Connectome

by Sebastian Seung

  [>] London taxi drivers: Maguire et al. 2000.

  [>] In musicians: Hutchinson et al. 2003; Gaser and Schlaug 2003. My statement “thicker cortex” is a bit glib, because the measurements rely on a method called voxel-based morphometry, which can’t distinguish between thickening and other structural changes. Thickening is just one possible interpretation.

  [>] Bilinguals: Mechelli et al. 2004.Mechelli et al. 2004.

  [>] severe mental disorder: Kessler et al. 2005.

  [>] symptoms of autism: Frith 2008.

  [>] unable to function: There are also milder forms of autism, which involve some but not all of the symptoms. For example, Asperger’s syndrome is defined by social impairment and repetitive behaviors but not linguistic difficulties. The term autism spectrum disorders has been introduced to include the entire range, from mild to severe forms of autism. Fombonne 2009 estimated the incidence of full-blown autism as two per 1,000 people, and that of the autism spectrum disorders as several times higher.

  [>] “beautiful child”: Frith 1993.

  [>] defined the syndrome: The Viennese pediatrician Hans Asperger is also credited with having defined autism a few years earlier.

  [>] large heads: Kanner 1943.

  [>] heads and brains: Redcay and Courchesne 2005. Interestingly, autism provides counterevidence to the maxim that bigger is better. Phrenologists might respond by pointing to autistic “savants,” who exhibit impressive displays of memory, numerical calculation, or other mental abilities like the fictional character in the movie Rain Man (Treffert 2009). Perhaps these enhanced mental abilities could be explained by the enlargement of autistic brains. But most autistic children are not savants, and even savants have disabilities. Perhaps it’s fairer to conclude that the phrenological approach of studying brain size is an oversimplification.

  [>] frontal lobe: Carper et al. 2002.

  [>] first-person account: BGW 2002.

  [>] less effective for the negative symptoms: Second-generation, or “atypical,” antipsychotic drugs were marketed as superior for negative symptoms, but this claim is now being questioned. For more on this controversy, see Murphy et al. 2006 and Leucht et al. 2009. The atypicals are less likely to produce movement disorders as side effects, which were common with first-generation, or “typical,” antipsychotics.

  [>] overall brain volume: Steen et al. 2006; Vita et al. 2006. The difference exists even in patients receiving their first psychiatric treatment, so it does not appear to be a long-term effect of antipsychotic medications.

  [>] lateral and third ventricles: Steen et al. 2006.

  [>] “graveyard”: Plum 1972.Plum 1972.

  2. Border Disputes

  [>] infant brain grows rapidly: Voigt and Pakkenberg 1983.

  [>] philosophy of education: Spurzheim was actually quite sophisticated for his time, acknowledging that other changes in the brain might take place besides growth: “The growth of the organs [brain regions], however, is not the only or even most important advantage to be derived from proper exercise. . . . [T]he size of the organ . . . will not augment in proportion to its being exercised, but its fibres will act with more facility” (Spurzheim 1833, pp. 131–132).

  [>] through simple mazes: The Hebb–Williams test of animal intelligence was a battery of twenty-four problems, each involving finding food in a simple maze. Donald Hebb pioneered this type of research on the effects of environmental enrichment. It’s briefly noted in Hebb 1949, which is better known for its presentation of Hebb’s theories of the cell assembly and synaptic plasticity (see Chapters 4 and 5 below).

  [>] Mark Rosenzweig: Rosenzweig 1996. The test of statistical significance was based on comparisons of siblings born in the same litter. The change in cortical size was not due to an overall change in brain size. In fact, the noncortical areas of the brain were slightly smaller. The change was not due to an increase in body size either. The environmentally enriched rats were actually somewhat lighter, owing to their increased activity.

  [>] learning to juggle balls: Draganski et al. 2004; Boyke et al. 2008.

  [>] intensive study for exams: Draganski et al. 2006. Draganski et al. 2006.

