Predictive processing is sometimes called a grand unifying theory of the brain. An important guide to the field from Andy Clark shows the idea’s strength – but also how far it has to go to fulfil that.
 |
| Optical illusions fool us because of the way we make predictions about size and distance Mario Gyb/Alamy |
Andy Clark (Allen Lane)
ON A building site, there is a scream of pain. A worker has jumped down from scaffolding and landed on a long nail that is now emerging from the top of his boot. In clear agony, the man is taken to hospital, where his footwear is cut away to reveal that the nail passed between his toes without even breaking the skin. The pain was entirely in the man’s head – yet it was very real to him, a paradox explained by Andy Clark in his new book The Experience Machine: How our minds predict and shape reality.
In contrast to what we might expect, and what neuroscientists used to believe, the way we perceive the external world isn’t just based on raw data coming in through our senses. It is a merging of our brains’ predictions combined with that new data.
Known as predictive processing theory, this is one of the hottest topics in neuroscience at the moment, and it has been described as a grand unifying theory of the brain.
The process might seem overly complicated, but in fact it is a very efficient way for our brains to run, explains Clark. Consider how video files are compressed by computers – say, a movie of a man running down a corridor. The only difference between frame 4 and frame 5 is a slight forward motion of the runner, says Clark. All that needs to be transmitted to capture frame 5 are those few differences to frame 4, which take up far less data than a new value for every pixel in frame 5.
For the same reasons, our brains do something similar when interpreting inputs from our senses: making predictions based on existing knowledge and only transmitting onwards the differences, which are known as the “prediction errors”. This helps explain why it is so easy to be fooled by optical illusions, as we unconsciously make predictions about the size and distance of objects based on cues from their surroundings.
This can also be illustrated, he says, when you listen to heavily distorted sounds, such as the “Brainstorm/green needle” clip (YouTube: https://bit.ly/3opzzsD ). What you hear changes depending on whether you read the words “brainstorm” or “green needle”.
We don’t just use predictive processing for external inputs, but also when interpreting internal signals from our bodies. This may account for what are sometimes called psychosomatic illnesses (now often known as “functional disorders”). An example would be someone who is convinced that part of their body is paralysed, despite there being no physical cause.
Some neuroscientists think psychosomatic illnesses could be explained by faulty predictive processing: the conscious belief that the body part is paralysed, say, is overruling the real incoming sensory data. Supporting this idea is the fact that one of the most successful treatments for functional disorders uses distraction techniques that shift attention away from the malfunctioning body part, allowing the real internal data to take over.
As someone who has been involved in the field for years, Clark makes a knowledgeable tour guide – although, if I have one gripe, I would have preferred a more impartial account. Predictive processing can feel like a “just-so” story when it is used to explain any puzzling aspect of the brain, from the placebo effect to our susceptibility to mental illness, without much solid experimental evidence.
Also, we don’t yet have good candidates for how predictive processing works at the molecular and neuronal level, but we learn little about these caveats.
Even so, for those who want to know more about an important and growing field of neuroscience, The Experience Machine is an excellent primer.
0 Comments