Connectome

by Sebastian Seung

  [>] A drug is an artificial molecule: The line between artificial and natural is blurred by “biologics.” Vaccines are the classic example, but newer ones are proteins identical or similar to the ones that occur naturally in the body. These can still be viewed as artificial, in the sense that they are synthesized or introduced by non-natural means. Biologics are distinguished from “small molecules,” which contain many fewer atoms and are the classic kind of drug.

  [>] the attrition rate: Kola and Landis 2004.

  [>] between the first and last stages: Markou et al. 2008.

  [>] humanized mouse models: Legrand et al. 2009.

  [>] animal behavior that is analogous: Nestler and Hyman 2010.

  14. To Freeze or to Pickle?

  [>] probability theory: The founding of probability theory is recorded in a series of letters between Pascal and another famous mathematician, Pierre de Fermat. See Devlin 2010.

  [>] a searing religious vision: The two-hour vision took place the evening of November 23, 1654, which has come to be known as Pascal’s “night of fire.” We know of it only because Pascal recorded the event on a document that was sewn into his coat, and discovered by a housekeeper after his death. See O’Connell 1997.

  [>] one thousand living members: According to an Alcor web page, as of July 31, 2011, there are 955 members and 106 cryopreserved “patients.”

  [>] want to live forever: Some friends tell me that they wouldn’t want to be immortal. This position has also been argued by philosophers, notably Charles Hartshorne. I find this ironic, as I saw Hartshorne a few times in my father’s office at the University of Texas, and he seemed practically immortal—he rode a bicycle well into his eighties and lived until age 103. But I agree with Camus that suicide is more interesting as a philosophical problem, since immortality doesn’t seem like a realistic option anyway.

  [>] apocryphal, alas: Peck 1998.

  [>] the court sorcerer Xu Fu: Howland 1996.

  [>] laid the athlete’s remains to rest: The Ted Williams story is told in Johnson and Baldyga 2009. The Ted Williams story is told in Johnson and Baldyga 2009.

  [>] “Miracle of the Sun”: There are many books on the miracle. Bertone and De Carli 2008 was written by a cardinal and endorsed by the pope. The apparition of the Virgin Mary revealed three secrets to the shepherd children. The Vatican claims to have disclosed all of them to the world, but has been accused of holding back part of the third, “Last Secret of Fatima.”

  [>] believe in miracles: Pew Forum on Religion 2010.

  [>] 270,000 customers: Markoff 2007.

  [>] According to Clarke’s: Clarke 1973 lays out three laws. The first and second are: (1) When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is very probably wrong. (2) The only way of discovering the limits of the possible is to venture a little way past them into the impossible.

  [>] Sonny Graham received a heart: Dudley 2008.

  [>] Cheryl had been married five times: Wigmore 2008.

  [>] Sperm survive the best: Woods et al. 2004.Woods et al. 2004.

  [>] ice inside cells is lethal: Mazur, Rall, and Rigopoulos 1981. Mazur, Rall, and Rigopoulos 1981.

  [>] still damaging to cells: I use the term salty here for convenience, but in reality other solutes beside salt ions are also important.

  [>] oocytes and embryos: Woods et al. 2004.

  [>] vitrified kidney: Fahy et al. 2009.

  [>] Peter Mazur: Mazur 1988.

  [>] “respirator brain”: Towbin 1973.

  [>] determination of death: Laureys 2005; President’s Council on Bioethics 2008.

  [>] “If the brainstem is dead”: Laureys 2005.

  [>] vicious cycle continues: President’s Council on Bioethics 2008.

  [>] discarded information: Conversely, some of the information in the connectome might be irrelevant to personal identity, because it’s just random “noise” created as the brain wired itself up during development.

  [>] After mechanical ventilation: Agarwal, Singh, and Gupta 2006.

  [>] types of damage present: Rees 1976; Kalimo et al. 1977.

  [>] still intact in the EM images: However, many are depleted of their vesicles containing neurotransmitter. Recall that the strength of a synapse is related to its size, and one measure of size is the number of vesicles. Therefore, information about synaptic strength—information that can be regarded as part of the connectome—may be difficult to recover.

  [>] Eric Drexler: Drexler 1986.

  [>] Charles Olson: Olson 1988.

