Lessons from a digital rat brain

There is nothing mysterious or miraculous about the brain.

It doesn’t sound like much, but it’s a huge step. Painstaking analysis of a small part of the brain of a juvenile rat has enabled scientists to re-create that part inside a computer. What is more, the behaviour of the digital brain sliver shows high fidelity to the original and tiny tweaks produce realistic effects.

The first lesson here is that there is nothing mysterious or miraculous about the brain. You don’t have to do things like invoke the strange quantum behaviours of its constituent atoms and molecules to get the same outputs as you would from real brain tissue. Understanding and re-creating the characteristic properties of the brain is a challenge because of the organ’s complexity. But it is an engineering, not a metaphysical challenge. When it comes to the brain, there is no need for superstition.

The work started with experiments to measure the properties of real neurons, map out their interconnections using fine slices of rat brain and differentiate between the various subtypes of neuron involved. It doesn’t help that not all neurons are equal. The researchers, based in Lausanne, had to deal with 207 different types of neuron in their reconstruction. In total, they modelled 31,000 neurons and all their 2,000 interconnections. That involved
characterising 40 million of the synapses – the interconnection hardware – a task that has taken 82 scientists 20 years.

If the numbers in this small project are mind-boggling, that is only the beginning. A typical rat’s brain contains about two hundred million neurons, with a few hundred billion synapses. A digital reconstruction of the entire rat brain is a daunting task. But it is a task worth attempting, if the results of the Lausanne study are anything to go by.

Once the simulation was complete, researchers were able to switch this small section of virtual brain on and see how the neurons fired. What they saw reflected real-world neural activity. They could make the cells fire in sync, for instance, mimicking some observed firings in sleep. But they could also change the resources available to the neurons, such as the number of calcium ions, a vital ingredient in neural firing processes. This produced effects, seen in the real brain, that computer models of individual neuronal activity could not predict. The brain, in other words, is more than the sum of its parts. But not mysteriously so: you just have to build it to see it start to work.

As this work progresses we will achieve an ever-clearer understanding of how brains process signals from the senses and integrate them with other facets of our neural processing. We may even learn how they create a conscious experience. This is remarkable. Understanding conscious experience once seemed impossible: it’s why the Australian philosopher David Chalmers coined the phrase “the hard problem” to describe the scientific exploration of consciousness. But given the experimental explorations, digital simulations, technological breakthroughs and Rottweiler-like levels of tenacity we have achieved so far, there is no reason to think we can never solve the mystery.

We are mapping the brain at ever-higher resolution and are now able to discern changes in a single neuron that show memories being encoded. At some point soon – maybe in just a few decades – we are likely to build something in a computer that is probably conscious. It’s not something to worry about; it is something to welcome. We will finally overcome the last of our superstitions.