  [>] Korbinian Brodmann: His map spanned the neocortex, which is the predominant part of the cerebral cortex. Confusingly, the term cortex often serves as an abbreviation for the neocortex alone. Brodmann divided the cortex into forty-three areas (Brodmann 1909), but not all are visible in Figure 11, which includes only one view of the cerebrum. If you look closely, you’ll notice that 52 is the map’s largest number, not 43. That’s because Brodmann skipped 12–16 and 48–51. He reserved these numbers for cortical areas in animals that appeared to have no analogues in the human cortex. Brodmann used a microscope to delineate the areas, as I’ll describe in Chapter 10. However, the areas line up roughly with the cortical folds, so they can be located approximately even without a microscope.

  [>] after three months: Cramer 2008.

  [>] after stroke: Cramer 2008.

  [>] removing one hemisphere: Mathern 2010. The procedure is justified, for example, when MRI reveals a one-sided brain abnormality that is clearly the cause of the seizures.

  [>] walk and even run: Vining et al. 1997. For inspiring testimonials by patients, see http://hemifoundation.intuitwebsites.com.

  [>] migrate to the right hemisphere: Basser 1962 discusses very early childhood; Boatman et al. 1999, late childhood. The phenomenon was already noted by Broca in the nineteenth century.

  [>] Miguel Nicolelis: Nicolelis 2007.

  [>] crude act of butchery: Bagwell 2005. By medieval times the church had taken over the practice of medicine. A 1215 papal edict forbade the clergy from practicing surgery, because contact with blood or bodily fluids was considered contaminating. Surgery was left to barbers, who may have been more effective healers than the university-trained physicians.

  [>] tie off large arteries: Finger and Hustwit 2003.

  [>] unremarked for so long: The history of phantom limbs from Paré to Mitchell is surveyed in Finger and Hustwit 2003.

  [>] phantom is not real: Reilly and Sirigu 2008.

  [>] irritated nerve endings: This explanation is credited to Descartes by Finger and Hustwit 2003.

  [>] this didn’t help: Ramachandran and Blakeslee 1999.

  Wilder Penfield: Penfield and Boldrey 1937.

  [>] V. S. Ramachandran: Ramachandran, Stewart, and Rogers-Ramachandran 1992. An entertaining and readable account of this research is provided in Ramachandran and Blakeslee 1999. Ramachandran’s discoveries in humans were probably not surprising to Mike Merzenich and other neuroscientists who had already found similar phenomena in animals, as reviewed in Buonomano and Merzenich 1998.

  [>] sensation of a phantom limb: This description may sound incomplete, because I’ve spoken only of functions and avoided inputs and pathways, which are discussed later in this book. It’s more revealing to say that amputation deprives the lower-arm territory of inputs from sensory pathways. Remapping replaces these with sensory inputs from the face and upper arm.This description may sound incomplete, because I’ve spoken only of functions and avoided inputs and pathways, which are discussed later in this book. It’s more revealing to say that amputation deprives the lower-arm territory of inputs from sensory pathways. Remapping replaces these with sensory inputs from the face and upper arm.

  [>] stroked the face of an amputee: There was even a one-to-one mapping between facial locations and digits of the phantom hand (cheek to thumb, chin to pinkie, and so on).

  [>] Functional MRI: More precisely, fMRI measures the BOLD (blood oxygen level dependent) signal, which was discovered by the Japanese scientist Seiji Ogawa. This is defined as the ratio between the oxygenated and deoxygenated forms of hemoglobin, the molecule in the blood that ferries oxygen from the lungs to the rest of the body. Using a brain region has two opposing effects on the BOLD signal. First, the region burns more energy, which deoxygenates hemoglobin. Second, blood flow increases, which carries in more oxygenated
hemoglobin. (Many believe that blood flow increases in response to use, because the brain precisely regulates blood flow to fulfill the energy needs of each region.) Since either of these effects can dominate, using a brain region can either increase or decrease the BOLD signal, which confuses the interpretation of fMRI. On a related note, since the BOLD signal reflects energy consumption, some people quip that using fMRI to understand the brain is like trying to understand the engine of a car by measuring where it gets the hottest.