  [>] fixing them in place: Formaldehyde and glutaraldehyde are used to link protein molecules together. An even more toxic fixative, osmium tetroxide, has the dual function of binding together fat molecules and staining the membranes to which they belong.

  [>] Figure 53, left: The tissue is embedded in Epon, an epoxy resin, and appears black because of the osmium staining.

  [>] Lenin was embalmed: Modern embalming methods began to develop in the seventeenth and eighteenth centuries. Most notoriously, the eccentric London dentist Martin van Butchell embalmed his dead wife in 1775 and displayed her in the window of his home office. See Dobson 1953.

  15. Save As . . .

  [>] “mind uploading”: In his 1955 story “The Tunnel under the World,” Frederik Pohl wrote: “Each machine was controlled by a sort of computer which reproduced, in its electronic snarl, the actual memory and mind of a human being. . . . It was only a matter . . . of transferring a man’s habit patterns from brain cells to vacuum-tube cells” (Pohl 1956). The first mention in the scientific literature may have been in Martin 1971: “We shall assume that developments in neurobiology, bioengineering and related disciplines . . . will ultimately provide suitable techniques of ‘read-out’ of the stored information from cryobiologically preserved brains into nth generation computers capable of vastly outdoing the dynamic patterning of operation of our cerebral neurones.”

  [>] requires dying first: After being resurrected, Jesus is said to have ascended to heaven without dying again. Shoemaker 2002 describes how Christians have argued for millennia over whether the Virgin Mary also entered heaven without dying first. Being carried up to heaven by God is called “Assumption,” to distinguish it from the “Ascension” of Jesus, which happened by his own power. In 1950, Pope Pius XII promulgated Munificentissimus Deus, which decreed that Mary, “having completed the course of her earthly life, was assumed body and soul into heavenly glory.” This dogma recognized the importance of the Assumption but didn’t really settle the debates, because its wording was ambiguous. Christians have also long argued over whether the Old Testament figures of Elijah and Enoch were assumed into heaven without dying first.

  [>] “brain in a vat”: The story “Where Am I?” in Dennett 1978 is a wonderful example. For an actual attempt to keep an isolated guinea pig brain alive and functioning, see Llinas, Yarom, and Sugimori 1981.

  [>] pyramidal tract contains: Lassek and Rasmussen 1940. For another way of tallying the numbers, let’s categorize the neurons of a nervous system by their connection to the outside world. Sensory neurons convert external stimuli into neural signals. For example, the photoreceptors of the retina produce electrical signals when stimulated by light. Motor neurons make synapses onto muscles and convert neural signals into movements. The remainder are called interneurons, because they are interposed between sensory and motor neurons. In the C. elegans nervous system, sensory neurons, motor neurons, and interneurons are found in comparable numbers. But sensory and motor neurons make up a vanishingly small fraction of our nervous system. Saying that a neuron is an interneuron is no great distinction, because almost all are. Very few of the neurons in our brains “talk” with the outside world. They mostly talk with each other.

  [>] running on a gigantic computer: Bostrom 2003; Lloyd 2006.

  [>] Alan Turing: Turing 1950.

  [>] successful example of AI: There are some s
light differences in Turing’s original setup of the test. The interested reader should consult Turing’s paper, which is very readable.

  [>] a proper Turing test: Natalie Zemon Davis has argued that Guerre’s wife knew very well that the new Guerre was fake, but fell in love and conspired with him (Davis 1983, 1988). But no historians question that some of Guerre’s sisters and friends were genuinely fooled.

  [>] The more accurate the simulation: Then again, self-models are often not very accurate. Researchers have shown that most people have inflated opinions of their own abilities. This is called the Lake Wobegon Effect, after the humorist Garrison Keillor’s fictional town in which “all the women are strong, all the men are good-looking, and all the children are above average.”

  [>] Markram was one of the first: He also showed that the strength of a cortical synapse can fluctuate from spike to spike. In collaboration with theoretical colleagues, he introduced mathematical models describing this phenomenon, known as short-term synaptic plasticity.

  [>] simulation of a cat brain: Ananthanarayanan et al. 2009. Ananthanarayanan et al. 2009.

  [>] “Cat Fight Brews Over Cat Brain”: Adee 2009 also prints the full text of the letter.

