With his palace, Simonides identified an idiosyncratic trait of human memory. The technique works because we all have a fabulous spatial memory. To prove this, you only need to think of how many maps and routes (through towns or people’s homes, of buses, through cities and buildings) you can recall with no effort whatsoever. This original seed of a discovery came to fruition in 2014, when John O’Keefe and the Norwegian couple May-Britt and Edvard Moser won the Nobel Prize in Medicine for finding a system of coordinates in the hippocampus that articulates this formidable spatial memory. It is an age-old system that is even more refined in small rodents–who are extraordinary navigators–than in our mastodon minds. Situating ourselves in space has always been necessary. Unlike remembering countries’ capitals, numbers and other things, which our brains never evolved to do.
Here we see an important idea. An ideal way for us to adapt to new cultural needs is by recycling brain structures that evolved in other contexts to fulfil other functions. The memory palace is a very paradigmatic example. We all struggle to remember numbers, names or shopping lists. But we can easily recall hundreds of streets, the nooks and crannies of our childhood homes, or the houses of our friends when we were growing up. The secret of the memory palace is establishing a bridge between those two worlds, what we want to remember, but find it difficult to, and space, where our memory is right at home.
Read this list and take thirty seconds to try to remember it: napkin, telephone, horseshoe, cheese, tie, rain, canoe, anthill, ruler, tea, pumpkin, thumb, elephant, barbecue, accordion.
Now close your eyes and try to repeat it in the same order. It seems difficult, almost impossible. Yet someone who has built their palace–which requires a few hours of work–can easily remember a list like this. The palace can be open or closed, an apartment building or a house; then what you have to do is go through each room and place, one by one, each of the objects on the list. You have to do more than just name them. In each room you have to create a vivid image of the object in that place. The image must be emotionally powerful, perhaps sexual, violent or scatological. The unusual mental stroll, in which we peek into each room and see the most bizarre images featuring those objects in our own palace, will persist in our memories much more than the words.
So a prodigious memory is based on finding good images for those objects we wish to remember. The task of memorizing is somewhere between architecture, design and photography, all creative tasks. Memory, which we tend to perceive as a rigid and passive aspect of our thought, is actually a creative exercise.
In short, improving our memories doesn’t mean increasing the space in the drawer where the memories are kept. The substratum of memory is not a muscle that grows with exercise and is strengthened. When technology made it possible, Eleanor Maguire confirmed this premise by investigating the very factory of memories. She discovered that the brains of the great champions of memory are anatomically indistinguishable from anyone else’s. Nor were those champions more intelligent or better at remembering things outside of the realm they had studied, just like virtuoso chess players. The only difference was that the prodigious memorizers use the spatial structures of their memories. They have managed to recycle their spatial maps in order to remember arbitrary objects.
The morphology of form
One of the most spectacular cerebral transformations occurs when we learn to see. This happens so early in our lives that we do not have any memory of how we perceived the world before seeing. From a stream of light, our visual system manages to identify shapes and emotions in a tiny fraction of a second, and what is even more extraordinary is that it happens without any sort of effort or conscious realization that something must be done. But converting light into shapes is so difficult that we have yet to create machines that can do it. Robots go into outer space, play chess better than the greatest masters, and fly aeroplanes, but they cannot see.
To understand how the brain pulls off such a feat, we must find its limits, see exactly where it fails. To do that, we will look at a simple yet eloquent example. When we are trying to think about how we see, an image is definitely worth more than a thousand words.
The two objects in the following figure are very similar. And both, of course, are very easy to recognize. But when they are submerged in a sea of dotted lines something quite extraordinary happens. The visual brain works in two completely different ways. It is impossible not to see the object on the right; it’s as if it were another colour and was literally popping out. Something different happens with the object on the left. We see the dotted lines that make up the snake only with much effort, and our perception is unstable; when we are focused on one part, the rest vanishes and blends into the texture.
We can think of the object we see easily as a melody in which the notes follow each other harmonically and are naturally perceived as a whole, while the other object is more like random notes. Just as with music, the visual system has rules that define how we organize an image and that dictate how we perceive and what we remember. When an object is grouped in a natural, integrated way that does not require much effort, it is said to be gestaltian, named for the group of psychologists who, in the early twentieth century, discovered the rules by which the visual system constructs shapes. These rules, like those of language, are learned.
Let’s look at how this works in the brain. Can we train and modify the brain to detect any object almost instantaneously and automatically? In the process of answering this question, we will sketch a theory of human learning.
