In 1665 the Danish anatomist Nicolaus Steno argued that our brain functions like a machine. If we want to understand what the brain does and how it does it, we need to treat it like a machine and take it apart to see how it works. For over 350 years we have been following Steno’s suggestion—peering inside dead brains, removing bits from living ones, recording the electrical activity of nerve cells (neurons) and, most recently, altering neuronal function with the most astonishing consequences.
We can now make a mouse remember something about a smell it has never encountered, turn a bad mouse memory into a good one and even use a surge of electricity to change how people perceive faces. In some species we can change the brain’s very structure at will, altering the animal’s behavior as a result. Some of the most profound consequences of our growing mastery can be seen in our ability to enable a paralyzed person to control a robotic arm with the power of their mind. And yet we still have no clear comprehension about how billions, or millions, or thousands, or even tens of neurons work together to produce the brain’s activity.
We know in general terms what is going on—brains interact with the world, and with our bodies, by sending stimuli across neural networks. Brains predict how those stimuli might change in order to be ready to respond, and as part of the body they organize its action. This is all achieved by neurons and their complex interconnections, including the many chemical signals that envelop them.
However, when it comes to really understanding what happens in a brain at the level of neuronal networks and their component cells, or to being able to predict what will happen when the activity of a particular network is altered, we are still at the very beginning.
A key clue to explaining how we have made such amazing progress and yet have still barely scratched the surface of the astonishing organ in our heads is to be found in Steno’s suggestion that we should treat the brain as a machine. “Machine” has meant very different things over the centuries, and each of those meanings has had consequences for how we view the brain.
In Steno’s time machines were based on either hydraulic power or clockwork mechanisms. The insights that could be gleaned about the structure and function of the brain were limited, and no one now looks at the brain this way. In the nineteenth century the brain was seen first as a telegraph network and then as a telephone exchange, allowing for flexible organization and output. Since the 1950s our ideas have been dominated by concepts from computing—feedback loops, information, codes and computation.
Even the simplest animal brain is not a computer like anything we have built, nor one we can yet envisage. The brain is not a computer, but it is more like a computer than it is like a clock, and by thinking about the parallels between a computer and a brain we can gain insight into what is going on inside both our own heads and those of animals.
Over the centuries, each layer of technological metaphor has added something to our understanding, enabling us to carry out new experiments and reinterpret old findings. But by holding tightly to metaphors, we end up limiting what and how we can think.
A number of scientists are now realizing that, by viewing the brain as a computer that passively responds to inputs and processes data, we forget that it is an active organ, part of a body that interacts with the world and that has an evolutionary past that has shaped its structure and function. We are missing out key parts of its activity. In other words, metaphors shape our ideas in ways that are not always helpful.
The tantalizing implication is that tomorrow our ideas will be altered yet again by the appearance of new technological developments. We will reinterpret our current certainties, discard some mistaken assumptions and develop new theories and ways of understanding. When scientists realize that how they think—including the questions they can ask and the experiments they can imagine—is partly framed and limited by technological metaphors, they often get excited and want to know what the Next Big Thing will be and how they can apply it to their research. If I had the slightest idea, I would be a very wealthy person.
The history of how we have understood the brain contains recurring themes and arguments, some of which still provoke intense debate today. One example is the extent to which functions are localized in specific areas of the brain. That idea goes back thousands of years, and there have been repeated claims up to today that certain parts of the brain appear to be responsible for a specific thing, such as the feeling in your hand, or your ability to understand syntax or to exert self-control.
Such claims were often quickly qualified by the revelation that other parts of the brain also influence or supplement those actions, and that the brain region in question is also involved in other processes. Repeatedly, localization has not exactly been overturned, but it has become far fuzzier than originally thought.
The reason is simple. Brains, unlike machines, have not been designed. They are organs that have evolved for over five hundred million years, so there is little or no reason to expect them to function like machines that people construct.
This may also underlie why some researchers sense we are approaching an impasse in how we understand the brain. This might seem paradoxical — every day we hear about new discoveries that shed light on how brains work, along with the promise (or threat) of new technology that will enable us to do such far-fetched things as read minds, detect criminals or even be uploaded into a computer. In contrast to all this exuberance, there is a feeling among some neuroscientists that our future path is unclear. It is hard to see where we should be going, apart from simply collecting more data or relying on the latest exciting experimental approach.
That does not mean that everyone is pessimistic — some confidently claim that the application of new mathematical methods will enable us to understand the myriad interconnections in the human brain. Others (like myself) favor studying animals at the other end of the scale, focusing our attention on the tiny brains of worms or maggots and employing the well-established approach of seeking to understand how a simple system works, and then applying those lessons to more complex cases. Many neuroscientists, if they think about the problem at all, simply consider that progress will inevitably be piecemeal and slow, because there is no Grand Unified Theory of the brain lurking around the corner.
The problem is twofold. Firstly, the brain is mind-bogglingly complex. A brain—any brain, not just the human brain, which has been the focus of much of the intellectual endeavor described here — is the most complex object in the known universe. Second, despite the tsunami of brain-related data being produced by laboratories around the world, we are in a crisis of ideas about what to do with all that data, about what it all means.
I think that this reveals that the computer metaphor, which has served us so well for over half a century, may be reaching its limits, just as the idea of the brain as a telegraph system eventually faded away in the nineteenth century.
Some scientists are now explicitly challenging the usefulness of some of our most basic metaphors about the brain and nervous system, such as the idea that neuronal networks represent the outside world through a neuronal code. This suggests that scientific understanding may be chafing at the framework imposed by our most deeply held metaphors about how the brain works.
It may prove to be that even in the absence of new technology, developments in computing, in particular relating to artificial intelligence and neural networks—which are partly inspired by how brains do things—will feed back into our views of the brain, giving the computational metaphor a new lease of life. Perhaps. But leading researchers in deep learning—the most fashionable and astonishing part of modern computer science—cheerfully admit that they do not know how their programs do what they do. I am not sure that computing will provide enlightenment as to how the brain works.
Properly understanding the human brain, with its tens of billions of cells and its incredible and eerie ability to produce the mind, may seem an unattainable dream. But science is the only method that can reach this goal, and it will reach it, eventually.
There have been many similar moments in the past, when brain researchers became uncertain about how to proceed. In the 1870s, with the waning of the telegraph metaphor, doubt rippled through brain science and many researchers concluded it might never be possible to explain the nature of consciousness. One hundred and fifty years later we still do not understand how consciousness emerges, but scientists are more confident that it will one day be possible to know, even if the challenges are enormous.
Understanding how past thinkers have struggled to understand brain function is part of framing what we need to be doing now, in order to reach that goal. Our current ignorance should not be viewed as a sign of defeat but as a challenge, a way of focusing attention and resources on what needs to be discovered and on how to develop a program of research for finding the answers. This highlights why the four most important words in science are “We do not know.”
