by Dr. Randal A. Koene
(This is a work in progress.)
SIM, the concept of substrate-independent minds. We call a mind substrate-independent when its self-same functions that carry out thinking processes can be implemented through the operations available in a number of different computational platforms. For example, if we can carry out the function of a mind both in a biological brain and in a brain that is composed of computer software or neuromorphic hardware (a hardware architecture with design principles based on biological neural systems), then that mind is substrate-independent. The mind continues to depend on a substrate to exist and to operate, but there are substrate choices.
The goal of substrate-independence is to continue personality, individual characteristics, a manner of experiencing and a personal way of processing those experiences. Your identity, your memories can then be embodied physically in many ways. They can also be backed up and operate robustly on fault-tolerant hardware with redundancy schemes. Achieving substrate-independence will allow us to optimize the operational framework (i.e., the hardware) to challenges posed by novel circumstances and different environments. But where did all this start? How did this particular concept of SIM and the organized efforts seeking to accomplish it come about?
The term Substrate-Independent Minds (SIM) is fairly recent. Many of the core ideas have been previously referred to by the terms mind uploading or off-loading. Unfortunately, those older terms have also led to some confusion, especially among those new to the ideas. Why "uploading", why not "downloading" or "off-loading"? I have heard all three terms used with regards to memory. And in each case, those terms really only describe the act of moving data from one storage location to another. Storage does not tell us anything about the use of that data. In this sense, taking a magnetic resonance image (MRI) of a brain could constitute "uploading". The most important part of the objective is of course that data is not simply recorded, but used to re-implement functions of mind. The objective is to carry on the functions of a specific mind. A re-implementation of functions must operate on some substrate, but when you can do this in a number of sufficiently powerful computational substrates then the mind has become substrate-independent in that way. We call that a substrate-independent mind, a SIM - which is the objective.
Mind uploading is a process, and that is what the term is now used for. By convention, we refer to mind uploading as the process by which that which constitutes a specific mind is transferred from one substrate (especially the biological brain) to another substrate (e.g., an implementation in-silico).
Beyond the process of moving to other substrates, we are especially interested in enhancement of the mind. We seek to achieve far greater adaptability and therefore greater competitive strength in a wide range of challenging circumstances. This goal is more than preservation or life-extension. I paid special attention to that aspect of SIM in "Pattern survival versus Genesurvival" (R.A. Koene, KurzweilAI.net, 2011). Enhancement or augmentation is where the multidisciplinary requirements for SIM intersect with research and development of brain-machine interfaces (BMI) or brain-computer interfaces (BCI). As a brain-machine interface, SIM enables increasingly intimate man-machine merger that integrates with us the capabilities of our creations.
There are numerous technological proposals for the accomplishment of SIM (R.A. Koene, International Journal of Machine Consciousness, 2012). At present count, there are at least six main tracks. Our understanding is still limited with regards to the manner in which fundamental computational elements of the brain participate in the vast interaction of concurrent processes from which mind emerges. For the vast majority of tasks that a mind can deal with, we do not understand the top-down set of strategies at each level of processing. For this reason, high-level approaches that begin with abstract assumptions about functions of mind and how those might be recorded and recreated are difficult to justify and validate.
Someday, extremely insightful methods of uploading from a biological mind to a SIM may be feasible and well-supported. And it may become possible to convert or compile functions of mind that were generated by processes in the neurophysiology directly into a form that is optimized to make use of the features of the target platform, while still achieving at least equivalent and satisfactory re-implementation of the specific mind. That is not yet feasible.
At present, nearly everyone who is actively working in the field of SIM and in closely related domains takes a much more conservative approach. That approach is to faithfully re-implement by emulating the basic computational functions carried out by elements of the neurophysiology, while at the same time faithfully re-implementing the connectivity as it exists between those elements in the neuroanatomy. The problem is decomposed into much smaller physical pieces for which so-called system identification must be feasibly carried out. At that level, there are still suppositions about scope and resolution that need to be tested. E.g., do ensembles of neurons, individual spiking neurons, morphologically detailed neurons, or molecular processes in synaptic channels attain the requisite resolution for emulation? Still, the acquisition and recreation of function and structure are feasible with understanding at a level within reach of current neuroscience, and with tools that we can construct today. In 2000, I named this approach descriptively as Whole Brain Emulation (WBE). The term caught on and was eventually adopted by related research aspirations, where it is sometimes abbreviated to "brain emulation" if the whole brain is not the scope of a project.
On the other tracks to SIM, significant practical work is also taking place to develop neural interfaces and brain-machine interfaces. Such interfaces, may provide a path of “Augmentation up to SIM”. It is an interesting possibility, because that path is highly incremental. We have always augmented ourselves with tools, and not just by relying on computer networks and smart phones. Today, we already see that a prosthesis can help an individual outperform the unaugmented. This is the case with Oscar Pistorius, the famous South African runner who has "Cheetah blades" instead of legs. As in that example, there is a case to be made for BMI as an approach that benefits from “market pull”.