  [>] neurons of the same type: For example, when neuroscientists inject electrical current into an inhibitory neuron of the neocortex, it can generate spikes for a long time without faltering (Connors and Gutnick 1990). But when they stimulate a pyramidal neuron, it slows down after the first few spikes, as if it were becoming “fatigued.”

  [>] Once all neuron types: It will also be necessary to classify synapses into types. Here I’ve taken the view that neuron types already include all information about synapse types. According to Dale’s Principle, a neuron secretes the same neurotransmitter (or set of neurotransmitters) at all of the synapses it makes onto other neurons. That’s why all the outgoing synapses of a pyramidal neuron secrete glutamate. There are many variants of glutamate receptor molecules. The particular variant that occurs at a synapse may be a property of the neuron type of the receiving neuron. In other words, the type of a synapse may be determined by the types of the neurons that it connects. If this turns out not to be true, then connectomes will have to include separate information about synapse types as well as neuron types.

  [>] millions of ion channels: This numerical estimate is courtesy of Michael Hausser and Arnd Roth. The multicompartmental models are based on the aggregate behaviors of large populations of channels. This has some similarity to the way in which pollsters keep track of the percentage of voters who support a candidate. Each compartment represents some part of the neuronal membrane. It contains multiple populations of ion channels, one population for each channel type. Therefore, if a neuron is divided into one hundred compartments, and there are ten types of ion channel, then the model contains a thousand variables for specifying the states of the ion channels. That may sound like a lot of variables, but it’s still much less than the total number of ion channels in the neuron.

  [>] multicompartmental model neurons: Multicompartmental models are essential when different parts of a neuron function independently. The dendrites of a single starburst amacrine cell of the retina, for example, detect multiple directions of visual motion and send different signals to other neurons (Euler, Detwiler, and Denk 2002).

  [>] Peters’ Rule: This was first stated in its general form by Braitenberg and Schüz 1998, and named in honor of Alan Peters for formulating a specific case of the rule.

  [>] more difficult for C. elegans: Lockery and Goodman 2009.

  [>] The only information unique: More realistically, the properties of each neuron type might vary slightly across normal people. These variations might be predictable from their genomes. If so, we’d have to say “You are your connectome plus models of neuron types plus your genome.” But again, a genome contains much less information than a connectome, so “You are your connectome” would still be a good approximation.

  [>] about one hundred types: White et al. 1986.White et al. 1986.

  [>] diffusion of neurotransmitter: Electronic circuits sometimes behave differently from their simulations, in which components can interact only if they are connected by wires. A real circuit can contain interactions mediated by “thin air” rather than wires. For example, one wire can set up an electric field that is felt by a nearby wire, a phenomenon known as “stray capacitance” that is analogous to extrasynaptic interactions in the brain. This type of deviation from the model can be extremely difficult to identify and troubleshoot.

  [>] almost beyond imagining: If you’re up to the mind-bending task of thinking about such a simulation, you can consult Tipler 1994, which proves that it should be possible in this universe.

  [>] all the positions and velocities: I’m avoiding the issue of whether quantum physics is important for the functioning of the brain. Tegmark 2000 provides some insight into the subject.

  [>] Ralph Merkle: Merkle 1992. Some of the earliest writings about connectomics were penned by proponents of cryonics and uploading, although the term connectome was not coined until later. In his 1989 technical report, “The Large Scale Analysis of Neural Structures,” Ralph Merkle reviewed the state of the art in serial electron microscopy. He knew that the C. elegans connectome had been mapped, and speculated about scaling up to the human brain.

  References

  Abeles, M. 1982. Local cortical circuits: An electrophysiological study. Berlin: Springer.

  Aboitiz, F., A. B. Scheibel, R. S. Fisher, and E. Zaidel. 1992. Fiber composition of the human corpus callosum. Brain Research, 598 (1–2): 143–153.

  Abraham, Carolyn. 2002. Possessing genius: The bizarre odyssey of Einstein’s brain. New York: St. Martin’s Press.

  Adee, S. 2009. Cat fight brews over cat brain. IEEE Spectrum Tech Talk Blog. Nov. 23.

  Agarwal, R., N. Singh, and D. Gupta. 2006. Is the patient brain-dead? Emergency Medicine Journal, 23 (1): e05.