A monster with slow processors
Most silicon computers today work with just a few processors. These computers calculate very quickly, but can only respond to one thing at a time. Our brains, on the other hand, are parallel machines–to a massive degree; which is to say, they are simultaneously making millions and millions of calculations. Perhaps this is one of the most distinctive aspects of the human brain and, to a large extent, it allows us quickly and effectively to solve problems that we are still unable to delegate to even the most powerful computers. This is one of the areas of most intense effort in computer science and yet the attempt to develop massively parallel computers has produced only elusive results. These researchers are presented with two fundamental difficulties: the first is simply to find a way to produce that number of processors economically; and the second is to get them all to share information.
In a parallel computer, each processor does its job. But the result of all that collective work has to be coordinated. One of the most mysterious aspects of the brain is how it manages to unite all the information processed in parallel. This is profoundly linked to consciousness. Which is why, if we understand how the brain brings together the information it calculates on a massive scale, we will be much closer to revealing the mechanism of consciousness. And we will have discovered the processes of how to learn.
The secret of virtuosity is in being able to recycle this parallel machinery and adapt it to new functions. The great mathematician sees mathematics. The chess master sees chess. And this happens because the visual cortex is the most extraordinary parallel machine known to humankind.
The visual system is comprised of superimposed maps. For example, the brain has a map that is devoted to codifying colour. In a region called V4,* modules of approximately a millimetre in size, called globs, are formed, and each one identifies different subtleties of colour in a very precise region of the image.
The big advantage to this system is that recognizing something does not require sequentially sweeping over it point by point. This turns out to be particularly important in the brain. It takes a long time for a neuron to load and route information from one to the next, which means that the brain can process between three and fifteen computing cycles in a second. That is nothing compared to the billions of cycles per second of the tiny processor in a mobile phone.
The brain solves the intrinsic slowness of its biological fabric with an almost infinite army of neural circuits.* So the conclusion i
s simple, and as I will argue in the next few pages will be the key to the enigma of learning: any function that can be resolved in the parallel structures (maps) of the brain will be done effectively and efficiently. It will also be perceived as automatic. On the other hand, functions that use the brain’s sequential cycle are carried out slowly and are perceived with great effort and fully consciously. Learning in the brain is, to a great extent, parallelizing.
The repertoire of visual maps includes movement, colour, contrast and direction. Some maps identify more sophisticated objects as two contiguous circles. In other words, eyes that watch us. That is what produces the strange sensation of turning your head quickly towards someone watching you. How did you know they were watching you before turning your gaze? The reason is that the brain is exploring the possibility that someone is watching us throughout the entire space and in parallel, often without any conscious register of it. The brain detects a different attribute in one of its maps, and generates a signal that communicates with the attention and motor control system of the parietal cortex as if** saying: ‘Turn your eyes over there because something important is happening.’ These maps are like factory settings of a range of innate skills. They are efficient and at the same time fulfil a very specific task. But they can be modified, combined and rewritten. And in this lies the key to learning.
Our inner cartographers
The cerebral cortex is organized into columns of neurons, and each one carries out a specific function. That was discovered by David Hubel and Torsten Wiesel, earning them the Nobel Prize in Physiology. When studying how those maps developed they found that there were critical periods. The visual maps have a natural programme of genetic development but they need visual experience to consolidate. Like a river that needs water running through it in order to maintain its shape.
The retina, especially in the first phases of development, generates spontaneous activity, stimulating itself in complete darkness. The brain recognizes this activity as light, without differentiating whether it comes from outside the body or not. Therefore the development based on activity begins before we open our eyes. Cats, for example, are born with their eyes closed. They are actually training their visual system with inner light. In cats, humans and other mammals, these maps develop in early infancy and are already consolidated after a few months. Hubel and Wiesel’s discovery converges with another myth: that learning certain things as an adult is an impossible mission. We are going to revisit this idea and offer some moderate optimism: learning later in life is much more plausible than we imagine but requires much time and effort–the same amount we devote to such tasks in our early infancy, although we have now forgotten that. After all, babies and kids spend hours, days, months and years of their lives learning to speak, walk and read. What adult puts everything else to one side to devote all their time and effort to learning something new?
In retrospect, some of this is obvious. Radiologists, as adults, learn to see x-rays. After much work they can easily identify oddities that no one else can see. It is the clear result of a transformation in their adult visual cortex. In fact, for radiologists, such detection is quick, automatic and almost emotional, as when we have a visceral response of annoyance when we see ‘grammatical‘ errors. What happens in the brain that can so radically transform our way of perceiving and thinking?
