Of wanting and knowing

What you can imagine begins with what you already know. David Hume made the point in 1748, in An Enquiry Concerning Human Understanding. Thought, he wrote, seems to have unbounded liberty. It can carry you to the far side of the universe in an instant, and form monsters no one has ever seen with no more effort than it takes to picture a chair. But look closer and that freedom turns out to be narrow. All the creative power of the mind, in his words, “amounts to no more than the faculty of compounding, transposing, augmenting, or diminishing the materials afforded us by the senses and experience.”1 You have never seen a golden mountain, yet you can picture one, because you have seen gold and you have seen mountains. A person born blind, Hume noted, can form no notion of colour at all.

Hume’s example of a missing shade of blue shows how imagination can extend what experience gives us. Picture someone who has seen every shade of blue except one. Lay out the whole range in order with that single shade missing, and he will see the gap, and can imagine the colour that belongs in it, though he has never seen it.2 So imagination can extend beyond what we have directly experienced. In this case, the surrounding shades give the mind enough material to supply the missing one. Imagination works with what experience gives it, combining, transforming and sometimes extending it.

More than two centuries later, researchers studied how the brain constructs experiences that have never happened. In 2007, researchers at University College London, in a study led by a young neuroscientist named Demis Hassabis, who would go on to found DeepMind, asked patients with damage to the hippocampus to imagine new scenes, such as lying on a white sandy beach. These patients had severe difficulties forming new memories. Four of the five also struggled to imagine coherent scenes: they could describe elements of a scene, but had difficulty putting them together into a unified spatial setting. The study suggested that remembering past experiences and constructing imagined scenes depend in part on shared processes in the hippocampus.3

Experience gives you material to imagine with. Understanding gives you ways to work with what you have experienced: to recognise relationships, draw implications and imagine how things could be different. As your understanding grows, you can conceive of possibilities you would not have thought to pursue before. Agency, the ability to act toward something you choose, includes deciding which of those possibilities you want to make real.

What you understand shapes the possibilities you can imagine, what you can imagine shapes the futures you might want, and what you want gives direction to what you choose to do.

Where ideas come from

Learning across different fields gives imagination more ways to put that material together. Leonardo da Vinci painted, dissected bodies, designed machines and investigated the movement of water. His knowledge of currents suggested a way to understand the heart: swirling blood might help close a valve after each beat. He devised an experiment using water and grass seeds in a glass model to investigate it.4 Something he understood about rivers had given him a possibility to consider inside the human body. Learning about one part of the world can change what you are able to imagine in another.

What understanding is worth

For most of history, understanding mattered for a simpler reason: it was how the work got done. That reason is fading. You can now outsource your output, and more and more of your thinking. What you cannot outsource is your understanding. Something else can explain a thing to you, but only you can understand it, and only what you understand becomes material for the next thing you imagine.

The obvious objection is that none of this survives a machine that knows more than any of us ever will. Why build understanding you will never use to compete with it? But the point was never to compete. You will not out-know a superintelligence, and you do not need to. What you need is to be able to point it somewhere worth going, and to tell when it has arrived.

During the transition, there is an economic reason to care about this too. When something becomes cheap, value moves to whatever has to be combined with it to be useful.5 Understanding the problem and what a good answer means there helps you put that intelligence to use. And when you hand work to something that knows far more than you, the less you understand, the less you can ask for, check or refuse, so the value goes to whoever does understand, or to whoever owns the machine.6

How long that lasts, no one knows. Every earlier revolution left a higher rung to climb to: from fields to factories, from factories to offices. This one is taking both ends of the ladder at once, with machines that think and, more and more, machines that move. It may leave people with no economic value at all, and if it brings real abundance, an economy built to divide what is scarce may stop making sense altogether.

But the case for understanding never rested on the economy. Aristotle opened his Metaphysics by saying that all people by nature desire to know7, and he was not giving career advice. Understanding, imagining and wanting are not jobs we hold until something cheaper comes along; they are what we mean by human.

A machine can find the best way to get almost anything, and it can suggest what you might want. But wanting happens in you, out of what you are able to imagine. Hand that over too, and you are no longer the one acting. You are the one being acted on. People and machines can make each other better, but people must retain the ability to question, refuse and help direct that relationship. That was true before anyone was paid to think, and it will be true after, only harder to ignore.

What we might become

Hume’s golden mountain was made from things already known: gold and mountains. What we can imagine begins with what we have experienced and come to understand. But the capacities through which we encounter the world may themselves be open to change.

A symbiosis with machines could reach into our biology. Brain–computer interfaces, biotechnology and nanotechnology could eventually help us extend our senses and augment our minds. We could gain new kinds of experience and more capacity to understand them, expanding what we are able to imagine, want and pursue.

We have spent a great deal of time asking how much more machines might become. We should be at least as curious about how much more we might become with them.

Whatever that future brings, understanding more of the world and realising more of our own possibilities is already a life worth pursuing.

Notes

  1. David Hume, An Enquiry Concerning Human Understanding (1748), Section II, “Of the Origin of Ideas”. The golden mountain and the blind man are in the same section. ↩

  2. Hume, Enquiry, Section II. Hume called the case “so singular” that it did not change his general rule. Philosophers have argued about it ever since. ↩

  3. Demis Hassabis, Dharshan Kumaran, Seralynne D. Vann and Eleanor A. Maguire, “Patients with hippocampal amnesia cannot imagine new experiences”, Proceedings of the National Academy of Sciences 104, no. 5 (2007): 1726–1731. ↩

  4. Leonardo da Vinci, The aortic valve, c.1512–13, Royal Collection, RCIN 919082. See also Stanford University, “Water and Circulation”, on the connection between his studies of water and the heart. ↩

  5. Economists call these complements: when the price of one good falls, demand for the goods that have to be combined with it rises. ↩

  6. This is the principal–agent problem: whoever delegates to someone who knows more cannot fully see what they are getting. See Michael C. Jensen and William H. Meckling, “Theory of the Firm”, Journal of Financial Economics 3, no. 4 (1976): 305–360. ↩

  7. Aristotle, Metaphysics, Book I, 980a21. In the Nicomachean Ethics, Book X, he goes further and argues that the life of understanding is the best human life precisely because it is valued for its own sake. ↩