One of the great endeavors of modern science is to understand the brain. This organ, the most complex machine we know, is a miracle of evolutionary biology. It processes a potent firehose of information to set goals, achieve tasks and navigate complex environments, often in ways that put the world’s most powerful supercomputers to shame. Remarkably, it weighs about the same as a bag of flour and runs on little more than a bowl of porridge.
And yet, at the heart of this amazing capability is a paradox, say Jieyu Zheng and Markus Meister at the California Institute of Technology in Pasadena. Human senses pump information into the brain at an impressive rapid rate and yet the information that comes out in the form of language and actions is vastly slower. It’s as if opening the floodgates at the Hoover Dam released nothing more than a dribble.
How come? Zheng and Meister explore this question and say it represents a profound problem of neuroscience that is ripe for experimental investigation. Their call to action has important implications for our understanding of the way the brain works, for the nature of neural networks and for practical applications such as assistive technologies for the blind and for brain computer interfaces in general, such as Elon Musk’s Neuralink technology.
Information Paradox
Zheng and Meister set out the nature of the paradox by reviewing work that has measured the rate at which the brain receives and processes information. It turns out that our senses capture huge amounts of data.
For example, they say that photoreceptors in the eye convert the amount of light they receive into a continuously varying membrane potential at a rate equivalent to about 270 bits per second. And since a single eye contains about six million receptors, that’s a total data rate of 1.6 gigabits per second.
This information passes into the nervous system, which works in a rather different way. Neurons transmit information via action potentials or spikes, which each carry 2 bits. Neurons fire at rates up to 50 times a second and so can achieve rates of around 100 bits per second.
The optic nerve contains about 1 million neurons, so the eye sends information to the brain at a rate of around 100 megabits per second. In other words, the visual system has already begun to compress this data. “The circuits inside the retina already compress the visual signal substantially, by at least a factor of 10,” say Zheng and Meister, adding that visual cortex processing takes this process even further.
That raises the interesting question of how much compression has taken place by the time this information reaches the level of perception. Put simply, how quickly do we think?
Zheng and Meister turn to various experiments that measure the information output of the brain. These range from studies examining the data throughput involved in language, touch typing, video gaming, memory tasks and so on.
In all these scenarios involving perception and action, the brain processes information at more or less the same rate. “These studies span the better part of a century with a remarkably concordant result: Humans operate at an information rate of about 10 bits/s,” they say.
That’s an extraordinary result. It implies that the rate of data perception is tiny compared to the rate at which we encounter data in the environment. Zheng and Meister give the example of a home WiFi network, which operates at 100 megabits per second to stream Netflix shows, even though our brains will never absorb more than 10 bits per second of that stream.
The hugely limited rate at which we absorb information has important implications for the total amount we can store. The researchers point out that if the human brain absorbs data at 10 bits per second, then in 100 years, operating 24 hours a day, it can soak up no more than about 4 gigabits.
Of course, DNA encodes information that determines the structure and functioning of the brain. But even taking this into account adds less than another gigabit. “After including the information from the genome, this still fits comfortably on the key-chain thumb drive you carry in your pocket,” they conclude.
The researchers have a sobering message for Elon Musk and his attempts to increase the bandwidth from the brain by creating a direct connection with the outside world. They point out it’s easy for people to imagine that their internal monologue is far richer than they are able to communicate. But this is an illusion.
The brain perceives the world at a rate of about 10 bits per second, which in theory allows 2^10 possible thoughts per second. “Because we could engage in any one of the 2^10 possible actions or thoughts in the next second, it feels as though we could execute them all at the same time,” say Zheng and Meister.
Data Illusion
This is an illusion known as subjective inflation. In reality, each thought occurs in sequence. “Based on the research reviewed here regarding the rate of human cognition, we predict that Musk’s brain will communicate with the computer at about 10 bits/s,” say the pair.
If they are right, the implications for assistive technologies are clear. Synthetic retinas, for example, attempt to pump information into the nervous system at gigabit rates. “As one might have predicted, that approach has been spectacularly unsuccessful: After decades of efforts, all the implanted patients remain legally blind,” say Zheng and Meister.
Instead, a better approach might be to feed preprocessed information into the brain at a rate it can easily process, in other words at about 10 bits per second. This might include the identity and location of objects and people in a scene, in a way that allows interaction, obstacle avoidance and so on. “This can be done comfortably using natural language: A computer translates the visual scene to speech in real-time and narrates it to the user according to their needs,” say the researchers.
Similar arguments can be made for technologies for assist people who are paralyzed or disabled in other ways.
That’s important work that the researchers hope will inspire a new generation of experiments to better characterize how the brain operates in this way. And it raises numerous exciting, unanswered questions. For example, how fast other species experience perception, how best to exploit our newfound limits on cognition and why the brain perceives information sequentially rather than in parallel. As Zheng and Meister put it: “Why is cognition restricted to one task at a time, rather than pursuing many strands – potentially thousands to millions – all in parallel?”
Beyond all this is a phenomenon that even Zheng and Meister dare not mention—the phenomenon of consciousness. This term is so laden with controversy that the authors have studiously avoided even mentioning it. But if little else can be said for certain about it, consciousness is surely a phenomenon of information and the rate at which it is processed.
The paradox at the heart of this paper – that the experience of ‘being’ is extraordinarily slow compared to the firehose of data that informs it – must surely offer a promising avenue of exploration for approaching consciousness.
Ref: The Unbearable Slowness of Being : arxiv.org/abs/2408.10234
