Today, many scientists are aiming to mimic the functionalities of the human brain down to its littlest details. Why? Many of them believe that unlocking the mysteries that surround the human brain and its functionalities is key to resolving many of the ailments that affect the brain, such as Parkinson’s disease and many other illnesses.
The Blue Brain Project
One of the most popular projects that aim to simulate the human brain is called the Blue Brain Project. Its goal is to make brain reconstruction possible by reverse-engineering or digitally deconstructing the brain’s circuitry to know how it functions.
The Blue Brain Project began in 2005 under the Brain and Mind Institute of the École Polytechnique Fédérale de Lausanne (EPFL) and spearheaded by Henry Markram, the neuroscientist responsible for the Human Brain Project.
After at least 15 years, the project team successfully released their first-ever digital 3D brain atlas that includes insightful information about neurons and their locations in the many regions of the brain. In one of his talks, Idan Segev, who is among the computational neuroscientists involved in the project, said they are close to testing the entire cortex of a rodent brain through virtual electroencephalography (EEG) experiments or using the electrophysiological monitoring method to record the electrical activities of the brain. The problem? The supercomputers currently in use for the project can no longer run the model.
Challenges Encountered in Brain Simulation Projects
Despite years spent on these projects, actual brain simulation remains merely an idea, a goal that may prove too far-fetched. Why? There are several reasons that include:
Limited Scalability Capacity
The human brain contains billions of neurons, each of which fires thousands of synapses simultaneously. Duplicating all these neurons and placing them in a computer would push the boundaries of even the supercomputers we have right now, as Segev pointed out. Simulating this important functionality of neurons would mean building an exascale computer capable of calculating at least 1 quadrillion operations within a second.
So far, the most extensive neural simulation that computers can do has a capacity of only 4 million neurons. As such, their full reconstruction capacity can only hold 31,000 biophysical models of rodent cortical neurons, which has 36 million synapses.
Complexity of the Brain
Duplicating the entire brain means coming up with an unlimited set of parameters since each and every detail and component must be faithfully incorporated into models. That includes all the brain’s molecular processes and extracellular interactions.
Currently, most brain simulation projects, such as the Blue Brain Project or Human Brain Project, for example, are developing extensive databases of cell types and their corresponding species. However, it can be challenging to complete data gathering in a non-invasive way, making it impossible to achieve for the human brain.
While there are brain complexities that are difficult to duplicate, proponents of brain simulation hope that merely uncovering the primary principles of brain functions would be enough to program algorithms. However, identifying which features can be used remains unclear.
Keeping Up to Speed with Processes
A human brain can take years to develop fully. Even the learning process can span decades. And developing a technology that can match these large-scale simulations in real-time, if not faster, is non-existent at the moment. Models that currently run ongoing projects do so at a slower pace. Achieving this means working on improving supercomputing.
While quantum computing might work, looking at neuromorphic computing is also promising since it utilizes analog circuitry, which can copy neural architectures in great detail. Investing in developing these technologies can somewhat make models work faster than real-time. But there is more to just the speed of carrying out processes. The system used must also be able to deal with complicated procedures to simulate learning processes.
Integration Requirements
To develop models that factor in brain-wide networks or regions in the brain that show functional connectivity, proponents must come up with smaller representations of brain regions that can be integrated. These are essential to grasp the overall functionality of the brain better—how it can process synapses in real-time, achieve flexibility, and do everything with high efficiency. The major stumbling block here is lack of an iron-clad theory of how the brain works.
That said, several aspects of the mind like consciousness, understanding, and agency will never be fully captured by even the most sophisticated digital brain simulations because of poor representation of consciousness that can limit full understanding.
Will We Ever Achieve Brain Simulation?
While many see the brain as a system full of electrical spikes that facilitate communication and encoding, there is so much more to it that remains unknown. And while we have yet to achieve full brain simulation, the possibility is still out there.
Scaling algorithms with supercomputers may make full simulation possible. When this happens, it can change the course of how we provide healthcare. We will better understand brain disorders. Artificial intelligence (AI) research and work on neural networks will also change in terms of viewing deep learning.
—
For years, researchers and scientists have been working tirelessly to simulate the human brain. Initiatives like the Blue Brain Project are a step toward that direction.
If you are interested in learning more about Blue Brain Technology, you check out our article What is Blue Brain Technology.