  Albertson, D. G., and J. N. Thomson. 1976. The pharynx of Caenorhabditis elegans. Philosophical Transactions of the Royal Society of London, Series B, Biological Sciences, 275 (938): 299–325.

  Amari, S. I. 1972. Learning patterns and pattern sequences by self-organizing nets of threshold elements. IEEE Transactions on Computers, 100 (21): 1197–1206.

  Amit, D. J. 1989. Modeling brain function. Cambridge, Eng.: Cambridge University Press.

  Amit, D. J., H. Gutfreund, and H. Sompolinsky. 1985. Spin-glass models of neural networks. Physical Review A, 32 (2): 1007.

  Amunts, K., G. Schlaug, L. Jäncke, H. Steinmetz, A. Schleicher, A. Dabringhaus, and K. Zilles. 1997. Motor cortex and hand motor skills: Structural compliance in the human brain. Human Brain Mapping, 5 (3): 206–215.

  Ananthanarayanan, R., S. K. Esser, H. D. Simon, and D. S. Modha. 2009. The cat is out of the bag: Cortical simulations with 109 neurons, 1013 synapses. In Proceedings of the Conference on High Performance Computing Networking, Storage, and Analysis, p. 63. ACM.

  Andersen, B. B., L. Korbo, and B. Pakkenberg. 1992. A quantitative study of the human cerebellum with unbiased stereological techniques. Journal of Comparative Neurology, 326 (4): 549.

  Antonini, A., and M. P. Stryker. 1993. Development of individual geniculocortical arbors in cat striate cortex and effects of binocular impulse blockade. Journal of Neuroscience, 13 (8): 3549.

  ——— . 1996. Plasticity of geniculocortical afferents following brief or prolonged monocular occlusion in the cat. Journal of Comparative Neurology, 369 (1): 64–82.

  Azevedo, F. A., L. R. Carvalho, L. T. Grinberg, J. M. Farfel, R. E. Ferretti, R. E. Leite, F. W. Jacob, R. Lent, and S. Herculano-Houzel. 2009. Equal numbers of neuronal and nonneuronal cells make the human brain an isometrically scaled-up primate brain. Journal of Comparative Neurology, 513 (5): 532–541.

  Bagwell, C. E. 2005. “Respectful image”: Revenge of the barber surgeon. Annals of Surgery, 241 (6): 872.

  Bailey, A., A. Le Couteur, I. Gottesman, P. Bolton, E. Simonoff, E. Yuzda, and
M. Rutter. 1995. Autism as a strongly genetic disorder: Evidence from a British twin study. Psychological Medicine, 25 (1): 63–77.

  Bailey, P., and G. von Bonin. 1951. The isocortex of man. Urbana, Ill.: University of Illinois Press.

  Bamman, M. M., B. R. Newcomer, D. E. Larson-Meyer, R. L. Weinsier, and G. R. Hunter. 2000. Evaluation of the strength-size relationship in vivo using various muscle size indices. Medicine and Science in Sports and Exercise, 32 (7): 1307.

  Barlow, H. B. 1972. Single units and sensation: A neuron doctrine for perceptual psychology. Perception, 1 (4): 371–394.

  Basser, L. S. 1962, Hemiplegia of early onset and the faculty of speech with special reference to the effects of hemispherectomy. Brain, 85: 427–460.

  Baum, K. M., and E. F. Walker. 1995. Childhood behavioral precursors of adult symptom dimensions in schizophrenia. Schizophrenia Research, 16 (2): 111–120.

  Bear, M. F., B. W. Connors, and M. Paradiso. 2007. Neuroscience: Exploring the brain, 3rd ed. Baltimore: Lippincott, Williams, and Wilkins.

  Beard, M. 2008. The fires of Vesuvius: Pompeii lost and found. Cambridge, Mass.: Harvard University Press.

  Bechtel, W. 2006. Discovering cell mechanisms: The creation of modern cell biology. Cambridge, Eng.: Cambridge University Press.

  Benes, F. M., M. Turtle, Y. Khan, and P. Farol. 1994. Myelination of a key relay zone in the hippocampal formation occurs in the human brain during childhood, adolescence, and adulthood. Archives of General Psychiatry, 51 (6): 477–484.