Kurzgesagt – In a Nutshell
Sources – Mind Upload
We want to thank the following experts for their valuable input:
Dr. Anders Sandberg
Computational Neuroscientist at Oxford University
Dr. Christof Koch
Chief Scientist & President of the Allen Institute for Brain Science
– Mind is one of these words that are really hard to define. It is thought to be the collective abilities of your consciousness and intelligence, the thing that lets you imagine, recognize and dream.
#Mind, Encyclopaedia Britannica, retrieved 2020
https://www.britannica.com/topic/mind
Quote: “Mind, in the Western tradition, the complex of faculties involved in perceiving, remembering, considering, evaluating, and deciding.”
#Mind, American Psychology Association, retrieved 2020
https://dictionary.apa.org/mind
Quote: “Broadly, all intellectual and psychological phenomena of an organism, encompassing motivational, affective, behavioral, perceptual, and cognitive systems; that is, the organized totality of an organism’s mental and psychic processes and the structural and functional cognitive components on which they depend.”
– Mind uploading is the hypothetical concept of making a copy of this inner world and transferring it into a computer to run a simulation of your consciousness.
#Minduploading.org, retrieved 2020
Quote: “Mind uploading is a popular term for a process by which the mind, a collection of memories, personality, and attributes of a specific individual, is transferred from its original biological brain to an artificial computational substrate. Alternative terms for mind uploading have appeared in fiction and non-fiction, such as mind transfer, mind downloading, off-loading, side-loading, and several others. They all refer to the same general concept of “transferring” the mind to a different substrate.”
– Assumption 1: Your mind is in your brain’s structure, arrangement and biochemistry. The idea that everything about the mind can be found in the brain is called physicalism and it keeps our discussion within the domain of natural law.
#The Identity Theory of Mind, Stanford Encyclopedia of Philosophy, 2008
https://stanford.library.sydney.edu.au/archives/sum2008/entries/mind-identity
Quote: “The identity theory of mind holds that states and processes of the mind are identical to states and processes of the brain. Strictly speaking, it need not hold that the mind is identical to the brain.[...] Consider an experience of pain, or of seeing something, or of having a mental image. The identity theory of mind is to the effect that these experiences just are brain processes, not merely correlated with brain processes.”
#The presentation of the mind-brain problem in leading psychiatry journals, Brazilian Journal of Psychiatry, 2018
https://www.scielo.br/scielo.php?script=sci_arttext&pid=S1516-44462018000300335
Quote: “According to physicalism, mind is a material or physical process, a product of brain functioning. In contrast, nonphysicalism claims that mind is something different from, and may exist beyond, the brain.”
–Assumption 2: At some point, we will understand the brain well enough and possess the technology to simulate all of its aspects to make a digital mind copy.
#Uploading to Substrate-Independent Minds, Randal A Koene, 2013
https://www.researchgate.net/publication/300463595_Uploading_to_Substrate-Independent_Minds
Quote: “Accomplishing SIM [Substrate-Independent Minds] is a problem that researchers in neuroscience and related fields can grasp and simplify into sensible granular parts. To our present knowledge, there are no aspects of the problem that lie beyond our physical understanding or beyond the ability to engineer solutions. As such, it is a feasible objective and one that can be dealt with through a hierarchy of projects and by the allocation of such resources as are needed to carry out the projects within the time-span desired. If that time-span is the span of a human life or a career then we should carry out project planning and resource allocation accordingly. It is possible.”
– Assumption 3: Computer software can host your mind. Which means the mind is computable. There is no physical property in the brain, including consciousness, that cannot be simulated accurately, even if it requires a lot of code.
#An argument for the scientific and technical plausibility of mind uploading, Brain preservation Foundation, 2015
Quote: “ All of our current theories of the human mind are computational and imply
that we are like a program in this sense –we can in principle be copied and can have many
‘instantiations’ running simultaneously. ”
#The Computational Theory of Mind, Stanford Encyclopedia of Philosophy, 2020
https://plato.stanford.edu/entries/computational-mind/#ClaComTheMin
Quote: “ According to CCTM [Classical Computational Theory of Mind], the mind is a computational system similar in important respects to a Turing machine, and core mental processes (e.g., reasoning, decision-making, and problem solving) are computations similar in important respects to computations executed by a Turing machine. These formulations are imprecise. CCTM is best seen as a family of views, rather than a single well-defined view.”
