Chapter 7

The Battle of the Giants

Do we invent because we systemize, as I have argued, or because we have language? Which of these two powerful devices in the brain is the best explanation for human invention?

No one will deny that an obvious big difference between humans and other animals is that we have language and they don’t. And this is not a new idea: the nineteenth-century linguist Friedrich Max Müller wrote that language is the “one great barrier between the brute and man,” and Darwin, while acknowledging that other animals have forms of communication, discussed how they lacked the complexity of human language.¹ Contemporary writers have also highlighted the importance of language in enabling flexible thought. For example, paleoanthropologist Steve Mithen has argued that the inventions of art and sculpture 40,000 years ago reflected a shift from the modular mind of our hominid ancestors to the more integrated mind of modern humans, via language.²

It is easy to imagine that, once we evolved language, we could entertain hypothetical ideas of an if-and-then nature, enabling a capacity for invention.³ It would be nice if the battle between these two big contenders as theories of invention could be settled by chronology—which came first, the Systemizing Mechanism or language? Sadly, this option is not available to us because the Systemizing Mechanism and language likely originated around the same time, around 70,000 to 100,000 years ago. So a different way to resolve this question is to use the principle of parsimony: Does invention need language, or can invention happen in the absence of language? Expressed differently, can the Systemizing Mechanism account for invention without needing language?

The word language is actually not that helpful, because it’s an umbrella term. So, let’s start by breaking language down into some of its key components, beginning with speech. We can rule out speech as being sufficient for the capacity for invention because the physiological apparatus for speech was present in our hominid ancestors at least 600,000 years ago. We know this because the hyoid bone in the front of the neck—thought to be key for speech and articulation—existed (in modern human form) in Homo erectus and also in Neanderthals.⁴ In contrast, as we saw in chapter 5, generative invention only appears in the archaeological record from about 70,000 to 100,000 years ago. So, speech alone can’t explain the flowering of human invention during the cognitive revolution.

What about communication, which is different to speech? Many species have communication systems even though they don’t have speech. For example, bees do the “waggle” dance to signal to other bees where to find the pollen, and birds sing (often at dawn) to attract opposite-sex members of their species to find them in the mating season.⁵ Vervet monkeys make alarm calls if they spot a tiger, a snake, or an eagle, causing other vervet monkeys to climb a tree, look at the grass, or look up at the sky, respectively, depending on which predator has been “announced.”⁶ And yet, as we saw in chapter 6, we don’t see other species inventing. So, having a communication system per se also doesn’t seem to be sufficient for the capacity to invent.

So, is there some other aspect of language that might have enabled humans to invent? What about recursion, which, according to linguist Noam Chomsky, is the unique feature of human language?⁷ (Recursion is when a procedure includes the procedure itself, and which can repeat indefinitely.) I’m going to explore recursion in a bit more depth because it’s so remarkable, and because it’s a good contender for how we invent.

One example of recursion is “nesting.” If you take the sentence “Alex has a red car” and nest within it “whom you know very well,” you can make the sentence “Alex, whom you know very well, has a red car.” The remarkable power of recursion is that we can keep nesting, building up more and more layers, like Russian dolls. So, the phrase “that is parked there” can be nested into our previous example, to become “Alex, whom you know very well, has a red car that is parked there.” Such recursion can go on, in theory, indefinitely. One could easily imagine how, once humans had this capacity for taking phrases, nesting them inside others to build ever more complex linguistic structures, this could have been very useful for the more general capacity to invent new things. Essentially, these are effectively the building blocks of a sentence (the clauses and subclauses of grammar).

A second example of recursion is how, with a finite number of words, we can create an infinite number of sentences. Neurobiologist Andrey Vyshedskiy invites us to imagine having a language with 1,000 nouns, including “bowl” and “cup.” He gives the example of adding the spatial preposition “behind” to our 1,000-word vocabulary. Suddenly we have a huge number of three-word phrases, such as “bowl behind cup,” or “cup behind bowl.” In fact, the number of distinct mental images we can refer to increases from 1,000 to 1 million (his calculation is 1,000 × 1 × 1,000).

Now Vyshedskiy invites us to imagine adding a second spatial preposition, such as “on.” We can now generate a huge number of five-word phrases, such as “cup on plate behind bowl.” Suddenly we’ve got a lot more we can talk about than just pointing at a bowl and saying “bowl.” In fact, he calculates, by adding these two spatial prepositions to our 1,000-word vocabulary, the number of distinct mental images we can refer to jumps to 4 billion (1,000 x 2 x 1,000 x 2 x 1,000). This is another remarkable example of recursion. Vyshedskiy calls this powerful increase toward an infinite number of sentences “magic.”

