THE PREDICTION MACHINE
How your beliefs shape your reality
It was just a few nights before Christmas, and the drones seemed to be everywhere and nowhere at the same time.
The drama began at nine p.m. on December 19, 2018, when a security officer at London Gatwick Airport reported two unmanned aerial vehicles—one flying around the perimeter fence, another inside the complex. The runway was soon closed for fear of an impending terrorist attack. It was only nineteen months after the Islamist bombing at the Manchester Arena, after all, and there had been reports that members of ISIS were planning to carry explosives on commercial drones.
The chaos escalated over the following thirty hours as dozens of further sightings kept the airport in lockdown. Try as they might, however, the security officers and police just couldn’t locate the drones, which seemed to disappear as soon as they were sighted. Even more astonishingly, their operators appeared to have found a way to avoid the military’s track-and-disable system, which was unable to detect any unusual activity in the area, despite a total of 170 reported sightings. The news soon spread to the international media, which warned that similar attacks might occur in other countries.
By six a.m. on December 21, the threat finally seemed to have passed, and the airport reopened for business. Whoever was behind the attack—be it a terrorist or a joker—had achieved their aim of chaos, disrupting the travel of 140,000 passengers with the cancelation of more than a thousand flights. Despite offering a substantial reward, the British police have been completely unable to find a culprit, and there is not a single photo offering evidence of an attack—leading some (including members of the police) to question whether there were ever any drones at all.1 Even if there was, at one point, a drone near the airport, it’s clear that the vast majority of the sightings were false, and the ensuing chaos was almost certainly unnecessary.
With so many independent reports from dozens of sources, we can easily rule out the possibility that this was some kind of lie or conspiracy. Instead, the event demonstrates the power of expectation to change our perception, and—occasionally—to create a vision of something that is entirely false.
According to an increasing number of neuroscientists, the brain is a “prediction machine” that constructs an elaborate simulation of the world, based as much on its expectations and previous experiences as on the raw data hitting the senses. For most people, most of the time, these simulations coincide with objective reality, but they can sometimes stray far from what is actually in the physical world.2
Knowledge of the prediction machine, and its workings, can help us to understand everything from ghost sightings to disastrously bad calls by sports referees—and the mysterious appearance of nonexistent drones in the winter sky. It can help us to understand why the name we call a beer changes its taste; and it shows how to someone with a phobia the world looks much more terrifying than it really is. This grand new unifying theory of the brain also sets the stage for all the expectation effects that we’ll examine in this book.
THE ART OF SEEING
The seeds of this extraordinary conception of the brain were sown in the mid-nineteenth century by the German polymath Hermann von Helmholtz. Studying the anatomy of the eyeball, he realized the patterns of light hitting the retina would be too confusing to enable us to recognize what is around us. The 3D world—with objects at various distances and odd angles—has been flattened onto two two-dimensional disks, resulting in obscured and overlapping contours that would be difficult to interpret. And even the same object may reflect very different colors depending on the light source. If you are reading this physical book indoors at dusk, for example, the page will be reflecting less light than a dark gray page in direct sunlight—yet in both cases, the page looks distinctly white.
Helmholtz suggested that the brain draws on past experiences to tidy up the visual mess and to come up with the best possible interpretation of what it receives, through a process he called “unconscious inference.” We may think we are seeing the world unfiltered, but vision is really forged in the “dark background” of the mind, he proposed, based on what it assumes is most likely to be in front of you.3
Helmholtz’s theories of optics influenced postimpressionist artists like Georges Seurat,4 but it was only in the 1990s that the idea really started to take off in neuroscience—with signs that the brain’s predictions influence every stage of visual processing.5
Before you walk into a room, your brain has already built many simulations of what might be there, which it then compares with what it actually encounters. At some points, the predictions may need retuning to better fit the data from the retina; at others, the brain’s confidence in its predictions may be so strong that it chooses to discount some signals while accentuating others. Over numerous repetitions of this process, the brain arrives at a “best guess” of the scene. As Moshe Bar, a neuroscientist at Bar-Ilan University in Israel who has led much of this work, puts it: “We see what we predict, rather than what’s out there.”
A wealth of evidence now supports this hypothesis, right down to the brain’s anatomy. If you look at the wiring of the visual cortex at the back of the head, you find that the nerves bringing electrical signals from the retina are vastly outnumbered by the neural connections feeding in predictions from other regions of the brain.6 In terms of the data it provides, the eye is a relatively small (but admittedly essential) element of your vision, while the rest of what you see is created “in the dark” within your skull.
By measuring the brain’s electrical activity, neuroscientists like Bar can watch the effects of our predictions in real time. He has observed, for instance, the passing of signals from frontal regions of the brain—which are involved in the formation of expectations—back into the visual cortex at the earliest stages of visual processing, long before the image pops into our consciousness.7
There are lots of good reasons why we might have evolved to see the world in this way. For one thing, the use of predictions to guide vision helps the brain to cut down the amount of sensory information it processes, so that it can focus on the most important details—the things that are most surprising, and which do not fit its current simulations.
As Helmholtz originally noted, the brain’s reliance on prediction can also help us to deal with incredible ambiguity.8 If you look at the image here—a real, albeit poor-quality, bleached photograph—you will probably struggle to identify anything recognizable.
If I tell you to look for a cow, however—facing you, with its large head toward the left of the image—you may find that something somehow “clicks,” and the image suddenly makes a lot more sense. If so, you’ve just experienced your brain’s predictive processing retuning its mental models to make use of additional knowledge, transforming the picture into something meaningful.
Copyright © 2022 by David Robson