  [>] “spots on brains”: These images give the misleading impression that a person uses a small fraction of the brain for any given task. However, each image is actually obtained by subtracting two images corresponding to two similar mental tasks. A “lit-up” region was used more in one task than the other. One should not conclude that all the other regions lay idle. Many of them were active, but the level of activity was similar in both tasks.

  [>] the shift occurred: Lotze et al. 2001 also demonstrated a similar remapping of area 4 in amputees, and measured brain activity caused by imagined movements of the phantom hand. Researchers also used fMRI to demonstrate remapping of area 4 in stroke patients. The hand representation moved up or down within area 4, depending on the location of brain damage. Further studies found that stroke can cause remapping on a larger scale, affecting distant areas on the same or the other side of the brain (Cramer 2008).

  [>] left-hand representation: Elbert et al. 1995 used magnetic source imaging rather than fMRI. They found a shift in the average location of the left-hand representation within area 3, which they interpreted as a change in area. But a direct measurement of the size of the representation showed no statistically significant change. They couldn’t prove that the shift was caused by musical training, because of the possibility of selection bias. However, the size of the shift was correlated with the age at which musical training began. See Amunts et al. 1997 for a related study using MRI.

  [>] crippling disorders: Elbert and Rockstroh 2004.

  [>] focal dystonias: A famous example is the pianist Leon Fleisher, who lost the use of his right hand for thirty-five years but recently made a comeback with both hands after receiving treatment based on injections of Botox into his arm muscles.

  [>] violin and Braille: Sterr et al. 1998 not only showed an expanded hand representation but argued that the arrangement of the fingers in the representation was disorganized, which might distinguish Braille reading from violin playing.

  [>] frontal lobe in schizophrenics: Glahn et al. 2005.

  [>] about brain disorders: Kaiser et al. 2010 and Bosl et al. 2011 are two recent studies characterizing activity in the autistic brain.

  [>] strength with a machine: Actually the scientific studies use isometric measurements, meaning that the force is measured while the joint angle is held fixed. This is more controlled, because force depends on joint angle. Muscle size is quantified by cross-sectional area (CSA), which is expected to be roughly proportional to the number of fibers and hence to strength.

  [>] correlation coefficients: You might think it’s silly to research this correlation, since common sense tells us it must be strong. Actually this has been surprisingly difficult to establish empirically. Maughan, Watson, and Weir 1983 reported lower correlation coefficients and took the contrarian view that “strength is not a useful predictive index of muscle cross-sectional area.” More recent studies like Bamman et al. 2000 and Fukunaga et al. 2001 appear to agree on stronger correlations, possibly thanks to improvements in measurement methods. Still, many interesting questions remain unanswered. For example, is the relationship between size and strength different for powerlifters and bodybuilders, or for elite athletes and regular people?

  [>] boundaries of Brodmann’s map: Lashley and Clark 1946.

  [>] cortical equipotentiality: Lashley 1929.

  [>] over 100 million: The estimate that Brodmann area 17 contains over 100 million neurons is from Huttenlocher 1990.

  3. No Neuron Is an Island

  [>] Figure 13: Although in the brain no neuron is an island, isolated neurons can be artificially cultured in a plastic dish, as shown in Figure 13. Even this neuron is not truly island-like, though, as its branches actually extend far outside the borders of the image, to form connections with other neurons in the dish. The image was obtained by scanning electron microscopy.

  [>] one million: If we don’t restrict ourselves to the brain, neurites can be longer still. Some neurites travel from the brain to the spinal cord, and others connect the spinal cord to the toes and fingers. And let’s not forget that giraffes and whales have neurites too.

  [>] marked “ax” and “sp”: “ax” marks an axon, and “sp” a dendritic spine, which sticks out of the dendrite like a thorn.

  [>] do not really touch: Invisible in the image in Figure 14 are various molecules that span the cleft between the membranes of the two neurons and bring them into direct contact. But the whole notion of “touching” starts to break down at even higher magnification. What we call matter consists mainly of empty space between its constituent particles.