– Around 100 billion neurons are communicating via one million billion connections, that are sending signals 5 to 50 times each second, about one quadrillion of events every second of your waking life!
#Brain Facts and Figures, University of Washington, retrieved 2020
http://faculty.washington.edu/chudler/facts.html
Quote: “ Average number of neurons in the brain = 100 billion”
Quote: “ Number of synapses for a "typical" neuron = 1,000 to 10,000”
#Frequency Coding in the Nervous System, PhysiologyWeb, 2014
Quote: “Physiologically, action potential frequencies of up to 200-300 per second (Hz) are routinely observed. Higher frequencies are also observed, but the maximum frequency is ultimately limited by the absolute refractory period. Because the absolute refractory period is ~1 ms, there is a limit to the highest frequency at which neurons can respond to strong stimuli.”
Note: Up to 1000 signals per second can be sent, but the average is far lower.
#Synaptic Energy Use and Supply, Julia J.Harris et Al., 2012
https://www.sciencedirect.com/science/article/pii/S0896627312007568
Quote: “ As an example of the energetic cost of information transmission through synapses, if we set typical physiological values of s = 0.01 (implying a firing rate of S = 4 Hz) and p = 0.25, Equation 3 states that, out of the 32 bits/s arriving at the synapse, 6.8 bits/s are transmitted, and from the estimate by Attwell and Laughlin (2001) of the underlying synaptic energy cost”
Note: Based on the energy cost of creating a signal, having 4 signals per synapse on average is a reasonable estimate.
And it is not just neurons, there are billions of supporting and immune cells, of various types performing different jobs.
#Cells of the Brain, DANA Foundation, 2019
https://www.dana.org/article/cells-of-the-brain/
Quote: “ The brain is a mosaic made up of different cell types, each with their own unique properties. The most common brain cells are neurons and non-neuron cells called glia. The average adult human brain contains approximately 100 billion neurons, and just as many—if not more—glia. Although neurons are the most famous brain cells, both neurons and glial cells are necessary for proper brain function.”
On a macro level the brain can be divided into sections with different roles, from breathing and heart rate to coordinating movement and involuntary reflexes. The most developed parts, the neocortex or the outermost layer of the brain, hold memories, our ability to plan, think and imagine, hope and dream.
#Anatomy of the Brain, Mayfield Clinic, 2018
https://mayfieldclinic.com/pe-anatbrain.htm
– Where exactly the “you” part of your brain is situated is not entirely clear. We know that areas like the precuneus cortex have the greatest influence on our consciousness but also that several areas can network together to share tasks none of them can do alone.
#Core Concept: Resting-state connectivity, Helen H. Shen, 2017
https://www.pnas.org/content/112/46/14115
Quote: “Intrigued by what the brain might be doing during supposedly inactive periods, Raichle and others began to explore this so-called “default mode network,” which seemed to be involved in high-level cognitive processes, such as self-awareness and memory. ”
Note: There are brain functions that rely on networks between multiple regions of the brain, challenging previous conceptions of specific areas having specific roles.
#The precuneus: a review of its functional anatomy and behavioural correlates, Andrea E. Cavanna and Michael R. Trimble, 2006
https://academic.oup.com/brain/article/129/3/564/390904
Quote: “However, recent functional imaging findings in healthy subjects suggest a central role for the precuneus in a wide spectrum of highly integrated tasks, including visuo-spatial imagery, episodic memory retrieval and self-processing operations, namely first-person perspective taking and an experience of agency.”
– The brain’s building blocks are not exactly simple either. Neurons are not just wires, they alter and process information. Synapses, where signals are handed over from one neuron to the next, contain receptors for hundreds of chemical signals, opening them up to outside influence.
