What are Substrate-Independent Minds (SIM)?
The functions of mind that we experience are originally implemented through neurobiological mechanisms, the neural circuitry of our brains. If the same functions are implemented in a different operating substrate, populated with parameters and operating such that they produce the same results as they would in the brain, then that mind has become substrate-independent. It is a substrate-independent mind (SIM) by being able to function in different operating substrates.
The popular term "mind uploading" can refer to the process of transfer, moving a specific substrate-independent mind from one operating substrate (e.g., the biological brain) to another.
When an operating substrate for mind functions is created by carefully replicating many of the principles of the neurobiology, physiological and architectural characteristics, then we call that approach whole brain emulation (WBE).
Concise definition of terms:
See the article 'Substrate-Independent Minds' by Dr. Randal A. Koene for an introduction and an overview of projects.
The term Advancing Substrate-Independent Minds (ASIM) was used as a mission statement for carboncopies.org in its first year and through the ASIM-2010 workshop series.
What presuppositions do we make about the brain and about the mind?
The concept of substrate-independence, here applied to functions of the mind, is nicely encapsulated by the Church-Turing thesis. This may be summarized in the following way:
Everything that is computable can be computed by a Turing machine.
The strong neuroscientific consensus is that behavior and experience, phenomena correlated with what we consider mental processes of the mind - as forms of information processing and the consequences that this elicits by interfacing with body and environment - emerge from biophysical functions that are adequately described in terms of classical physics. These information processing functions of mind (and in fact, even quantum physical processes) are computable. It follows that the mind is computable; our brains are machines that carry out the information processing and the interface with body and environment. The Church-Turing thesis implies that one Turing machine can implement another.
Of course, this has already been implicitly assumed true by all those who work on the development of neural prostheses.
Where does the concept of SIM come from?
SIM depends on developments in many disciplines. From a technical perspective, some of the foremost are neuroinformatics, neuroprosthetics, artificial general intelligence, high-throughput microscopy and brain-computer interfaces. Conceptually, there are also strong associations with applied bioinformatics and life-extension research.
The notion that the human mind is central to the experience of our existence and the realization that the brain can be understood as a biological machine have both been raised many times throughout the history of science. Following the development of computers and serious attempts to create mind-like function in artificial intelligence, there are now multiple high-profile projects directly aimed at reimplementing brain structure and functions of neurophysiology. To name the most obvious current candidates: the Blue-Brain Project, and the DARPA Synapse Project. Finally, converging developments in the areas of neural interfacing, optogenetic techniques and high-throughput microscopy, we arrive at the very real possibility to learn from and re-implement structure and function of specific brain samples.
Why would you want substrate-independent minds?
Remarkable as our brains are, it is clear that there are many limitations. We experience limited working memory, finite and unreliable long-term memory, the inability to multi-task effectively, and our sensory experience and comprehension are a small subset of what may be possible. Substrate-independence can enable us to exceed those limitations and to suitably adapt to novel environments, to explore other ways of thinking, and to experience virtual environments from a truly first-person perspective.
Furthermore, human embodiment, while in all likelihood responsible for much of the astonishing capability of our minds within that embodiment, is also increasingly a constraint on the possible capabilities of both mind and body. Our brains, evolved to suit the local environment and embodiment are not easily redirected at other possible optimizations. Consider for example, that we do not think fully even in three spatial dimensions when we build our cities. More complex geometric problems entirely elude our natural capabilities.
Similarly, our potential embodiments and the corresponding experiences are limited by the need to sustain a local environment suitable for the biological substrate of brain and body. We could fly – but we must do so within an unnecessarily complicated aircraft that is designed to incorporate and sustain human passengers. We could simultaneously interact with the marine environment of the Great Barrier Reef while also experiencing the freedom of near vacuum on the far side of the moon. But, we must always temper such pursuits for the needs of human physical embodiment.
In addition to these limitations in terms of capability or environment, we have the limitations of speed and time. Our ability to experience and act is constrained by the rate at which we can process thoughts. Time is limited by the mortality of our biology. For those who wish to improve their capabilities in these dimensions, substrate-independence offers unique potential safeguards against these limitations, while at the same time removing concerns about human over-population within the bounds of our current biosphere.
What kind of research or developments in basic neuroscience relate closely to SIM?
Neuroinformatics investigations that seek to map the detailed “connectome” of the human brain are essential for the whole brain emulation (WBE) approach to SIM. Knowledge gained in that field can be used to create a virtual brain laboratory at the scale and resolution required for SIM. Well-known examples of research in this field are the Blue-Brain Project headed by Henry Markram and large-scale brain models created by Eugene Izhikevich.
In the Blue-Brain Project, the structure and neurophysiology of the mammalian neocortical column are synthesized. Such columns are superficially highly similar arrangements of specific neural cell types with regular arrangements of local connectivity between those neurons and long-range connectivity beyond the column. The rat neocortical column, which was the first to be reverse engineered and synthesized, contains about 10,000 neurons. In humans, a neocortical column is about 2mm tall, 0.5mm across and contains about 60,000 neurons. The synthesized column is composed of compartmental neuron models, with dynamics described by the Hodgkin-Huxley equations. Where necessary, reverse engineering included studies at the molecular level.