So, is the recursive property of language a rival theory of human invention? I don’t think so, for several reasons.

First, recursion is not just found in language—it’s also a critical feature of music.⁸ Given the key role of systemizing in the invention of music, as we discussed in chapter 5, this suggests that the Systemizing Mechanism likely enabled recursion, not the other way around. Consider how if-and-then reasoning could handle the earlier recursion example: “if I take the phrase ‘Alex has a red car,’ and nest within it ‘whom you know very well,’ then the phrase becomes ‘Alex, whom you know very well, has a red car.’”

Secondly, people who lose language as a result of a stroke, or who never developed much language in the first place, can still be remarkable musicians.⁹ This again suggests that you don’t need linguistic recursion, but you do need the Systemizing Mechanism, to be able to invent music.¹⁰

Thirdly, a human mother can mesmerize her infant with varying rhythmic patterns during simple games like “Pat-a-cake, pat-a-cake, baker’s man,” long before her infant can handle linguistic recursion.¹¹ This suggests that infants can pick up if-and-then patterns without linguistic recursion.¹² Such rhythms can include nested ones, where different sequences can be nested within each other. For example, a mother could chant the “pat-a-cake” rhythm, then switch to the “incy-wincy-spider” rhythm, and then switch back to the “pat-a-cake” rhythm, and an infant would be able to follow the rhythm simply by using if-and-then reasoning.

Let’s turn to a final key feature of human language, syntax. Syntax is remarkably powerful. It allows us to change the phrase “dog bites man” to “man bites dog,” each with a very different meaning, just by breaking the phrase down into its individual units (three words, in this case) and then changing the sequence of the first and last words (the subject and the object). It is easy to imagine how syntax allowed the mind to create new images or ideas, which is essentially invention.¹³

But again, I don’t see the capacity for linguistic syntax as a challenge to the Systemizing Mechanism theory of invention, because syntax is also a property of the Systemizing Mechanism. Consider how the Systemizing Mechanism runs the if-and-then algorithm: if a numerical sequence is 1-2-3, and the first and last digits are swapped around, then the numerical sequence is 3-2-1. Word sequences can be pushed through the same if-and-then pipeline: if the phrase is “dog bites man,” and the first and third words are swapped around, then the phrase becomes “man bites dog.” Nor do I see that syntax is essential for invention, but, according to my theory, if-and-then reasoning is a requirement. Without it, there would be no invention.

So, if-and-then thinking conferred on humans 70,000 to 100,000 years ago the capacity to rearrange variables within any system: “if I take a straight bladed tool, and change its shape to a curve, then it can become a fishing hook.” The Systemizing Mechanism allowed—and still allows—powerful magic: infinite invention. As a side benefit, if-and-then thinking enabled us to perform operations like recursion and syntax. These in turn would have transformed a simple language into a complex one. Undoubtedly it was a two-way street, with language facilitating if-and-then reasoning by allowing us to put our new ideas into words, and then play with words to come up with new ideas.

But the very existence of autistic savants, some of whom have very minimal language but who are hyper-systemizers and who can invent, suggests that systemizing and language are independent of one another.¹⁴ Two beautiful examples of such autistic savants are Nadia, an autistic girl who could draw horses from any perspective, when she had almost no language; and Stephen Wiltshire, who could draw buildings with remarkable accuracy and from any perspective, even as a child and when he had very limited language. (Both Nadia and Stephen eventually developed some language.)

In sum, in my view, language is powerful in its own right, but is not a rival explanation of human invention.¹⁵

Four other psychological theories have been put forward to explain the capacity for invention, and I’m going to deal with them briefly.

The first is that we invent because we can integrate two ideas into one new one. Vyshedskiy argues that this is the role of the lateral prefrontal cortex in the brain. According to him, this is what enabled early humans 40,000 years ago to take two separate concepts (“man” and “lion,” for example) and combine them into a synthesized concept (“lion-man”) to make a sculpture of such a fictional entity. Vyshedskiy suggests that only the lateral prefrontal cortex can combine objects from memory into a novel mental image. He calls his alternative theory of human invention “prefrontal synthesis.”¹⁶