#Neurotransmitters, Synapses, and Impulse Transmission, Lodish H et Al, 2000
https://academic.oup.com/brain/article/129/3/564/390904
Quote: “Many neurons secrete neuropeptides, a varied group of signaling molecules that includes endorphins, vasopressin, oxytocin, and gastrin. Neuropeptides are stored in a different type of vesicle than classic neurotransmitters. Exocytosis of both types of transmitter is triggered by a localized rise in cytosolic Ca2+, but neuropeptides are released outside the synaptic zone.”
Note: Neurons communicate with more than just electrical signals. Signalling chemicals like neuropeptides are an example of non-electrical communication.
#Nerve Cells Structure and Function, Adaptations & Microcopy, MicroscopeMaster, retrieved 2020
https://www.microscopemaster.com/nerve-cells.html
Note: Overview of the many parts and functions of the neuron cell bodies.
– We have a basic understanding of how these work, and we can accurately predict their behaviour at small scales, but there’s a lot more to the brain than just nerve signals. Hormones play a huge role, like serotonin which affects our mood or histamine which helps us learn. The brain is influenced by our other parts too, from heart nerves to gut bacteria.
#Hormonal Influences on Cognitive Function, Siti Atiyah Ali et Al., 2018
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6469458/
Quote: “The hippocampus is highly sensitive to changing levels of glucocorticoids, and fluctuating levels of this hormone are highly associated with hippocampal atrophy, Alzheimer’s disease and major depressive disorders. In short, intense stress levels do appear to have a negative impact on memory ability.”
Note: Hormones are an indirect method for the body to influence the brain. Without taking them into account, we could be missing a critical piece of our computer model of a mind.
#The Gut-Brain Axis: Influence of Microbiota on Mood and Mental Health, Jeremy Appleton, 2018
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6469458/
Quote: “The gut-brain axis is a bidirectional communication network that links the enteric and central nervous systems. This network is not only anatomical, but it extends to include endocrine, humoral, metabolic, and immune routes of communication as well. The autonomic nervous system, hypothalamic-pituitary-adrenal (HPA) axis, and nerves within the gastrointestinal (GI) tract, all link the gut and the brain, allowing the brain to influence intestinal activities, including activity of functional immune effector cells; and the gut to influence mood, cognition, and mental health.”
– To get this wildly interconnected mess of cells and meat and chemicals into a computer, we need a model that we can simulate in our digital world. Some sort of scan.
#Electron Imaging Technology for Whole Brain Neural Circuit Mapping, Kenneth J. Hayworth, 2012
https://www.gwern.net/docs/ai/2012-hayworth.pdf
Quote: “The model suggests that our unique “software" is mainly digital in nature and is stored redundantly in the brain's synaptic connectivity matrix (i.e., our Connectome) in a way that should allow a copy to be successfully simulated”
– Unfortunately our scanning technology, like fMRI machines, is not nearly good enough to attempt such a thing.
#Overview of Functional Magnetic Resonance Imaging, Gary H. Glover, 2011
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3073717/
Quote: “Resolution in fMRI is limited primarily by SNR [...]. Accordingly, the typical fMRI pixel size is 3–4 mm, although with higher field magnets (7T) a pixel size of 500 microns or less may be readily achieved”
Note: The size of neurons is between 4 and 100 microns, so an fMRI cannot distinguish individual neurons.
– But there is a different method that seems very promising: Cutting a brain into tiny slices and scanning them with a high resolution electron microscope to create an accurate map of all the cells and connections.
#Multi-Beam Scanning Electron Microscopy for High-Throughput Imaging in Connectomics Research, Anna Lena Eberle and Dirk Zeidler, 2018
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3073717/
Quote: “Multi-beam SEM technology has just recently become available and has the potential to fulfill the throughput and resolution requirements of future connectomics experiments needs.”
