Hi All

In the 1960s a lone neuropsychologist posited that individual neurones store or represent individual concepts - for example your grandmother. Thus was born the semi-perjorative phrase "grandmother neurons" to described this - at the time - overly simplistic theory.

But now we can actually measure individual neuronal activity it seems to really be true - concepts have neurons of their own...

Grandmother Neurons

...which might not seem that space related until you realise that it is another step towards representing human-like intelligence in silico. A Belgian theorist of the brain, Armand de Callatay, argued for grandmother neurons in a fascinating book on brain-modelling which I recommend even a browse of...

Artificial and Natural Intelligence

...in which he attempts a high-level model of the brain rather than the usual "reduce to neurons" approach. Yet he obviously has a lot to say about neurons too, such as the grandmother neurons and also the locus of long-term memory in the equivalent of simple "switches".

That's one idea I can't let go just yet - I'm not convinced that some vague "long-term potentiation" is the cause of memory. Instead I think that neuronal connections change physically and "lock in" the new memory - if it is repeated, reviewed after short-term storage in the hippocampus.

That aside it means that real AI space-machines inch closer to reality, the ultimate being the human-level (and beyond) von Neuman machines some imagine populating our Galaxy. While I take issue with that particular naive image I think there's potentially a lot of vNM activity, but even over the age of the Solar System it need not have been a large fraction of our system's free mass - as Chris Boyce argued cogently over 25 years ago in his "Extraterrestrial Encounter".

But also understanding our brains and the ability to record and analyse them leads us towards real immortality.

Adam