Gorillas and chimpanzees are our closest evolutionary relatives (there is some controversy on which is closer to us), but they are not extremely close. Based on DNA similarity, our common ancestor with these species was about 7-10 million years ago (paleontological evidence suggests a much longer time) so a total of at least 15 to 20 million years worth of independent evolutionary events separate us from them, half going from us up to the common ancestor, 7 to 10 million years ago, and the other half coming down to the chimpanzee or gorilla. [See figure 2] To get a perspective on the differences that can accumulate in such a period, it should be realized that nearly all the differences between cows, goats and deer, have occurred in the same period, and this is roughly the length of time separating deer and giraffes! It is often stated that our DNA is 99% similar to that in lower primates, but this figure tells us nothing. The vast bulk of our DNA, about 97%, is of no clear functional significance (so-called “junk” DNA) and it diverges between species at a constant time rate. So the similarity of this part of the DNA is simply another way of saying that we diverged from the other primates 7 to 10 million years ago. In the functional DNA, there is no proportionality between the percent similarity and similarity of function. A few DNA base changes in a gene made up of thousands of bases will, in some cases have drastic effects on function, while in other cases much larger number of changes will have no detectable effect. There are no simple rules of proportionality, and the claim that we share some percent of our DNA with some other species contains no information about our anatomical or physiological similarity to that species in any particular respect. One of the claims to similarity is concerned with the central nervous system and its functioning. To what extent is a chimpanzees brain simply a smaller version of then human brain, with a detailed structural similarity of its different parts to the parts of the human central nervous system? [See figure 3] The analysis of speech is particularly revealing. Chimpanzees and gorillas make a variety of grunting, screeching and hooting noises (“vocalization“) using their tongues, vocal cords and lips From electrical stimulation and surgical experiments it is known which regions of the chimpanzee brain are concerned with various aspects of this vocalization. Roughly the same areas in the human brain are related to speech and speech comprehension, as revealed from swdies of brain injuries, so the temptation is strong to claim that Mr. Jiggs grunts are simply the primitive form of Lincoln’s Gettysburg Address. But when the details are examined, a. much more interesting situation is revealed.
When the motor areas of the brain are electrically stimulated in monkeys, they produce grunts. When the same area of the brain is stimulated in humans they produce grunts When a region called Broca’s area is damaged in humans, various speech disorder, aphasias, result that are not motor malfunctions, but have to do with making and comprehending sentences. Broca’s aphasics cannot use words like “with,” “and,” “by,” “or,1’ and although they can move their tongues, lips, and mouths on command, they get confused when they are given the sequenced command, “Move your tongue and then your lips and then your mouth.” When Broca’s area is stimulated in lower primates, they move their tongues, lips and mouths, but they do not make any vocalizations. So a region of the human brain that is associated with speech production and comprehension, is nothing but a motor area in lower forms. Yet another area of the brain, Wernicke’s area, is concerned in humans with understanding heard speech. If the connection between Wernicke’s and Broca’s are is broken, the patient cannot repeat a heard sentence, although he or she can understand and act on it, ln lower primates Wernicke’s area is concerned with distinguishing sounds that the individual itself makes from those made by other individuals. Speech and speech communication are not simply grunting writ large.
What happened in the evolution of human beings is that certain regions of the primate brain were recruited from their simple motor functions to create quite new functions, the functions of speech. At the same time it was possible for these regions to hold on to their old functions, probably because the brain grew so much larger and so had nerve connections to spare. There is no homologue of speech in chimpanzees, certainly not grunting. Speech is a novel function whose anatomical basis was created by the process of recruitment of parts that could be spared, just as our inner ear apparatus has no homologue in fish, reptiles and amphibia, but is a totally new machine, made up of parts that could be spared from the jaw hinge. There is no reason to suppose that a study of the brain or behavior of a gorilla will tell us anything about human language.
A great deal has been made of the experiments in which lower primates have been taught to form phrases out of available bits and pieces, using computers to help them “say”, “Me pour water.” But dolphins can be taught to distinguish the sentence, “Put the ball in the ring”, from the sentence, “Put the ring on the ball.” Yet dolphins have not had a common ancestor with humans since close to the origin of the mammals, and we are no closer to them than we are to rats. Whatever it is that animals are being taught to do when they are taught to communicate with us by computer, it is not homologous with speech.
When we move from speech communication to such behavioral traits as aggressiveness, temperament, affection, sexuality, sociality, etc. the problem of similarity becomes yet more acute. Is what is labelled as “aggression” in gorillas the same in any interesting and informative way to “aggression” in humans. The immediate difficulty is that “aggression” in humans is not clearly delineated. The individual feeling of “aggression” one may experience when someone has been particularly nasty or rude is not the same as the “aggression” committed by Hitler in 1939 against Poland. Political and national aggression is an expression of economic and political self-interest of a group, and does not arise from the individual aggressive feelings against other individuals. If it did, there would be no need of conscription and war propaganda. Is the intellectual “aggressiveness” of an ambitious professor, the same as the “aggressiveness” of antisocial violence? Perhaps it is, but the fact that it is not obviously true shows why comparison with aggressiveness of other species is not helpful.
