Biology

Surprisingly slow speed of human thought calculated by Caltech

Surprisingly slow speed of human thought calculated by Caltech
Artist's impression of the brain's surprisingly slow speed of thought
Artist's impression of the brain's surprisingly slow speed of thought
View 1 Image
Artist's impression of the brain's surprisingly slow speed of thought
1/1
Artist's impression of the brain's surprisingly slow speed of thought

We all like to think of ourselves as quick-witted, but a new study from Caltech calculates that our brains process information at the extremely slow speed of around 10 bits per second. This leisurely pace may have long evolutionary roots, despite our sensory systems gathering data about 100 million times faster.

The human brain is often said to be the most powerful computer in the world, and its efficiency is undisputed. But how fast does it actually work, in computer terms? Biologists at Caltech have quantified the rate of human thought in terms of bits. Digitally speaking, one bit is a single 1 or 0, and a string of them are used to code information.

The researchers on the new study first set out to define one bit in human terms. Of course, it varies based on the specific mode of information processing – with reading and writing, for example, they quantify one bit as one character of text, and in listening to speech it’s essentially one sound.

For writing, they started with the example of a professional typist. They can type at a speed of 120 words per minute, and with an average of five characters per word that comes out to 10 keystrokes, or bits, per second. With audio, the recommended rate to make sure your speech can be understood is up to 160 words per minute. Applying the same math, this comes out to 13 bits per second.

The team goes on to calculate similar bit rates for extreme examples of human information processing. That includes solving Rubik’s cubes at world record speeds, which comes out at 11.8 bits/sec, digit memorization challenges (4.9 bits/sec), professional Tetris (7 bits/sec), and speed card challenges (17.7 bits/sec), which involve studying a randomized deck of cards and recalling the order.

From this, the team concludes that around 10 bits per second is a good average for the rate of human thought. Compared to artificial data transmission systems, that’s extremely slow – Wi-Fi speeds, for example, are usually measured in hundreds of millions of bits.

It’s even slow compared to our own hardware – sensory organs and the nervous system. The team calculates that a single cone photoreceptor in the human eye can transmit information at around 270 bits/sec, which comes out to a staggering 1.6 billion bits/sec per eye. The optic nerve seems to then compress it down to about 100 million bits/sec. However, that still dwarfs our rate of thought, especially considering the huge amounts of input streaming in from our other senses at the same time.

"Every moment, we are extracting just 10 bits from the trillion that our senses are taking in and using those 10 to perceive the world around us and make decisions,” said Markus Meister, corresponding author of the study.

Individual neurons have the capacity for much faster data rates, but in practice they usually operate at around 10 bits/sec, the team says. This could be a holdover from our most ancient ancestors, who needed to focus on just moving towards food and away from predators. As such, we can really only have one “train of thought” in the forefront of our mind at a time.

“Our ancestors have chosen an ecological niche where the world is slow enough to make survival possible,” the researchers write in the paper. “In fact, the 10 bits per second are needed only in worst-case situations, and most of the time our environment changes at a much more leisurely pace.”

The implications for future technology are a little worrying. The researchers say that computers, robots and AI can process information much faster than us, and will only get faster.

“The discussion of whether autonomous cars will achieve human-level performance in traffic already seems quaint: roads, bridges, and intersections are all designed for creatures that process at 10 bits/sec,” the team writes. “When the last human driver finally retires, we can update the infrastructure for machines with cognition at kilobits/sec. By that point, humans will be advised to stay out of those ecological niches, just as snails should avoid the highways.”

It’s also bad news for anybody hoping to augment their cognitive abilities with AI, through systems like Elon Musk’s Neuralink. No matter how fast the computer half gets, our squishy, antique hardware will still be the bottleneck.

“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/sec,” the paper reads. “Instead of the bundle of Neuralink electrodes, Musk could just use a telephone, whose data rate has been designed to match human language, which in turn is matched to the speed of perception and cognition.”

The study does require a little bit of number-fudging though. Brain bits and computer bits aren’t perfectly comparable – for example, systems like ASCII take seven bits to encode each character, compared to the new claim of one character per bit in the brain. Plus, it’s hard to quantify bits for different human activities.

Either way though, it’s an intriguing study that will get you thinking – just, not too fast.

The research was published in the journal Neuron.

Source: Caltech

7 comments
7 comments
Techutante
Considering we see at around 30-60 frames per second and we do it with dual inputs that are synced, and can still think and move our bodies I think the overall equation is probably wrong. Plus we only need 800 calories a day on average to run it all. That's about 3300 watts of power in a day to keep us running, think, see, and move. My GPU is using 350 watts of power a hour give or take. Not counting the CPU and the RAM and whatnot. I dunno what I'm saying, how do any of these numbers relate even?
JøhP
This is a good example of bad science.
What they are measuring is, of course not, the performance of the entire brain, but only on the small shallow sliver of the mind, we call consciousness.
The rest of the mind, the subconsciousness, the reptile brain and all the other parts are what is doing the real processing and making all the decisions. Those parts are performing processing comparable to today's supercomputers at a fraction of the energy expenditure.
The only decisions the consciousness will ever be able to take, if there is time, is whether to veto a decision made in other parts of the brain before it is executed.
These are all well established fact from actual relevant research from the last 50 years.
see3d
Thinking is an enormously parallel process in the human mind. The results of this massive computer generate new output states at an average of about ten times a second. That is not a single bit but millions of analog control signals operating in parallel! However, it is a good rate to respond to our environment with the size and mass of our physical bodies. Brains can operate at a much higher speed if it makes economic sense for the organism. For instance, a hummingbird, with its low mass and high-speed actions, requires a much higher brain processing rate to survive. It comes at a cost of a much higher energy consumption per gram of weight. I suspect a giant tortoise would be at the other end of the scale.
I tend to "think" in pictures. How many "bits" does it take to generate a 3D picture movie in my mind as I work on a problem? It is megabits. Ten bits fails the common sense test.
Global
I must be running a quantum computer in my noggin, being able to catch a spray bottle that had started to topple off my bench as I was focused & crouched cleaning out the litterbox on the floor, yet managed to hear, & see out of the peripheral vision, & react with my right hand, arm and fingers perfectly caught the bottle before it hit the floor. Milliseconds reaction time to process, & react...10 bits per second HAHAHAH
paul314
It's more along the lines that conversion to the kinds o I/O that the researchers want to measure is extraordinarily inefficient. If you forced a computer to communicate by controlling a robot that bashed a teletype keyboard at 110 baud it wouldn't look very efficient either. Even if we just go from typing to speech, we find that (some) people can reliably produce up to 5 words per second, at somewhere between 5-10 bits average per word (said Shannon) which would mean 25-50 bits per second. And of course (some) people can reliably read upwards of 25,000 words per minute, which is more in the neighborhood of 400-800 bps. So it's all about how you choose to measure.
(And of course humans can recognize some images at multiple frames per second when it would take a computer trillions of operations to achieve the same thing from scratch....)
lon4
I tend to agree with the comments of Techutantea and See3d that visual processing is faster than verbal processing. Remember the old adage that one picture is worth one thousand words. As a Design Engineer 99% of my thought process is visual. Employing my imagination to visualize what doesn’t exist. And to make a movie in my mind that lays out the steps to create that vision in reality. 10 Bits - BS!
MCG
It's the ones who process at 10 bits that complain the loudest lol. Jokes aside, the measurements in this article seem to equate to our ability to crunch language in our brains, but as one mentioned here, we can perceive beyond language.