– In 2019 scientists successfully mapped a cubic millimetre of mouse brain - roughly the size of a big grain of sand, it contained 100,000 neurons with a billion synapses and 4 kilometres of nerve fibres. This grain of brain was cut into 25,000 slices. Five electron microscopes ran continuously for five months, collecting more than 100 million images. It took three months to assemble the images into a 3D model. The completed data set fills up 2 million gigabytes of cloud storage.
#How to map the brain, Nature, 2019
https://www.nature.com/articles/d41586-019-02208-0
Quote: “The microscopes ran continuously for five months, collecting more than 100 million images of 25,000 slices of mouse visual cortex, each just 40 nanometres thick.”
Quote: “The balloons proclaim the size of the completed data set, spelling out “2PB” (2 petabytes, which is equivalent to 2 million gigabytes) in blue and silver letters.”
– Even worse: to correctly simulate a brain, we might have to map out much smaller building blocks to include the billions of underlying proteins or even individual molecules that cause all the behaviours we see at the cellular level. Which might produce more data than the capacity of all data storage on earth.
#Whole Brain Emulation: A Roadmap, Anders Sandberg and Nick Bostrom, 2008
https://www.fhi.ox.ac.uk/brain-emulation-roadmap-report.pdf?utm_source=morning_brew
Quote: “Brain emulation needs to take chemistry more into account than commonly occurs in current computational models. Chemical processes inside neurons have computational power on their own and occur on a vast range of timescales (from sub‐millisecond to weeks). Neuromodulators and hormones can change the causal structure of neural networks, e.g. by shifting firing patterns between different attractor states”
Note: The highest level of detail, which corresponds to mapping out single molecules, corresponds to 3.14 * 10^14 Terabytes of data storage. The total data storage capacity in the world is estimated to be about 2 * 10^9 Terabytes (https://www.statista.com/statistics/638593/worldwide-data-center-storage-capacity-cloud-vs-traditional/).
– While all of these issues are annoying, the real question is how we turn the static blueprint of the brain into an active thing. Even if we have a scan, down to the level of synapses, we need laws and rules that animate the wiring diagram, to endow this static structure with life, update it with the various laws of chemical binding, of electro-dynamics to animate the simulation.
#What is a Connectome?, Brain Preservation Foundation, retrieved 2020
https://www.brainpreservation.org/content-2/connectome/
Quote: “A connectome* is the complete map of the neural connections in a brain. It is sometimes referred to as a “wiring diagram” of the molecular connections between neurons, trading on the analogy of a brain to an electronic device, where axons and dendrites are wires and neuron bodies are components”
– It all hinges on the nature of the problem:
Are the brain and mind just complicated and a lot of work to figure out? Or are they complex in a way that we can’t solve? In the worst case, consciousness is more than the sum of the parts of the brain in a way that we don’t realize yet. Complex in a way that we can’t solve by getting better scans. Just having a list of the ingredients might not be enough to get a good consciousness cake.
#Why not connectomics?, Lodish H et Al, 2000
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4184185/
Quote: “Top ten arguments against connectomics:
-circuit structure is different from circuit function
-signals without synapses and synapses without signals
-‘junk’ synapses
-same structure, many functions
-same function, many structures
-statistical synapses should suffice
-the mind is no match for the complexity of the brain
-merely descriptive neuroanatomy, just more expensive
-not much was learned from the connectomes we have
-a connectome is static”
Note: There are many arguments against mounting a huge brain mapping effort. The paper responds to each argument and suggests that matching a simple structure map with the locations of more and more detailed components (like proteins) will address many issues.
#The Big Data Problem: Turning Maps into Knowledge, Florian Engert, 2014
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4857857/
Quote: “I would suggest that the essential ingredient that turns a useless map into an invaluable resource is the experimental design employed to gather and analyze the underlying data[...] This is where the hard work is—in formulating precisely the question of what we actually want to know, what an answer would look like, and what kind of insight we can take away from the experiment.”
Note: There’s more to mind uploading than just collecting data for a map. We have to know what to do with all that data!
– Right now we have a great starting point with tangible scientific results and an end goal, but the road to true simulation is unclear and requires a lot of innovation and research. Humans have historically been horrible at predicting the pace of progress.