However, the lateral prefrontal cortex is involved in a lot more than integrating two ideas into one new one. And this theory isn’t really an alternative because integrating two ideas is just an operation within the Systemizing Mechanism (it’s the and in if-and-then). Thus: “if I take (the idea of) the top half of a lion, and I attach it to the bottom half of (the idea of) a man, then I have (the idea of) a lion-man.” The power of the Systemizing Mechanism is that it can perform this and any other operation on input to produce an invention, whether the input is real, or is an idea, a word, a picture, or a model (such as sculptures representing objects), to produce fictional entities (like Spider Man).¹⁷

The second theory, briefly, is the idea that we became capable of invention because we could think symbolically.¹⁸ Archaeologist April Nowell has proposed exactly this. One meaning of the term symbolic thinking is the capacity to let one thing stand for another, or to imagine that one thing stands for another, as in algebra when we say, “If x represents the number of apples in this box,” or in drawing when we say, “If this big circle I have drawn in the sand represents the Earth.” So, the first meaning of symbolic thinking involves hypothetical thinking.

This was undoubtedly a huge step forward in humans’ cognitive power—to be able to entertain thoughts about hypothetical situations—and it’s unclear if any other species is capable of this. But hypothetical thinking is not really a challenge to the Systemizing Mechanism theory of human invention because hypothetical thinking is the if element in if-and-then systems-thinking. In addition to hypothetical thinking, someone also had to use if-and-then thinking to see how ochre could be used as paint, or to make a tool like a paintbrush, or a chisel to carve a sculpture. Systems-thinking (if-and-then reasoning) had to come first.

A second use of the term “symbolic thinking” is what psychologist Alan Leslie calls the capacity for meta-representation. During meta-representation, a proposition (“the moon is made of blue cheese”) is prefixed by a mental state (for example, “I imagine that…”). The result is a sentence: “I imagine that the moon is made of blue cheese.”¹⁹ This statement can be true even if the statement “the moon is made of blue cheese” is obviously false. Meta-representation enabled both theory of mind (imagining someone else’s thoughts) and self-awareness (thinking about one’s own thoughts), and it is one part of symbolic thinking (pretending or imagining that one thing represents another).

Meta-representation could be argued to be key to the capacity for human invention because it would enable the thought “I imagine that this hollow bone could be used to make sounds.” Meta-representation therefore allows us to imagine fictional possibilities and can explain our capacity for “make-believe.” Make-believe is fun (it allows us to joke around and pretend a banana is a telephone) and socially invaluable (it underpins our ability to have a theory of mind, so is part of the Empathy Circuit). Being able to think “I imagine that x” must again have been a huge step forward in cognitive terms, and we have no evidence that any other animal is capable of such a thought.

But by itself, meta-representation cannot explain the capacity to engineer a new product. You can imagine or joke around as much as you like, but actually engineering something to understand technical implementation, still requires if-and-then systems-thinking. This is not to diminish the huge importance of symbolic thinking—as part of the Empathy Circuit—for art, language, and thought, but symbolic thinking does not replace the need for systemizing in explaining human invention.

A third theory comes from Yuval Harari, who argues that human invention is possible because we are the only species that can think about collective fictions (like religion, a limited company, or money).²⁰ He is of course right to underline how unique humans are in this respect, and how powerful such collective fictions can be: collectively sharing the same fictional belief can coordinate the activity of thousands or millions of people. As I write this, I am in awe of how, by the last week of March 2020, virtually all 7.6 billion people on the planet, believing a lethal but invisible virus was all around us, stayed home for several months, leaving the whole planet deathly silent and devoid of people in public places.²¹ Such is the power of a collectively held belief (in this case, not a fictional one, but based on hard evidence of its reality) to mobilize large numbers of people into coordinated action.

However, again, we can think about and share fictional concepts as much as we like, but without a Systemizing Mechanism, we aren’t going to be able to implement our ideas at a technical level. Fiction-thinking, collective or otherwise, only gives us the if part of if-and-then reasoning. We also need the and part, typically a causal operation, and as we saw in chapter 6, non-human animals don’t appear to understand causality. And we need the then component, which allows us to see the results of observing or experimenting with causal operations. There is no convincing evidence that other animals can systemize the whole if-and-then pipeline.

The final psychological theory is that we can invent because we have a bigger working memory.²² Archaeologist Thomas Wynn and psychologist Frederick Coolidge, who proposed this theory, define working memory as the ability to hold something in mind in the face of distraction. They argue that for humans to have designed and used traps, for example, they must have needed working memory: you set the trap, and then you watch and wait, or come back later, to see if it worked. Ironically, although the word “memory” usually refers to past information, the term “working memory” is also used in relation to implementing a future plan. That’s because you have to be able to remember the steps of the plan.