Large-scale models by Eugene Izhikevich also included regularly reoccurring microcircuitry, implemented with spiking neuron dynamics. One second of activiity in 100 billion neurons with 1 quadrillion generated synapses was simulated by computation over 50 days. Characteristic recognizable patterns of activity were observed, such as rhythmic activity at alpha and gamma frequencies, and a propagation of clusters of activity.
What is the relationship to artificial general intelligence (AGI)?
SIM research is a subset of research into artificial general intelligence (AGI). When human intelligence is implemented in a substrate other than the human brain, then we have an artificial instantiation of the most general intelligence that is available to us today.
Investigation of procedures required for SIM are interesting to developments in AGI, because with the human brain we have a proof-of-concept system that is known to work, and by reimplementing it in another substrate we gain complete access to its functioning detail. A substrate-independent mind, or a virtual brain laboratory, provide the opportunity to test hypotheses, properly ground assumptions, and apply a clear set of initial metrics for improvement in the quest for AGI.
We can also make the argument that an AGI instantiated with the cognitive architecture of the human mind would operate and develop in a more comprehensible and potentially more desirable manner than one that was instantiated in a way that is more alien to the human mind. This makes sense from our human perspective, where we will tend to select for something that matters to us.
We may envision a substrate-independent mind, or human-instantiated AGI, eventually determining how to reconfigure itself in order to be able to tackle a greater variety of problems.
SIM may also benefit from artificial intelligence (AI) and AGI. Learning more about engineering architecturally similar intelligence can teach us valuable things about the brain, and about the significant functions and parameters that must be measured and implemented for SIM. New AI and AGI capabilities can also be tools of great value for the development of processes for SIM.
Is SIM related to ideas about "mind uploading"?
Yes. There is a strong link between the objective to achieve substrate-independent minds and the notion of a procedure for mind uploading, as that term is used in certain discussions. An important distinction is that SIM refers to an objective, and that objective encompasses any solution in which a sufficient set of mind functions can be carried out equally well in a number of different computing substrates. By contrast, mind uploading, when used in this context, commonly refers to a transfer procedure, whereby the relevant data that describes a mind's operation and information content is moved from a biological brain to some other medium. In some cases the target medium referred to is intended merely to store the data, while in others it is also intended to carry out the functions, i.e. to emulate the "uploaded" mind.
That distinction and the different uses of the term "mind uploading" are indications of the overloaded nature of the term "mind uploading". It is not always clear what is meant by uploading, and sometimes the same ideas are referred to as "downloading", "off-loading", etc. Furthermore, "uploading" has very different meaning in the conversations of different communities. The term is not at all self-explanatory to the uninitiated. For these reasons, communications by carboncopies.org attempt to use clear and explicit terminology when referring to objectives, procedures and technological implementations.
Is a mind and consciousness 'computable'?
All known evidence indicates a 'Yes'. The question is raised frequently and approached from a number of angles, sometimes in connection with matters of algorithmic computability, such as in arguments by Sir Penrose, and sometimes in connection with the notion of 'qualia', etc.
To make a case for incomputability of mind or consciousness, a first requirement would be to posit and then demonstrate at least one phenomenon of mind that reliably trips up and eludes computational mechanistic explanations. We have yet to encounter such an example.
Penrose and others have supposed a number of theoretical claims to which yet others (among them neuroscientists) have published rebuttals (see for example, 'Gaps in Penrose's toilings'; additional arguments were presented in the corresponding carboncopies Facebook thread: https://www.facebook.com/groups/carboncopies/permalink/806797076015979/).
Can a mind and consciousness move from a biological brain to another substrate?
See the very cogent and reasonable article by Keith Wiley "Response to Susan Schneider's The Philosophy of 'Her'" in H+ Magazine.
FAQ Interview by Luke Muelhauser with Dr. Randal A. Koene
A March 2014 interview with Dr. Randal A. Koene, conducted by MIRI's Luke Muelhauser and published on intelligence.org takes the shape of a carefully crafted list of Frequently Asked Questions: http://intelligence.org/2014/03/20/randal-a-koene-on-whole-brain-emulation/
(The interview is also available on carboncopies.org in PDF form.)
FAQ Article by Dr. Ben Goertzel
A number of frequently asked questions are answered in this piece that Dr. Goertzel wrote in response to an article by Goerge Dvorsky, which contained eight points reflecting common questions: http://hplusmagazine.com/2013/04/21/goertzel-contra-dvorsky-on-mind-uploading/
Interesting and recent discussions from the carboncopies Facebook group are listed on our Discussions Permalinks page.
(More questions and answers are pending! If you have specific questions that you would like to see answered here, contact us!)