On mental models and the algorithmic brain

For every action there is a reaction.

This is a fragment of Newton's third law of motion, and it is also the formula for one of our most intrinsic mental models. Human beings possess inherent mental models that are like formulas applicable to different situations. The model I am referring to here is the one that leads us to think that a phenomenon was caused by an immediately preceding action. For example: I saw a black cat and was then run over, so the next time I see a black cat, I will know that something bad is going to happen. This model of thought is undoubtedly the reason deities were invented during the early years of civilization, as well as the reason superstition developed in general. Human beings try to answer many questions, and this model provides an answer.

There is a variation of the preceding model that consists of believing that, if we perform an action, we will cause a phenomenon to repeat. Consider the example of someone who, a few hours after arriving home from work, discovers that their keys are missing. They know the keys are in the house because they used them to get in, so they begin searching and, after a while, find them on the sofa. This person is clever and realizes what happened: upon arriving home, they sat on the sofa and the keys slipped out of their pocket. The next time this person loses their keys inside the house, where will they look first? Perhaps out of ten times that the keys go missing, they will find them on the sofa half the time. This has saved them a great deal of time because it was the first place they looked. It is like a mental algorithm, so we can say that the model works. It is true, the model works, but not because the keys will be found every time someone looks on the sofa. It works because this person always sits on the sofa after coming home from work and wears somewhat loose trousers, loose enough for the keys to slide onto the sofa. In other words, finding the keys is not a reaction to the action of looking on the sofa. The model stops working when we believe that it is.

The moment we treat the model as a law, it stops working. And speaking of laws, we can now return to Newton's third law, which we mentioned at the beginning. The Newtonian physics we were taught as absolute truth in school is a good example of how models work only on certain occasions. If you spoke with a physicist and asked whether Newton's laws work the same way on every occasion, at every scale, and at every moment in the universe, the physicist would tell you they do not. You see, Newtonian physics has proved to be a functional model for practical daily life. However, Albert Einstein's theory of General Relativity introduced a new notion of how the universe might work, one different from that of Newtonian physics, and from that point the theoretical movement governing physics today began to take shape: quantum mechanics. What a physicist would tell you today is that some laws appear to work at the large scale, such as in the world around us, while others seem to work only at the small scale, among microparticles. The greatest quest of modern science is to find a model capable of explaining how both the large and the small work in the universe.

The sole purpose of this entire digression has been to show how even what we consider "absolute laws" does not necessarily work in every case. Does this mean Newtonian physics is a lie and should no longer be taught in schools? Of course not. It is an incredibly useful model for describing the world around us; it is, as Professor Stephen Hawking would put it, a functional system. We only need to remember that no law is infallible.

Just as Newtonian physics remains useful to us, so do our models of thought. They are useful tools, but they are not infallible. Let us now examine a more unconscious model of thought: association. It has happened to all of us. When looking at a cloud, we suddenly discover a figure with an incredible resemblance to a face or an animal. We have seen it in the textures of the walls in our homes, in the shapes of rocks, and even in shadows. Our brains evolved in such a way that they had to make quick associations in order to survive. Imagine that you live in the jungle thousands of years ago, when we still hid in caves, and that you go out hunting. As you walk stealthily while following prey, you glimpse a moving orange figure at the edge of your field of vision. The first thing you think is—TIGER!—and you run for your life, feeling that the movement behind you is not the branches you disturbed as you passed, but the hungry beast running at full force. At last you reach your cave, the safety of the shelter, the protection of the tribe's fire. You have managed to survive one more day in that hostile jungle. Was it really a tiger, or was it orange fruit hanging from the branch of a tree moved by the wind? We do not know, but we know that for every X number of branches bearing fruit, there will be a tiger, and reacting differently could have cost you your life.

This mental model of association seems to be one of the most primitive, and it operates constantly throughout the day. For example, do you remember when you began to read, how you had to look at each letter and associate it with its sound before connecting those sounds into a word? The process is very different now, especially for those of us who read a great deal. Of course, all of us read a lot simply because our day and our work require it, but I mean those of us who truly read a great deal. At work, for example, I have to look at code all day, and I also have to read a substantial amount of prose consisting of documentation for that code. In my free time, what I do is read: hundreds of articles, essays, and... more documentation, as well as books. Those who need or habitually choose to read so much, have you ever analyzed your reading process? I have, and it is incredible, because I have realized that I do not actually need to read. Merely by seeing words, my brain is guessing them. It is receiving a series of pieces of information with which to associate the figures with words: surely the number of elements, the length of the word, letters with strokes rising upward such as "b" and downward such as "p". Without needing to read the word, the brain simply associates the figure. This is true of the mental process of almost everyone and happens almost without us noticing, but for those of us who read a great deal, there is a further step: we can generally run our eyes over the lines of a paragraph and automatically extract its most important information. For example, when reading a document in search of certain information, merely by running our eyes over the paragraphs we can determine which words are the paragraph's "key" words and thus find the information more quickly. It is as though the brain ran an algorithm that, upon seeing the paragraph, determined which words were of interest—for example, nouns and adjectives—while setting the others aside. And what do nouns and adjectives have in common? They are generally longer words. It is surely a process determined more by the shapes of the words than by their content.

This model of association is useful not only in the ancient jungle or in reading, but in our daily lives, and thanks to this model we have symbols that have allowed us to share a common language. Think of the sign outside a restroom, especially a men's restroom, anywhere in the world. Regardless of the language or culture, you will recognize that symbol as a person. It is only a figure composed of a circle and a rectangle with appendages, yet we are capable of associating that simple figure with the idea of a human being. Why? Our brains can compare those figures with every figure they have seen and determine that the circle combined with the rectangle and four appendages resembles a head with a torso, arms, and legs. This capacity for association is not merely learned from our surroundings; that is, we do not know that it is a human because someone told us that it is a human, but because of our brain's process of association. Proof of this is our ability to draw. Drawing means creating our own symbols, graphic representations of what we want to express. Children do not need to be taught to draw; they can do it on their own because their brains are developed to associate shapes.

