Technology and Research
"Computing power doubles approximately every two years."
- Moore's Law
Although downloading consciousness is still only the stuff of science fiction, recent research has led scientists to claim that an artificial brain could be constructed in as little as ten years (Fildes, 2009). One such study, led by Henry Markram and his team at the Blue Brain project, has already successfully simulated elements of a rat’s neocortical column, a complex layer of brain tissue common to all mammalian species. But as promising as Markram’s research is, most scientists admit that we still have a ways to go before we can even construct a functional model of the human brain, let alone download our own consciousness into a machine. As such, this section will cover the present state of mind uploading technology, focusing mainly on brain simulations, brain mapping techniques, and other technologies that might some day turn the worlds of Frederik Pohl and James Cameron into reality.
Currently, the closest we have come to building a functional model of the human brain, a crucial step on the pathway toward downloading consciousness, is a series of cortical simulations. These simulations, which have utilized the very best in computer technology, such as IBM’s Blue Gene supercomputer, have successfully emulated the processing power of a brain possessing 1.6 billion neurons—about the equivalent of a cat brain in terms of neuron numbers alone. But even these extremely complex models, run on some of the best computer hardware in existence, lag behind the actual processing power of their biological counterparts. Among other things, this is mainly due to the inability of computers to process information in parallel by doing many calculations simultaneously. What takes a cat one second might take the Blue Gene supercomputer ten or even a hundred seconds, depending on the complexity of the task.
A Blue Gene Supercomputer.
Other barriers to constructing human-scale brain simulations include storage space and sheer processing power. A complete map of the human brain containing detailed information about each neuron and synapse would occupy about 20,000 terabytes and require 1016 flops (floating point operations per second) of processing power to function. Currently, only the world’s fastest supercomputer possesses the capability of crunching that many numbers in a second. Nevertheless, the future is bright. If computing technology continues to follow recent trends such as Moore’s law, doubling in processing power every two years, it is likely that most supercomputers will be able to run an accurate simulation of the human mind within the next few years. Additionally, as storage capacities continue to increase, it will be more feasible and economical to store digital maps of human brains. As an indicator, in 2007, the largest cortical simulation contained about eight million neurons—the equivalent of half a mouse brain—and just four years later, scientists are capable of emulating brains comprised of over 1.5 billion such structures.
In addition to supercomputers and mind-modeling software, powerful brain-scanning technologies are also at the forefront of efforts to construct virtual brains that might eventually house human consciousness. According to many neuroscientists, the human mind is really just a complex computer whose function depends on electrochemical processes. In their eyes, if we are able to sufficiently emulate the neural networks that comprise the human brain, it is only natural that intelligence and consciousness should follow. As a result, neuroscientists have turned to magnetic resonance imaging (MRI), high angular resolution diffusion imaging (HARDI), and a series of other more invasive techniques in order to construct a detailed map of the human brain.
In recent years, neuroscientists have made use of emerging MRI techniques to probe not only the anatomical structure of the brain, but also to map the very connections that link the various regions of the human mind. According to Yap, et al., functional MRI’s (fMRI’s) are being used to track blood flow and oxygen consumption within the brain, providing accurate images of neuronal circuits that are activated under different simulated behaviors (Yap, Wu, & Shen, 2010: 1). Similarly, resting state fMRI’s, which detect fluctuations in brain activity of people at rest, are being used to locate coordinated networks within the brain, providing powerful insights into the different regions of the brain. High angular resolution diffusion imaging (HARDI) is also currently being used to measure water diffusion along fibrous tissue in the brain, and allows visualization of axonal bundles—groups of long, slender nerve cell projections that conduct electrical impulses within the brain.
While noninvasive imaging has certainly provided us with a better understanding of the structures that comprise the human brain, more invasive techniques are currently uncovering the minute details of synapses—the connections that link each individual nerve cell to the next. Researchers within our very own school of medicine have applied these techniques to the mapping of mice brains, achieving an unprecedented level of detail in synapse visualization. By thinly slicing portions of a mouse’s cerebral cortex, staining each section with antibodies designed to match seventeen synapse-associated proteins, and taking extremely high-resolution photographs of the resulting sections, they have created a 3-D mosaic of synapses that can be rotated, penetrated, and navigated, all while preserving the original anatomical context of each neuron (MedicalDaily.com, 2010). Combined with existing data on higher-level structures, these emergent techniques could help us improve the granularity of our digital brain maps and bring us one step closer to creating accurate models of the human brain.
Although a brief examination of our current technology reveals that we are still years away from downloading our conscious minds into other media, recent advances in supercomputing, brain mapping, and invasive imaging techniques are certainly a cause for hope. If we are able to generate a functional model of the human brain, many scientists argue that there is no reason why these models cannot be based on the brains of specific individuals. In addition, other futuristic technologies, such as brain-computer interfaces, may provide the necessary link between minds and machines, allowing us to eventually upload the consciousness of a living human subject. For more information about these interfaces and other theoretical aspects of downloading consciousness, click here.