WHY ‘NOT’?

Among the most beloved conversation topics of witty dinner parties are those words that only exist in some languages, while other languages are forced to express the same concepts by complex circumlocutions. A famous case is German schadenfreude, or ‘the pleasure derived by someone from another person’s misfortunes’ (according to the New Oxford American Dictionary). So varied is human experience that different peoples around the Earth have different words to talk about what matters to them most. At least, this seems to be the lesson of dinner parties.

But the opposite situation is also enticing to the scientifically curious mind. There are words common to all languages, regardless of the varieties of human experience. In some cases, an explanation is ready to hand. Words like ‘water’ and ‘tree’ are found nearly everywhere, and it’s easy to imagine why: referring to such things is often useful and always easy, water and trees being familiar parts of our environment. But another consistent finding are some words of interest to logicians: connectives (not, and), quantifiers (every, most), and modal and temporal operators (perhaps, always). These words are crucial to the reasoning abilities that make us special in the animal kingdom. So logical words are certainly useful. But not all that is useful in fact evolved: opposable thumbs are useful, but we’re the only ape that has them. And logical words don’t refer to anything we can easily point to, if they can be said to refer to anything at all.

Why, for instance, do we have the word ‘not’? Basic forms of negation exist in every human language, but it’s doubtful that other animals can deny (Schlenker et al. [2016]). There appears to be a gap in the natural order of development, but natura non facit saltum. Just to rule out a ‘jumpy’ view: Grammatical complexity, especially as exemplified by negation, did not arise in the space of a generation. It is very implausible that, perhaps around the time of Homo erectus, some of our more ape-like ancestors should suddenly have had children who could warn them, ‘If you don’t let me play, I’ll throw a hissy fit!’.

In our article, we outline an adaptationist view of negation based on a signalling model of communication. Adaptation is an evolutionary process by means of which a complex whole (such as a biological organism or a language) is ‘fitted’ to its environment (its habitat or an information space), particularly through the development of new features (such as organs or words) whose functions are responses that meet environmental challenges. A good example of adaptation in biology is the eye, developed for vision. It’s plausible that an organ whose function is the rapid detection of changes in light intensity is adaptive, because we find eyes developed independently in very distant animal species, such as in us and the octopus, whose most recent common ancestor lived around 750 million years ago.¹The story is nicely narrated by Peter Godfrey-Smith ([2016]).

So it is reasonable to suppose that negation is adaptive, since it is present in all natural languages, despite completely distinct developmental pathways. But this initial plausibility does not mean that the case is closed, for adaptation may mean many things. When someone says, ‘This is water’, she is talking about water; but when she says, ‘No water over there’, it would be odd to claim that she is talking about ‘negativity’ or ‘absence’. There’s no such thing around us. Instead, we argue that negation is not a direct adaptation, like ‘tree’ or ‘water’, but the result of co-option. That is, even if negation is commonly used to state that something is not the case (to say that something is false or non-existent), its origins are probably traceable to something else, and to some other purpose. What we need is a story in which negation is useful for communication, that does not appeal to implausible metaphysical ‘spooks’ in our environment, and that is backed by evidence we find in world languages.

Imagine a primitive interaction. There is a sentry who knows if predators or food are to be found around the hill. She must say something to her mate who can’t see behind the hill, but what she says depends on what’s going on. She can say, ‘Food over there’ or ‘Tiger over there’, if there is food or a tiger, respectively. But the sentry also has an incentive to deceive: to encourage her friend to run away for safety while she grabs all the food. But the companion is not hopelessly exposed to the sentry’s tricks: the friend can decide if she wants to trust the sentry or not. If she trusts the sentry, she will accept the message and act accordingly; if she doesn’t, she will reject and ignore it.²This naughty behaviour and reactions to it are documented in capuchin monkeys by Brandon Wheeler ([2009]). Information will flow through in the first case, but not in the second.

This game of cooperation or conflict—in which the speaker may be truthful or not and the listener may accept or reject the message—describes many familiar interactions. At a certain level of abstraction, it’s an ordinary conversation: the speaker decides between telling a truth or a lie, and the listener decides on whether to trust the speaker or not. It’s a double choice that is very common in our social lives.

We play this game over and over. We talk about tigers, food, water, trees, and so on. The information space for communication has virtually no limits. There is pressure to maximize informativity, for listeners would not accept too many ambiguous messages that do not make clear whether to run or feed or anything else. Hence more and more new words should be introduced: ‘tiger’, ‘food’, ‘water’, and so on—everything we need to describe different scenarios. On the other hand, there is pressure to minimize complexity, for neither speaker nor listener can exchange information in a language that has as many words as there are beliefs. That would be utterly impractical, because we don’t have an infinite memory.

A common view is that languages balance these two opposing forces by committing to memory a small stock of words, and then combining them in a rule-like fashion to express all information that is worth conveying. And as the game between speaker and listener is defined by the choice between acceptance and rejection, there is a device that could prove extremely advantageous: a simple word that the speaker can use to anticipate rejection. ‘Not’ can be a flag indicating that the message so marked (‘There is no water there’) should be accepted any time its unmarked alternative (‘There is water there’) would be rejected, and vice versa. The difference between A and not-A emerges in language to track the duality between acceptance and rejection. At the small cost of one extra word, the speaker can stay on top of the game of communication, avoiding ambiguity and memory overload. For she can now choose to say, ‘Food is there’, in case the listener would accept this claim, and ‘Food is not there’ if anything else she could say would be rejected. A primitive form of negation emerged as a device to flag signals so that A is accepted just in case not-A is rejected and vice versa. Negation is a tool for speaker and listener to exchange information to their mutual satisfaction in virtually all cases.

This is not the full story of how ‘not’ came to be and we are still far from the combinatorial possibilities of negation. What’s missing is a more complete picture of the evolution of grammar, but that is for another day. Yet there is evidence for the proto-negation we have described. In some languages there are markers of denial that do not combine freely with other words, but which can only be attached to sentences to mark an accept–reject reversal. An example is the Italian word ‘mica’, for mica-A is true just in case A is false, and vice versa. Although one cannot say, ‘If you mica let me play, I’ll throw a hissy fit’ (mica under if; see Frana and Rawlins [2019]). The limited combinatorial possibilities of mica may well suggest that it’s the missing link: it occupies a place in the history of negation prior to its full incorporation into grammatical structure. It does not have the complete functionality of negation, but does prove useful as a tool to prevent rejection.

In our view, the origins of negation are found in our practice of exchanging information. This points to a bigger picture in which logic in language and thought derives from the actions we perform. A cooperative speaker could secure efficient information transmission and avoid straining memory by means of the word ‘not’, which in combination with any message expresses the alternative to accepting a message—that is, rejecting it. And so the speaker can stay ahead of the game of cooperation and conflict, capable of expressing in words the choice the listener has to make in life. ‘In the beginning was the deed’, says Faust. In the evolution of negation, the deeds were the two incompatible actions of acceptance and rejection. That’s where not’s meaning of opposition comes from.