There’s no doubt that the human capacity for holding many more steps in mind must be a big advantage, but did increased working memory per se lead to our capacity to invent? The answer must be no, for several reasons. First, an animal can have a good working memory but still lack the capacity to invent. For example, squirrels have an excellent working memory for where they buried their nuts before the winter, yet don’t invent in any generative way. Some have claimed that crows and apes can even think about the future, although this is contested, yet they also don’t invent in a generative way. So, invention entails more than just working memory.²³

Making a trap, a good example of planning, clearly requires way more than working memory. At a minimum, it also needs systemizing: “if I attach a spring to a metal bar, and trigger the spring, then the metal bar will snap shut.” Or, “if a mouse nibbles the cheese, and this triggers the spring, then the metal bar will snap shut on the mouse’s head, killing her in a split-second.” Making a trap is also a sign of the capacity for deception and so entails cognitive empathy, or theory of mind (part of the Empathy Circuit), in that theory of mind is needed to appreciate that the mouse won’t know what’s about to hit her, or won’t appreciate how the spring mechanism works. But systemizing is needed to design the spring mechanism in the first place.

In summary, although these four psychological processes are contenders for explaining human invention, and undoubtedly helped the whole process of invention, none of them replaces the need for a Systemizing Mechanism, and none of them alone would lead to invention. Furthermore, part of what is needed is a theory not only of how we invent, but why we invent. Recall that Edison was inventing for the pure pleasure of inventing. He worked on many of his inventions, not to meet an unmet need, but just to see what happens and what’s possible. The Systemizing Mechanism is what drives curiosity.

We also need to briefly consider some alternative theories of invention that focus on evolutionary changes in the human body. Some have proposed our upright posture and our overall brain size as contenders, but Homo erectus also walked upright, and Neanderthals had an even larger brain than ours, and yet neither of them came close to our capacity to invent generatively.²⁴ Another theory focuses on humans’ opposable thumbs, which allow for more precise fine motor control, including a precision grip and a power grip, undoubtedly advantageous in advanced tool use. (Think how you grip an overhead rail in the subway compared with how you manipulate chopsticks.) But opposable thumbs can’t explain how we can invent, since Homo habilis, all Old World monkeys, and all the great apes have opposable thumbs and yet these other primates don’t invent generatively.

Finally, others have argued that our long childhood must be relevant. Undoubtedly this had an impact on our learning capacity: a protracted childhood means that human infants are born at a relatively more immature stage of development, so a bigger fraction of our knowledge is the result of experience rather than genetic pre-programming, increasing our behavioral flexibility. But a longer childhood per se doesn’t automatically lead to a capacity for invention.

A final challenge to the Systemizing Mechanism theory of human invention might come from archaeological evidence that challenges the date of the cognitive revolution: Are there not apparent inventions that predate the idea that the cognitive revolution occurred 70,000 to 100,000 years ago?²⁵ The evidence of burial, the existence of perforated shells, and the use of pigments go back hundreds of thousands of years and may be cases of invention that predate modern humans. However, archaeologists have argued that these may not meet the criterion for being genuine inventions, largely because they are one-off instances and are open to other interpretations.

So, for me, none of these alternative proposals—psychological or physical—are sufficient to explain our remarkable capacity for invention. If we just have the if (as in hypothetical thinking), that doesn’t get us to invention. If we just have the and (as in the concept of causality), that too doesn’t get us to invention. Equally, if we have if-then, that doesn’t help us understand how to invent, only that objects or events can change. To invent, I argue, we needed the whole process of if-and-then reasoning. Invention just can’t be done without the Systemizing Mechanism.

I’ve made the argument for why the Systemizing Mechanism is necessary for invention, and for it being partly genetic. This means hyper-systemizing qualities can be passed down from parents to their children. I’ve also presented evidence that the genes for systemizing partly overlap with the genes for autism. Indeed, this genetic overlap was one of the intriguing connections that we glimpsed at the beginning of this book, between the minds of hyper-systemizers such as inventors and the minds of autistic people, both of whom are drawn to seek if-and-then patterns in the world. This leads to a very specific prediction: that hyper-systemizing parents are genetically more likely to have an autistic child. To test if this is true, we need to observe what happens when hyper-systemizers breed. And the perfect opportunity to do so is found in places like Silicon Valley, where hyper-systemizers flock to work, then meet and start to make babies. It’s a natural experiment.