In 2003, an American startup from Iowa Primate Programming Inc. (PPI) announced that the company's programmers - chimpanzees, baboons, gibbons and gorillas - will do any work of the IT team, only cheaper. Chimpanzee Mitch, for example, took a JSP course and specializes in Crystal Reports, VB3, VB6 - 93 cents per hour. Orangutan Cloyd has completed his ASP.NET Training and offers XML und Web services at $ 1.10 per hour.
Nice try guys, but you can't see the primate IT pros taking over the world. Monkey start-up ended in the same way as 90% of all startups... Can monkeys really be taught to program? The question is not as absurd as it might seem. To understand how to answer it, you have to touch on such areas of knowledge as the theory of computation, theory of formal languages and evolutionary linguistics.
Communication with animals is more likely no than yes
The close links between natural and formal languages are beyond doubt. Maybe we should start by teaching the primate to keep a simple conversation? And then it comes to programming. The first difficulty arises immediately when trying to talk to any animal - they produce sounds differently. Unlike humans, animals have almost no control over spoken sounds and do not exchange specific details in the process of communication. For example, the chirping of a bird in general means nothing more than "I'm fine, life is beautiful." Monkeys' vocal apparatus is also not adapted to reproduce the sounds of human speech. (1)
But maybe it's worth trying non-verbal communication? Many researchers have tried more or less successfully to teach primates to express their thoughts and feelings through gestures. One of the most famous subjects of this kind of experiments is the chimpanzee Nim Chimpsky, who, under the guidance of linguists at Columbia University, mastered more than 125 gestures. As a result, the project was recognized as unsuccessful: He did not master the language to the extent that is inherent in a person. He could use already learned phrases with the help of gestures and repeat new ones after a person, but he did not form them himself according to the rules of syntax and grammar. His longest sentence was something like " Give me an orange to eat an orange to me to eat an orange give me you ." All his deeply philosophical statements can be read here .
Nim Chimpsky talks to his teachers
Animals as artful imitators
Noam Chomsky, an American linguist, after whom the talkative primate Nimes was named, coined the term "generative grammar" (or generativism ), meaning a set of rules by which a grammatically correct sentence is formed in any language. According to Chomsky's views, the grammatical rules underlying any language are innate and inherent only in humans . There is no need to dwell on these ideas in detail, since they have already been described on Habré, for example, here . In other words, Chomsky believes that generative grammar as a general set of rules is already built into the human brain from birth.... This innate human ability to speak any natural language and the ability to use formal languages, such as programming, are closely related. So far, no experiment has confirmed the ability of a humanoid animal to master generativeism. Does this mean that we will never see the monkey conjuring over the program code?
To test this experimentally, you have to overcome many difficulties: one of them is the animal's desire to copy patterns and memorize ready-made solutions without thinking about them. Physicist Richard Feynman described this problem very accurately in his book "You, of course, are kidding, Mr. Feynman":
“In 1937, the zoologist Paul Thomas Young . : , , , , . , , . , . , . — . , , , , . . — . , . … . , , , that makes the floor right in front of her . The scientist covered the floor in the rat corridor with sand, and only then the rats stopped repeating routine actions. Young's merit, among other things, is that he showed how carefully all factors must be taken into account when organizing an experiment of this kind with animals. "
And that's just rats. The monkey, in order to get food, will resort to complicated chains of routine actions. The experimenter's task will be to make sure that the animal is really solving the problem, and not how Nim uses his memory and environmental clues to find a suitable answer.
A hypothetical experiment with a palindrome machine
Let's imagine that we have technical and other possibilities to conduct such an experiment. It will be based, for example, on the language of palindromes - strings that can be read equally in both directions - as one of the simplest basic languages of context-free grammar. If an animal does not learn to build palindromes, the chances that it will be able to write a program in the simplest language tend to zero.
These are all palindromes - language
We have a machine with buttons, each button has some kind of symbol - let's say, fruit icons: orange, banana, pear, peach, and so on. The machine is programmed in such a way that when it is composed of palindrome symbols, it gives out food. By pressing successively "banana", "apple", "peach", "apple", "banana", you can earn lunch, but the combinations "banana", "apple", "apple", "peach" will not give anything. You could think of a better reward for long palindromes. In addition, we also need a line termination signal - for example, a backlight and a sound signal, which will turn off the buttons for a few seconds if the composed key combination is not a palindrome. Finally, the most important thing is to explain to the animal what we actually want from it. And this may just be the main problem - is it necessary to explain in advance to the primate,what is a palindrome? Wouldn't such an experiment be just a test of this understanding, and not a way to teach a monkey to solve assigned problems? The best way to test all these guesses is to conduct an experiment in reality. When someone succeeds, let us know.
What prevents animals from writing programs
Communication is not the most difficult task, which we and animals solve in different ways. Face recognition, for example, is an even more difficult task, which, nevertheless, both of our species can handle. But cars are much inferior to us in this. Computer programs are quite successful in creating meaningful texts according to the rules of natural languages, but with face recognition everything is not so radiant. Probably, animals are still capable of solving problems comparable in complexity to writing programs, it's just that these abilities of theirs evolved in other, non-linguistic areas. For example, pigs, like humans, have an interest in video games and are able to enjoy them. And monkeys can mentally control rather complex devices.
The plot of "Peppa Pig", apparently, is taken from real life.
We cannot evaluate the mental abilities of animals only by the criteria of communication and mastering formal and natural languages, neglecting their other abilities.
There are at least two things that allow us to program. The first is the ability to think, which covers a wide range of reasoning, from understanding cause and effect to mathematical proof of complex formulas. Obviously, this ability, albeit in a truncated form, is present in animals. The second thing is the ability to communicate, including the ability to make abstract, probabilistic judgments about the world. The absence of these key properties does not give animals a chance to create their own programs. They have enormous mental resources to memorize actions and compare with existing patterns, but they have not yet applied these capabilities to complex reasoning, similar to human ones. And it's not a fact that this will happen someday.
“Baboons were shocked and confused by the Java IDE, SunONE and Visual Age. None of the baboons ever managed to master the Java programming language. However, most baboons have learned the Visual Basic 3.0 programming language with ease. The baboons were able to independently change the program settings and even edit the parameters of the file attributes. Some researchers have argued that Visual Basic is too easy for baboons. They insist that the monkeys take another course in Java programming. However, for most baboons and chimpanzees, any Java component, according to scientists, causes stress ” (2) .
Alas, while such statements are somewhere in the middle between the trolling of the scientific community and the original PR campaign of a monkey startup.
Article preparedcrazytosser00 and talkjulyabased on materials from theorangeduck. Thanks to the author of the site for his ideas.