Imagine introducing simple, non-invasive augmentative technology that lawyers can use in the court room. When that technology gives lawyers who have it an edge, there will be demand. Other professionals, such as those working on stock markets, may then notice the appeal so that the market for augmentation expands. In addition to breadth, competition in the market can lead to a need for higher bandwidth integration and interfacing. Once there, you begin to bump into the same questions that need solving for technological advances that achieve SIM.
The core idea, that the significant aspects of a person's mental life can persist when properly transferred from body to machine, from machine to machine, or from body to body has been around for a long while. In those early forms, it was a fancy and a fantasy, the purview of magical transformations. With the scientific renaissance, modern philosophy and psychology, thought experiments along the same lines became more refined, and by the middle of the 20th century, serious science fiction writers were incorporating in their stories some ideas that quite closely resemble the current conceptions of mind uploading, whole brain emulation and substrate-independent minds.
My own thinking was influenced by reading "The City and the Stars" (A.C. Clarke, 1956), a 1956 re-write by science fiction legend Arthur C. Clarke of his first novel. In the story, the inhabitants of the city Diaspar take turns having 1000 year intervals of active life between periods of stasis in which they are stored as data patterns in the city's computer. Either shortly before or shortly after I read that story, sometime in 1984, was when I decided that substrate-independent minds would be the objective I needed to achieve first. (According to my thinking at the time, the next mission would require control of matter in the fashion that is now known as nanotechnology and molecular manufacturing.) I have explained the reasons for this elsewhere (R.A. Koene & S. Olson, 2011).
In the early 1990's, the rise of the Internet began to facilitate the formation of on-line interest groups and communities that would otherwise have had a very difficult time finding their peers on a global scale. One of these interest groups coalesced around the concept of mind uploading. A collection of web pages was maintained by Joe Strout, which was most notable for practical ideas about the reconstruction of brains by building compartmental models of neurons from structural scans. Joe Strout, at UCSD at the time, also established a mailing list called the mind uploading research group (MURG).
It is when I discovered and joined that mailing list in 1994 that I first realized that I was not alone. There were some early fellow travelers with the same destination, who understood the difference between impossible projects and ambitious projects, and who were willing to devise practical plans and dedicate their efforts to the necessary actions. The archives of the MURG list and some of the pages that I have retained on the site http://minduploading.org give insight to this period. It is during the early years of that web site and my tenure as curator of the MURG list that Whole Brain Emulation was coined for clarity and in an effort to move from good ideas to feasible projects. The focus on whole brain emulation also distinguished our efforts from approaches that shared many of the same ultimate ambitions, but had different methodological philosophies or different criteria for success (e.g., the Terasem Movement).
In 2007, the Future of Humanity Institute at Oxford University, and in particular Nick Bostrom and Anders Sandberg (a former computational neuroscientist) began to take a serious interest in Whole Brain Emulation. The first Whole Brain Emulation Workshop was convened at Oxford University. The result of this workshop was a technical report on the feasibility of WBE, a first attempt at a roadmap of sorts (A. Sandberg & N. Bostrom, 2008). The report already included key technologies such as the Knife-Edge Scanning Microscope (KESM), the Automatic Tape-Collecting Lathe Ultramicrotome (ATLUM, now called an ATUM) and functional recording work by Peter Passaro.
Thinkers from outside the domains of neuroscience, physics, and engineering wrote about future possibilities that included WBE from the social/philosophical (D.J. Chalmers, 2010) and economic (R. Hanson, 1994) perspectives. Around the time of the 2009 Singularity Summit in New York, I was working on the organization of systematic efforts to bring together the key pieces needed to achieve WBE and SIM. During the summit workshop, we were able to put forward WBE as a transformative technology to be considered and contrasted with Artificial General Intelligence (AGI) and its potential existential risks.
In 2010, WBE was for the first time included in the annual conference on Artificial General Intelligence (Lugano, Switzerland). There, I teamed up with Dr. Suzanne Gildert (D-Wave, Vancouver) and we revamped the organizational network around the multidisciplinary research and development efforts toward WBE. We introduced Substrate-Independent Minds (SIM) as a well-defined objective, employing the acronym ASIM for Advancing Substrate-Independent Minds to indicate the purpose of our new action-oriented organization carboncopies.org.
As of this writing, most of those working on SIM are focused on research and tool development aimed at the initial challenge to gain sufficient access to the biological human brain. In those efforts, we now include the world-class expertise of Jeff Lichtman, Ted Berger, Henry Markram, Sebastian Seung, Ed Boyden, George Church, Anthony Zador, Konrad Kording, Clay Reid, their laboratories and many others.