What Bodies Think About: Bioelectric Computation Outside the Nervous System (2018)

Prof. Michael Levin @ NeurIPS 2018

Link to talk (youtube)

My first introduction to bioelectricity and the work of Prof. Michael Levin.

Mind-blown by the stuff in this talk and everything I’ve seen about it since.

In summary:

  1. Unicellular organisms without brains are capable of pattern recognition and behavioural goals, as these combine to create multicellular organisms, no capabilities are lost, only the boundary of the self expands so that they behave as one “self”.

  2. Instead of communicating with each other as separate entities, gap junctions facilitate the passing of ions (electrical charges) across cells directly, such that the cells cannot distinguish its own “self” from that of the other cell, and as such expands the boundary of the “self” to cover multiple cells rather than one.

  3. Organisms, including humans, have all the tools and knowledge to (re)generate any part of their anatomy without the need for micromanagement, similar to a coroutine call in software; trigger the instructions to build an eye somewhere in the organism and the organism will take care of literally everything else.

  4. Regeneration is not an embryonic phenomenon; organisms are capable of regenerating any part of their anatomy at any point given the instructions to do so.

  5. Organisms have an electrical pattern across them which dictates the anatomy it creates.

  6. By manipulating this bioelectrical pattern you can change what anatomy will be built in the event of regeneration — the pattern is like a blueprint for the goal anatomy, not related to the current anatomy.

  7. If you manipulate the biolelectrical pattern at the tail end of a regenerative organism to match that of the head end, when you chop off the tail it will happily build a second head instead of a tail, complete with a working brain and persisted memories.

  8. If you teach a Planarian (a regenerative flatworm) something, then chop off its head, it will regrow a new head and a brain that remembers what you taught it. That data is persisted somewhere (???). Similarly, a butterfly remembers things that the caterpillar knew despite its entire nervous system being liquified during transformation.

  9. You are not restricted to the expected anatomy for that species. Planarians can be made to regenerate the heads of different flatworm species separated by 150 million years of evolution, and even head shapes that don’t exist in nature.

  10. This manipulation is permanent and the new anatomy will be regenerated in perpetuity no matter how many times you amputate the anatomy in question.

  11. Their research is the first steps in developing regenerative medicine — being able to induce tissues to regrow complex organs without genomic editing or the need to micromanage any of the process, just by delivering the right message to the organism (“build a new kidney here”).

  12. Bioelectric pattens can override genetic defects. By setting the correct bioelectric pattern with drugs that selectively open and close ion channels, a normal tadpole brain can develop despite the presence of a genetic defect that would usually result in a deformed brain.

  13. The end-game for their field is to be able to design organisms the way we design currently 3D models and to “export” the required bioelectric patterns that result in the designed organism growing.

  14. Learning and memory is not confined to the brain. Non-neural cells in the bone, heart, pancreas, etc, are all capable of learning and developing robust novel responses to stimuli. This might help the machine learning field to overcome the limits which arise from basing their architecture on the human brain.