#The Prospects of Whole Brain Emulation within the next Half- Century, Journal of Artificial General Intelligence, retrieved 2013
Quote: “Recently, proponents of WBE (Whole Brain Emulation) have suggested that it will be realized in the next few decades. In this paper, we investigate the plausibility of WBE being developed in the next 50 years (by 2063). We identify four essential requisite technologies: scanning the brain, translating the scan into a model, running the model on a computer, and simulating an environment and body.”
– In the best case it is just a matter doing the work and finding the right solutions. It might not be necessary to simulate every last cell, down to the last atom. Instead it may be possible to simplify elements into probabilistic models that could match the behaviour of the brain using a more manageable number of simpler systems.
#Analytic Performance Modeling and Analysis of Detailed Neuron Simulations, Francesco Cremonesi et Al., 2019
https://arxiv.org/pdf/1901.05344.pdf
Quote: “In this work we have demonstrated the applicability of the ECM analytic performance model to analyze and predict the bottlenecks and runtime of simulations of biological neural networks. The need for such modeling is demonstrated by the ongoing development efforts to optimize simulation code”
#Whole Brain Emulation: A Roadmap, Anders Sandberg and Nick Bostrom, 2008
https://www.fhi.ox.ac.uk/brain-emulation-roadmap-report.pdf?utm_source=morning_brew
Quote: “Single cell‐modelling can be done on roughly five levels of abstraction: as black box modules that generate probabilistic responses to stimuli according to some probability distribution, as a series of linear or nonlinear filters of signals, as a single compartment with ionic conductances, as a reduced compartment model (few compartments) or as a detailed compartmental model. The more abstract models are theoretically and computationally tractable and have fewer degrees of freedom, while the more detailed models are closer to biological realism and can be linked to empirical data”
– In the best case, this might put mind uploading well within the capabilities of our rapidly progressing computer technology. So we really don’t know if we will ever understand our brain and consciousness well enough to upload human minds.
#The Scientific Case for Brain Simulations, Gaute T. Einevoll et Al., 2019
#Whole Brain Emulation: A Roadmap, Anders Sandberg and Nick Bostrom, 2008
Quote: “We have presented arguments for why brain network simulators are not only useful but also likely critical for advancing systems neuroscience.”
Quote: “Brain network simulation is still in its infancy, and the simulators and associated infrastructure should be developed to allow for the study of larger networks and fully exploit the capabilities of modern computer hardware.”
– Successful mind uploading is functional immortality. Unless damaged or deleted, you will continue to exist as long as a copy is stored somewhere.
#Indefinite survival through backup copies, Anders Sandberg and Stuart Armstrong, 2012
https://www.fhi.ox.ac.uk/reports/2012-1.pdf
Quote: “However, there might be a way around it if we can make backup copies. If an
entity gets destroyed by something, restore them from the backup. Of course,
backups are also vulnerable to random destruction. But if we distribute the
copies widely so that their fate is independent of each other and the original (in
practice a very hard problem, see below) and whenever one gets destroyed it is
replaced by a copy of a surviving copy, then the probability of all N copies being destroyed is µ^N , where µ > 0 is the probability (per unit of time) of one copy being destroyed. Typically µ^N << µ even for modest N.”
Note: Staying immortal means having as many copies as possible, and adding more as you go. Each extra copy is one more chance to survive a catastrophe.
– How would mind uploading change your outlook on life? Will you feel safer knowing that death is not necessarily the end? Or would you try to be super safe to avoid dying before your mind is uploaded?
#Embodiment in Whole-Brain Emulation and its Implications for Death Anxiety, Charl Linssen and Pieter Lemmens , 2016
https://jetpress.org/v26.2/linssen_lemmens.pdf
Quote: “In biology, these fears and strategies are continuously reinforced by such daily reminders as the deaths of others and maladies of the body. In WBE, if the substrate is built to extreme standards of reliability, the frequency of such reminders could be drastically reduced.”