Once again, the model works, provided that we understand our brain can deceive us. If we are unable to control the model, the model will control us. That is why we frequently see in the news how an entire town has devoted itself to a mysterious stain on a tortilla whose shape vaguely resembles the iconography of the Virgin of Guadalupe. "You can even see her little eyes," they will say, and the more they look at it, the more associations their brains will be able to make.

An interesting variation of this mental model is "déjà vu", which could be translated from French as "seen before" and is the sensation that we have experienced a current event before, as though exactly the same thing that had already happened to us were repeating itself. It is an incredible experience. At least, I find it fascinating when I experience it because it becomes a riddle for me. You see, the sensation of déjà vu is simply an association our brain is making, not specifically with an image, but with any previous experience. Following the way association works, it is reasonable to assume that our brain evolved to warn us about dangers in our surroundings, and one way it does this is by recalling previous situations so we know how to react. We are not conscious of it, but our brain is constantly recording information: images, smells, sounds, temperature, tastes, et cetera. It seems as though an algorithm runs in the brain, determining when one or more factors in the current experience match one or more factors in a previous experience and immediately alerting us. It is like a call for our attention so that we know we had a similar experience before and remember how we responded to it. In our daily lives, within the safety of cities, at least relatively safe from beasts and the environment, this instinct does not appear to be very useful. But surely, if we were in a more hostile environment, déjà vu would serve as an injection of fear, alerting us to situations of possible danger. Déjà vu can therefore be triggered by many things: a sound, a smell, an object in our field of vision, et cetera. That is why, when I experience it, it is a riddle to me. I ask myself which of the elements I am perceiving at that moment I had experienced before. I have never solved the riddle, and the criteria under which this algorithm operates will surely remain a mystery.

An interesting case to study within this mental model of association is sound. Certain sounds produce sensations in us, and we do not know why. These sensations can range from calm to fear. Consider a sound that most people find quite unpleasant: fingernails scraping across a blackboard. It is a high-pitched sound, and the level of discomfort it causes varies among people; some are unable to tolerate it. Why does this sound cause a certain anxiety? Perhaps it is an association our brains developed many generations ago as we evolved. That sound might be associated with other sounds we experienced thousands of years ago and that frightened us. Let us return to our life in the ancient jungle, sometime after our encounter with the possible tiger. Imagine what it must have been like to live in a cave, sheltered from nocturnal predators, in the dark, alert to sounds coming from outside and constantly afraid of being attacked. At the darkest and quietest moment of the night, imagine being awakened by the howl of a wolf—or worse, by the death cry of another human being. I know this is a very strange jungle where tigers and wolves live together, but what interests me is imagining how certain sounds became fear triggers for creatures that evolved with those warnings, and how a sound can now cause us anxiety through association alone. More interesting still, it can be an association with something we may never have experienced, but earlier generations did. Even melodies can play with our brains. A masterpiece in this regard is the theme from the film Jaws, which creates tension with only a few notes. Surely, if we could use as a control group people who had never heard the melody or anything about the film, they would describe the sensation it creates as fear. The trick in this melody is that it repeats the notes faster and faster each time. It begins very slowly and gently, then grows increasingly intense, creating the feeling that something is approaching. Our brain, which is excellent at making associations, knows it is not good for something to approach in that way, so it alerts us.

There are mental models that may be less primitive, or that have at least undergone a greater process of refinement, such as prejudices. As their name suggests, prejudices allow us to form an idea about something before studying or experiencing it. Perhaps an extreme situation will make it easier to analyze how useful or deceptive prejudices can be. Imagine a woman walking alone down a street at night (I use a woman because culture has imposed an image of vulnerability on women, which I disagree with but which works well for our example). The woman notices that on the same sidewalk, walking toward her, is a man who under the light of a streetlamp displays characteristics we might associate with a gang member: tattoos all over his body, clothing like a gang member might wear, a shaved head, and a pipe in his hand. On the opposite sidewalk, also walking toward her, is another man. This one is wearing a suit and carrying a briefcase. What is our woman more likely to do: continue walking toward the supposed gang member, or cross the street so that she will instead pass the supposed office worker? We know what she is more likely to do based on the mental model of prejudice, but we do not know whether her decision was correct. The supposed gang member might simply be a plumber, perhaps with poor taste in clothing and a fascination with tattoos, but peaceful, hardworking, honest, and ethical, carrying a pipe for his work. The supposed office worker might in fact be a serial killer whose targets are precisely women like the one in our experiment and who is carrying some limbs from his latest crime in the briefcase as trophies. Do you want to know what happened to the woman? Make up your own story. What matters here is understanding that the model of prejudice can be useful but is not infallible. When we allow prejudices to cloud our ability to reason, we have ceased to be human; we are machines operating according to our brain algorithms.

Prejudices even influence our consumption habits. The packaging of a certain product may look better to us and make us think the product is of higher quality. Surely, far back in the evolutionary chain, we learned to determine the qualities of things quickly.

Prejudices also exist in nature. Many animals have defense systems that consist only of appearing to be predators when they are not, which keeps other animals away. One example is the owl butterfly (caligo idomeneus), whose wings resemble the face of an owl in order to deceive its predators.

If we manage to master these mental models, we will have tools that allow us to interact with the world more quickly and safely. But if the mental models master us, then we are lost.