Wetware - A Computer in Every Living Cell

Dennis Bray, 2009. BVTN 571.6 BRA

• p64 My argument is as follows: Living cells must be capable of some sort of logical analysis or they would never survive. How else could free-living microorganisms avoid obstacles, hunt for food, respond to stimuli, store memories? ... What is more, as just mentioned, they (proteins) have the ability to exist in two forms: two shapes that can have totally different properties. Flipping between these two states, in a manner sensitive to their surroundings, they perform like molecular switches. ... er, uh ...

  • p82 Living organisms, by contrast, seem incorrigibly analog. In other words, they do not change by stepwise increments, in ones and zeros, but as a continuous variable over a wide range. ... a little better ...

  • IMHO, the "logical analysis" analogy is inaccurate; cells do accumulate state information, but they do not arrive at it by logic, neither the human philosophical kind nor the digital computer kind. Cells can be decisive without being logical.
• p75 "For example, the addition of phosphate groups to a protein in nerve cells encodes long-lived memories in the brain.
  • Baloney. Neurons sculpt synapses, at a much larger scale than single proteins.
• p75 Kinases and phosphatases add and subttract phosphate groups to proteins ... and "underpins" computation ... a tunable device.
• p102 Computer animations of bacterial chemotaxis can mislead. Cells contain millions of molecules, each with tens of thousands of atoms, behavior dependent on chemical environment and recent history.
• p163 Histones; positively charged proteins associate with negatively charged DNA to shape chromosomes.
• p174 tissues like epithelia connect cells with junctions permitting passage of some molecules
• p211 typically 1000 synapses per neuron (some have many more), 100 billion neurons, 1e14 synapses

  • von Neumann 1958 The Computer and the Brain, 100 million bit (operations) per synapse

  • Landauer at Bell, 4 billion bits of accessable factoids
  • My opinion - specific connectivity, one particular neuron to another particular neuron, adds ln₂(C) bits of stored information, where C is the number of spatially possible connections that neuron can make. WAG, perhaps 1,000 possibilities weighted by likelihood, so 10 bits per synapse. And a brain over a lifetime can have many states, so a lifetime (rather than an instant) is a much larger record than 10 kbits per neuron. The neural hardware gets reused for new tasks, modifying old disused capabilities.
• p219 CamII kinase adds phosphate groups to synaptic proteins, changing the shape of dendritic spines
  • p220 individual proteins turn over in weeks, CamII modifies replacements

• p228 nuclear import signals receptors in nuclear membrane recognize and permit entry

  • motor proteins controlled by adapter proteins that read molecular addresses on RNA