Only after you have watched the game can you understand why the Brazilians still grumble about it—in spite of their later “revenge” on Germany in the 2016 Summer Olympics.
Massed learning
UNFORTUNATELY, MANY METHODS for learning (whether in high school, at the university level, in vocational training, or continuing education at the workplace) continue to rely on the basic concept that memorizing facts and data is a good idea. On the contrary, this method leads to a completely false strategy for learning, known scientifically as “massed learning,” in which you must pump yourself full of information in a short period of time in the hopes of retaining as much as possible in the future. This obviously doesn’t work, since our brain thinks data packets are totally uninteresting.
An orchestra doesn’t learn a new piece of music merely by playing a single note for one second and then waiting before processing the next informational packet (the next note) and so on for thousands of notes (this would be akin to “massed learning”). No, it learns best by quickly recognizing the relationship between the notes and the way in which, at a certain time and place, they develop into a whole melody.
The context is what allows us to learn effectively—namely, that we do not have to consciously concentrate on the idea that we are learning. This became evident in a study conducted by the work group of my colleague, Melissa Vo, who researched memory capacity among adults. Specifically, the study’s adult test subjects were asked to find objects that were pictured in an apartment setting (i.e., the soap in the bathroom). Although participants were not asked to remember these objects, they were much better able to recall them later than if they had been asked to memorize isolated pictures of the objects.5 When the same objects were isolated and presented to participants in front of a neutral background, the information was much less interesting to participants and thus not saved. A bar of soap makes much more sense in the context of a bathroom than surrounded by a green background. The object by itself is not interesting. It is only its situation in a particular context that gives the object a meaningful correlation, which we do not forget. Though this may seem illogical because it implies that we are required to note down additional information (namely, the object’s surroundings), this is, in fact, an ability that comes easily to us.
The lasagna-learning rule
IN ORDER TO understand this correlation of context and of the meaning of a word, the brain must learn differently than it might be used to—namely, with interruptions. In the last chapter, you already read that the brain sacrifices some pieces of memory to nonmemory (or even actively forgets them) in order to be able to actively combine them. Something similar takes place with learning. Learning is successful when breaks and distance are built into the process, a practice referred to as “spaced learning.” This would seem to go against our better intuition, as we assume that we will only be able to grasp correlations and concepts by processing as much information as possible at one time. If we deliberately incorporate breaks into our learning process, we fear we might forget things that could be important. But our brain is not interested in the sheer mass of information so much as it is interested in our ability to connect the information.
To research this, one study asked participants to identify the painting style of various artists. The subjects were divided into two groups. The first group was shown a series of six images, all of them works by one artist, followed by another series of six images, which were by the next artist, and so on for the next four artists. The second group was shown all of the images mixed up in no particular sequence, so that the various artistic styles alternated from image to image. The results were clear. The group that viewed the alternating images was able to identify a new image according to the particular style of the individual artist. Those in the first group, who viewed the images in sequential blocks, were less able to recognize the basic painting concept (artistic style). Despite the results, most of the test subjects indicated that they preferred learning in blocks (“massed learning”), as they believed it to be a more successful strategy.6
This result has been reaffirmed over and over again in studies. Taking breaks is what makes learning successful. Not only for learning about various artistic styles, but also vocabulary at school, movement patterns, biological correlations, or lists of words. The reason for this has to do with the way in which our nerve cells interact. An initial information impulse triggers a stimulus for structural change in the cells. These changes must first be processed to prepare the cells for the next informational push. Only after they have taken a short break are they optimally prepared to react to the recurrent stimulus. If it comes too early, it will not be able to fully realize its effect.7 It is only by alternating information that the brain is able to embed it in a context of related bits of knowledge. It’s not too different from making lasagna. You could of course choose to pour the sauce into the pan all at once and then pile the lasagna noodles and the cheese on top. That would be something like “massed cooking,” but it wouldn’t result in authentic lasagna. Only when you alternate the components do you get the desired, delicious dish—or, when it comes to the brain, a meaningful thought concept. This kind of conceptual thought is the brain’s great strength because it enables us to get away from pure rote learning. Only then is it possible for us to organize the world into categories and meaningful correlations and, thereby, to begin to understand it.
Don’t learn—understand!
ANYONE WHO CAN learn something can also unlearn it. But once you have understood something, you cannot de-understand it. Learning is not particularly unique. Most animals and even computers can learn. But developing an understanding of the things in the world is the great art of the brain, which it is able to master precisely because it does not consume and draw correlations from data in the same way a robot would. A brain creates knowledge out of data, not correlations. These are two vastly different concepts, though they are often equated with each other in the modern, digitalized world. But whereas the amount of data from :-) and R%@ is the same, the information conveyed is completely different. Not to mention the concept behind it—a smiling face. To a computer, the characters :-) and :-( are only 33 percent different. But to us, they are 100 percent different.
How do we learn such knowledge, such thought concepts? How do we understand the world? We can see how we don’t do it by marveling at computer algorithms. Specifically, the most modern algorithms in existence, the “deep neural networks.” These are computer systems that are no longer programmed to follow the classic A then B system of logic. Rather they “borrow” from the brain and copy its network structure. The software simulates digital neurons that are able to adapt their points of contact to one another depending on which pieces of data they need to process. Because the cells and their contacts are able to adjust themselves, the system is able to learn over time. For example, if the software needs to be able to identify a penguin, it is presented with hundreds of thousands of random images with a few hundred penguin images included among them. The program independently identifies the characteristics specific to penguins until it is able to recognize what a penguin might look like.
The advances that have been made in artificial neural networks are huge. Merely by regularly viewing images, such a system is independently capable of identifying animals, objects, or humans in arbitrary pictures. Facial recognition capabilities have even surpassed human ability (Google not only pixels out human faces in its Street View maps but also the faces of cows).8 But to put it all into perspective: a computer system like this is to the brain what a local amateur athlete is to an Olympic decathlon champion. The comparison is not even the same concept because computers do something that is very different than neurons, in spite of the pithy appropriation of the neurology terms by IT companies who claim they are building “artificial neural networks.” In reality, computers are neither replicating real neural networks nor a brain. It is nothing but a marketing trick by computer companies. For a deep learning network to learn to identify a penguin, it must first process thousands of images of one, in a method that follows the maxim “practice makes perfect.” But this is not necessarily how