Sometimes it is said that basic behavior has been established in species by the same force of natural selection, so similar species will have similar neurological and hormonal bases for similar behavior. By studying sexual behavior in chimpanzees which was, after all, selected in chimpanzees in order to guarantee offspring, we will understand sexuality and its variation in humans, since it has the same evolutionary basis. It is precisely in the study of sexuality that the error of the animal model approach becomes clearest. Human beings, have speech, a consciousness of history and of culture, an ability to manipulate the world technologically, and to transmit that abstract technological and cultural knowledge. These are evolutionary novelties that are shared by no other species. To keep matters in perspective, it must remembered that although primatologists may write books about teaching chimpanzees to speak, no chimpanzee has ever written a book about teaching primatologists. One of the remarkable consequences if human’s unique technology and culture is that there has been a complete transformation of sexuality from its origin as a reproductive mechanism. Human beings have sex without reproduction, reproduction without sex, and pregnancy without conception. Sex is used as a deliberate form of affirmation, of aggression, of domination, of commerce, of producing pleasure and inflicting pain. Human sexuality is a novel function that has been recruited from a biological impulse, and in the recruitment it has been freed of the biological constraints of reproductive sex.
The example of sexuality illustrates two general trends in human evolution and history that are particularly important for understanding the relevance of animal models. The first is the conversion of internally biologically constrained functions into functions that depend heavily on individual life histories and social environment. What may be “hard-wired” into the anatomy and physiology of lower primates will be much more subject to environmental history in humans. Suppose, for example, we give an injection of testosterone to a male gorilla and observe a stereotyped set of sexual displays and behavior toward females, including mating attempts and aggressiveness. In addition we record the changes in its metabolic rates and analyze its blood for hormones, sugar levels, etc. If we now give an equivalent dose to a human male there may indeed be an initial similarity in the pattern of hormonal responses and feelings of sexual arousal, but these are immediately modified by the conscious and unconscious mental responses of the subject. He may feel guilt, mental anxiety, a conscious desire to repress illicit feelings of sexuality, anger at the experimenter for creating conditions of mental and physical stress, or any of a variety of attitudes that accompany strong sexual urges in an inappropriate milieu. He may even fail to perceive his reaction as sexual at all, depending on his psychic and social history. But all of these mental reactions produce hormonal and metabolic consequences directly because the nervous system and the set of interacting hormones secreted in the body are not physiologically independent, but form what is called the “neuro-secretory” system in which nervous impulses trigger hormonal releases, which in turn signal other hormones and also affect the state of the nervous system. Thus when the human subject’s metabolism and blood contents are analyzed, not to mention his manifest behavior, there will be unpredictable differences from the gorilla, and the variation from one human male to another will be large. In general, metabolic states are in an effective interaction with mental states. Whatever the mental states of lower primates may be and whatever factors determine them, they are not the consequences of reading books, going to church, watching television, being given sex education in school, participating in puberty rites, or any of the other complex historically and socially determined factors that form human mental attitudes about sex. Apes may or may not be simply a sack of hormones, but humans certainly are not and their sexuality cannot be understood by using lower primates as models.
The second consequence of human evolution has been the elimination of certain biological limits to function, as a result of the novelty of language and technology. No human being can fly by flapping his or her arms, because no lift is generated by appendages with the shape and musculature of a human’s appendages. Thus, our anatomy seems to place an absolute constraint on our function. Yet human beings do fly, as a result of inventing the technological means to do so and having created complex social institutions like airports, petroleum refining companies, and flight instruction. Nor is it society that ifies. Individual human beings leave the ground in New York and land in San Francisco, but as a consequence of uniquely human social interactions. No chimpanzee will ever fly except at the will of human beings, because they share with humans the same limitations of anatomy of their arms, but fail to share with humans the unique anatomy of their brains.
The example of flying can be repeated over and over. Human beings share with crows the inability to remember accurately how many objects they have instantaneously glimpsed, if the number is greater than about seven. Yet individual human beings can recall at will all the details of the National Budget for 1934, by the simple expedient of looking it up in the Archives. Again, social institutions and technology abolish limits on individual function that would appear when the human organism is viewed outside a social context, or through animal models that lack that context. Human technology and culture are the consequences of a unique biological property of the human species, the detailed structure of the central nervous system. Any study of some other species that lacks that biological property, is the use of a model organism that lacks what is an essential and overwhelmingly important feature of the biology of Homo sapiens. With luck, some properties of lower primates will carry over well enough, but this cannot be the rule. That is why only humans are a good model for humans. Unfortunately, for many purposes, even that will not work, because the immense variation in culture and history, and the genetic differentiation among individual human beings means that it is difficult to generalize from an experimental subject to the generality of the species in many cases. Only the study of large and heterogeneous experimental populations of humans will give an adequate model for generalizing to the human species at large. If we want to study the human lot we need to study lots of humans.
References
Lewontin RC. An estimate of the average heterozygosity in man. American Journal of Human Genetics 1967; 19:681-685.
Lewontin RC. The apportionment of human diversity. Evolutionary Biology 1972;6:38l-392.
Lewontin RC. Single and multiple locus measures of genetic distance between groups. American Naturalist 1978, pp 1138-1139.