One of the great debates in neuroscience is how neurons encode information that is sent to and from the brain. At issue is whether the information is sent in digital or analog form or indeed whether the brain can process both at the same time. That’s important because it can change the way we think about how the brain works.
But solving this question isn’t easy. The digital signals used by conventional computers are entirely different from the analog signals used in devices such as old-fashioned TVs and radios. That makes them easy to distinguish,
But the same can’t be said of neural signals, where digital and analog signals are hard to tell apart. So a useful step forward would be a way to distinguish between neural signals that are analog and those that are digital.
Today, Yasuhiro Mochizuki and Shigeru Shinomoto at Kyoto University in Japan say they’ve come up with just such a technique. And these guys have used it to distinguish between analog and digital signals in the brain for the first time.
Neuroscientists have long known that neurons carry signals in the form of electrical pulses that they call action potentials or spikes. A series of these is known as a spike train.
Exactly how information is encoded in a spike train isn’t known but researchers have discovered at least two different encoding protocols. In the 1990s, neuroscientists found that the way a muscle becomes tense is determined by the number of spikes in a given time interval, the rate which they arrive. This kind of signal is either on or off and so is clearly digital.
But other neuroscientists say that information can be encoded in another way–in the precise timings between single spikes as they arrive. This is analog encoding.
The difficulty is in telling these two apart since they both depend on the pattern of spikes that travel along a neuron. And that causes much dispute in the neuroscience community because nobody agrees on when a signal is analog or digital.
Now Mochizuki and Shinomoto have come up with a way to automatically distinguish between these types encoding. Their approach is based on the idea that some statistical models are better at representing digital codes than analog ones and vice versa.
For example, an approach known as empirical Bayes modelling is specifically designed to simulate analog signals. By contrast, hidden Markov modelling is particularly good at capturing the properties of digital codes.
Mochizuki and Shinomoto’s idea is to exploit the strengths of each method to determine whether a neuronal signal is digital or analog.
Their method is straightforward. They analyse a neuronal signal and then try to reproduce it using the empirical Bayes model and then using the hidden Markov model. They then decide whether it is digital or analog depending on the model that best simulates the characteristics of the original signal.
So if the empirical Bayes model best simulates the signal, it must be analog. And if the hidden Markov model triumphs then the signal must be digital.
These guys have tested their approach by analysing the signals produced in different parts of the brains of long-tailed macaques. And they say their approach indicates that different parts of the brain rely on different forms of encoding. “Fractions of neurons exhibiting analog and digital coding patterns differ between the three brain regions,” they say.
That’s an interesting discovery. If their method proves sound, it could finally help to settle the question of how the brain encodes information to do different tasks. And it could also help engineers build chips that can recreate these kinds of signals to make better interfaces between humans and machines and even to replace nerve function once it has been irreparably damaged.
There’s certainly more to come on this topic. But in the meantime, interesting stuff!
Ref: arxiv.org/abs/1311.4035: Analog And Digital Codes In The Brain
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
Data analytics reveal real business value
Sophisticated analytics tools mine insights from data, optimizing operational processes across the enterprise.
Driving companywide efficiencies with AI
Advanced AI and ML capabilities revolutionize how administrative and operations tasks are done.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.