How The Mind Works

Pinker starts the book by asking: “Why are there so many robots in fiction, but none in real life?” (3). The answer stems from the fact that even the simplest actions that we take for granted, like a four-year-old complying with a request to put away a toy, walking, or reading, are complex and precisely organized behaviors. This organization happens in our mind, and understanding these behaviors involves having our mind consider itself. Unsurprisingly, that is no small feat. Even though the mind works seamlessly and as if by magic, there is order and structure to the system.

Consider all the seemingly simple things the human body does every day and the amazing complexity that goes into them. For instance, vision involves processing the brightness of individual patches in our visual field, identifying boundaries between objects, somehow creating their three-dimensional relationships from two-dimensional impressions, and then identifying those objects to make sense of the world. The visual system must handle changes in illumination to know that a snowball is a snowball both bright sunlight and when the sun goes down. Our brain carries out these steps so quickly, we don’t even notice they are all happening.

A robot could be programmed to handle some of the simplest steps quickly, such as processing the brightness of patches in the visual field, but how to three dimensions from two is still not understood, and identifying objects is more challenging that it seems. To identify an object, the robot would first need to have seen these objects (or something similar) before and be able to pull up possible objects based on the shape and size. Not only would the robot need a large catalog of objects, but it would need to understand their relations and context to pick the one most likely being viewed at a given moment.

Philosophers and others have performed experiments and created thought models to examine how robots would handle various tasks, and the conclusions are generally that robots would fail to handle several things humans would take for granted. Daniel Dennett argued that humans live by a set of core rules that we use to quickly learn and reason about the world around us, building knowledge that is culturally relevant and nuanced instead of, like a computer, receiving a download of facts to apply regularly and without alteration. The rules humans use are difficult to determine.

Pinker argues that the computational theory of mind helps us understand these rules and how they are used. First, the theory posits that information used by the mind doesn’t have to be physical. Beliefs and desires, such as “I want to go to grandma’s house,” are arrangements of symbols that have logical connections to other symbols in our brains. If we decide we want to go to grandma’s house, we can then draw up the symbols related to what we must do to get there (for example, the bus route or directions). Second, the magic of the brain’s ability to handle millions of different pieces of information with millions of different connections but to still look basically the same in each human must have to do with the pattern of activity. The brain produces its many nuanced actions through slight variations in patterns of activity. These patterns of activity are like computations and not really rules at all. Computations allow for infinite inputs and infinite outputs, just like the brain.

Humans, and therefore their brains, have evolved over time. Despite public resistance to this idea, there is considerable evidence that evolution has shaped the human brain. The vertebrate eye, with both its amazing ability to see and its intricate structure, could only have come about through natural selection. Similarly, the brain it is connected to could only have come about through natural selection, adaptation to the environment, and selection for those who thrive in that environment. However, evolution and natural selection are often poorly understood. Previous researchers have used reverse-engineering, which is a process of understanding a goal and all the factors that come to bear on why that would be a goal and how that goal could be accomplished. Bringing these factors together, some indications of natural selection emerge even for phenomena that don’t seem adaptive.

For instance, morning sickness does not appear adaptive on the surface. Making the mother throw up and avoid food isn’t helpful for growing a new human. There must be a benefit because morning sickness has been linked to healthier pregnancies and reduced risk of miscarriage. One plausible theory is that morning sickness exists to help mothers avoid foods that may contain toxins that could harm the baby. This theory makes sense when one considers that bitter, flavorful, and pungent foods are the top food aversions for pregnant women, and these characteristics are associated with higher likelihood of toxins, especially for our ancestors who did not process and test their food as we do today.

Importantly, natural selection doesn’t select for every possible human behavior; it happens on a species scale, meaning people can still have behaviors that don’t increase fitness and likelihood of passing on genes. Natural selection also happens over many generations because the adaptive behaviors must be passed down, and some people with some maladaptive behaviors will still survive. Additionally, natural selection acts on physical structures, like the brain and how it is arranged and operates, and not concepts such as adultery or love or beauty. Natural selection may select for hearing systems that like rhythmic patterns (music) because music brought our ancestors together. People who liked music formed groups that were more likely to survive by working together. In turn, cooperation became a standard behavior. Natural selection didn’t select for cooperation, but it pushed humans in that direction through affecting the hearing system.