Fluorescent triangles
Science has some curious repetitions. Those who come up with extraordinary, paradigmatic ideas are often the very same ones who later topple them. Torsten Wiesel, after establishing the dogma of the critical periods, got together with Charles Gilbert, one of his students at Harvard, to prove just the opposite, that the visual cortex continues to reorganize itself even in full adulthood.
When I arrived in Gilbert and Wiesel’s laboratory–by that point it had moved to New York–to begin my doctorate, the myth had already been turned on its head. The question was no longer whether the adult brain could learn, but how exactly it did. What happens in the brain when we become experts in something?
We devised an experiment to be able to carefully investigate this question in the laboratory. This required making certain concessions in a process of simplification. So instead of using expert radiologists, we created experts in triangles. Something with very little merit as a skill or profession but which has the laboratory advantage of being a simple way to simulate a learning process.
We showed a group of people an image filled with shapes that after 200 milliseconds disappeared in a flash. They had to find a triangle in that mess. They looked at us like we were crazy. It was impossible. They simply hadn’t had enough time to see it.
We knew that if the test had been finding a red triangle among many blue ones, everyone would have solved it easily. And we know why. We have a parallel system that, in eighty milliseconds, can sweep the space in unison in order to solve a difference of colour, but our visual cortices do not have a similar map devoted to identifying triangles. Can we develop that ability? If so, we would be opening a window on to how we learn.
Over hundreds of attempts, many of the participants were frustrated to find they saw nothing. But after hours and hours of repeating this boring task, something magical happened: the triangle began to glow, as if it were a different colour, as if it were impossible not to see it. So we know that with much effort we can see something that previously seemed impossible. And that it can be done as an adult. The big advantage to this experiment is that it allowed us to study what happens in the brain as we learn.
The parallel brain and the serial brain
The cerebral cortex is organized into two large systems. One is the dorsal, which–if your head is looking upwards–continues along the back of the body, and the ventral, which corresponds to the belly. In functional terms, this division is much more pertinent than the more popular division of the hemispheres. The dorsal part includes the parietal and frontal cortex, which have to do with consciousness, with cerebral activity that deals with action, and with a slow, sequential working of the brain. The ventral part of the cerebral cortex is associated with automatic and generally unconscious functions, and corresponds to a rapid, parallel way of working.
We found two fundamental differences in the cerebral activity of the triangle experts. Their primary visual cortex–in the ventral system–activated much more when they saw triangles than when they saw other shapes they hadn’t been trained to identify. And at the same time their frontal and parietal cortices deactivated. This explains why for them seeing triangles no longer entailed effort. This is not specific to triangles. A similar transformation is observed when a person trains to recognize something (for example, musicians learning to read scores, a gardener learning to recognize a parasite on a plant, or coaches who realize in a matter of seconds that their team is flailing out on the field).
Dorsal pathway: Produces learning
Ventral pathway: OK threshold
Dorsal pathway: Slow
Ventral pathway: Fast
Dorsal pathway: Mental effort
Ventral pathway: Automatic
Dorsal pathway: Sequential
Ventral pathway: Parallel
Dorsal pathway: Flexible and versatile
Ventral pathway: Rigid and Stereotypical
Dorsal pathway: Reads letter by letter
Ventral pathway: Automatic reading
Learning: a bridge between two pathways in the brain
The cortex is organized into the dorsal system and the ventral system. Learning consists of a process of transferring from one system to the other. When we learn to read, the slow, effortful system that works ‘letter by letter’ (dorsal system) is replaced by the other, which is capable of detecting entire words without effort and much faster (ventral system). But when the conditions are not favourable for the ventral system (for example, if the letters are written vertically) we go back to using the dorsal, which is slow and serial but has the flexibility to adapt to different circumstances. In many cases, learning me
ans freeing the dorsal system to automatize a process so we can devote our attention and mental effort to other matters.
The repertoire of functions: learning is compiling
The brain has a series of maps in the ventral cortex that allow it to carry out some functions rapidly and efficiently. The parietal cortex allows for the combination of each of those maps, but through a slow process that requires effort.
However, the human brain has the ability to change its repertoire of automatic operations. After thousands and thousands of rehearsals, a new function can be added to the ventral cortex. We can think of this as a process of outsourcing, as if the conscious brain were delegating this function to the ventral cortex. Conscious resources, requiring mental effort and limited to the capacities of the frontal and parietal cortices, could be devoted to other tasks. This is a key to learning how to read that is enormously relevant to educational practice. Expert readers, who read cover to cover with no effort, delegate the reading; readers who are learning don’t, and their conscious mind is fully occupied by the task.
The Secret Life of the Mind Page 20