  [>] same small set of neurotransmitters: Eccles et al. 1954 stated the principle that a neuron secretes a single neurotransmitter, and attributed it to Sir Henry Dale, who won a 1936 Nobel Prize for his studies of synaptic transmission. Eccles 1976 later revised Dale’s Principle to allow for multiple neurotransmitters. Eccles himself shared a 1963 Nobel Prize for his work on synapses. More recently, researchers have found a further exception: neurons are capable of switching from one neurotransmitter to another.

  [>] brain secretes thoughts: The eighteenth-century French philosopher and physiologist Pierre Cabanis wrote that “the brain secretes thought as the liver secretes bile.”

  [>] send them to specific targets: In most biological contexts, chemical signaling relies upon the specificity of molecular binding (the lock-and-key mechanism). That’s not sufficient to prevent crosstalk between synapses, because many synapses use exactly the same neurotransmitter.

  [>] minimize crosstalk: That’s not to say there is zero crosstalk. Some spillover of neurotransmitter is known to occur, and appears important for brain function in certain cases.

  [>] “most expensive loveseat”: Russell 1978.

  [>] 67 miles of tangled wire: Kolodzey 1981.

  [>] insulating material: Small amounts of crosstalk can still occur because of electrical fields, which penetrate the insulation.

  [>] millions of miles: The brain is over a million cubic millimeters in volume, and a large fraction of that is cortex. According to Braitenberg and Schüz 1998, a cubic millimeter of cortex contains several miles of neurites.

  [>] single axon, long and thin: This description holds for a very common type of neuron, the pyramidal neuron of the cortex. However, there are many other types of neurons, which have different appearances. The dendrite–axon distinction is not even valid for some types of neuron, especially in invertebrate nervous systems. For these neuron types, each neurite both sends and receives synapses. This description holds for a very common type of neuron, the pyramidal neuron of the cortex. However, there are many other types of neurons, which have different appearances. The dendrite–axon distinction is not even valid for some types of neuron, especially in invertebrate nervous systems. For these neuron types, each neurite both sends and receives synapses.

  [>] typical synapse is from: But there are also synapses from axon to cell body, dendrite to dendrite, axon to axon, and pretty much any other variation you can think of.

  [>] Figure 17: This figure shows a brief segment of the voltage signal recorded from a neuron in the hippocampus of a rat exploring a maze. The experiment is described in Epsztein, Brecht, and Lee 2011.

  [>] above the static: After the telegraph, the telephone was invented for analog communication—that is, the transmission of voice signals without encoding them into pulses. But now the telephone system has become digital again, utilizing something like Morse code. The encoding and decoding are invisible to the u
ser because they are done quickly and automatically by electronic circuits rather than human operators. Why would our sophisticated telephone systems return to the style of communication used in the primitive telegraph? One reason is that today’s systems are designed to transmit information at the highest possible rate. This requires operating at the limits set by noise, so the best strategy is again digital.

  [>] spike triggers secretion: I say “passing” because synapses mostly occur at locations along the axon, so that spikes propagate past them. Some synapses are located at axonal dead ends, however, so that spikes terminate at them.

  [>] a synapse converts: How receptors transform chemical signals into electrical ones will be explained in Chapter 6. How receptors transform chemical signals into electrical ones will be explained in Chapter 6.

  [>] toward the cell body: This is known as the Law of Dynamic Polarization. Neuroscientists sometimes violate the law by using electrical stimulation to initiate a spike that travels backward along the axon toward the cell body. Such “antidromic” propagation is opposite the normal direction, proving that signal transmission along the axon is two-way.

  [>] cells that support them: The nervous system also contains non-neuronal cells, known as glia. These come in a number of types, and are absolutely essential for keeping the brain alive and functioning. I will take the traditional view that glial cells are like the crew, supporting the cast of neurons that star in the mental show. Neurons and glial cells are about equally numerous (Azevedo et al. 2009). Much more about glia can be found in Fields 2009.

  [>] synapses onto muscle fibers: These are called neuromuscular junctions, to contrast them with ordinary synapses between neurons.

  [>] “To move things”: Sherrington 1924. Sherrington 1924.

  [>] 190 stations: Bradley 1920.

  [>] synapses are weak: Some contrarians believe that there are a small number of strong synapses, and these are the important ones for brain function.