Networking our brains?

The plan of Elon Musk to defend humanity from AI evolution appears to be bold: We all shall become Sensates by design and through engineering. For real. Constructive collaboration and instant understanding and empathy for everybody. It sounds like a dream championed by the Wachowski sisters (The Matrix) and Michael J. Straczynski (Babylon5) in a (all too early cancelled) Netflix series SENSE8 spanning the globe, gender, skills, emotions and mind-bodies of 8 people becoming connected telepathically. A cluster. Elon Musk wants to accomplish creating human clusters with technology to protect us from the singularity – AI which knows that it knows and starts to think and act on it´s own behalf. He dreams to decipher and apply the code of human thought with brute force and some material innovation / industrial approach. After wading through the reviews so far – that may even work. Or not. To dive into the depth of the converging field I would recommend an in-depth read of 38.000 words by wait-but-why author and stick figure artist Tim Urban about why Elon wants/needs/suggests to develop a “wizard´s hat” for humanity. Have fun with approximate 190pages “article” to put all “in a nutshell – maybe the best summary I haver encountered so far about evolution, the brain, BMI’s and AI. (see@

What hundreds of institutes and ten thousands of scientist with low-bandwidth slow language tools could not do worldwide so far (or only with slow progress) might be in the grasp of a notorious visionary outsmarting whole industrial complexes, creating agendas nobody can ignore. In mass scale. Musk may see brainscience as an engineering problem, where you solve the application first and then backpropagate or learn “why” it works or narrow down statistically with machine learning how concepts do work. Alone that approach, successfully applied with SpaceX and Tesla, is interesting enough, to hack the big question “what is the brain?” to find more out about this tool we all use when we plan and remember these endeavours. How the brain forms individual representations, processes and stabilises concepts to make sense of the world is not a trivial problem. The impairment through metaphoric, compressed language which did not really get an update in 50.000 years feels like the right analysis. We are helplessly in a disadvantage in regard to communication. But the brain is a powerful “processing unit” – not a computer, but a device which creates world views, does some fuzzy math pretty well and got evolved through millions of years, not just through 100. Every one of us boasts 80-100 billion units, nerve cells called neurons which communicate with each other and our sensors in our head and body at the speed of up t 100m/s. That potential we need to tap in directly, Musk suggests. This July his start-up Neuralink (, founded in 2016 has just announced to go into a round of recruiting for starting experiments in humans next year. The base is a contemporary understanding of the human brain:

“Neurons represent information in the rate statistics and precise timing of spikes.”

Elon Musk / Neuralink Launch Event / 16.July 2019

It looks like there must be already a profound error in this first underlying assumption, one may suspect. Descartes has being accused to have made one, a grave error, pointed out by contemporary cognitive scientists like Antonio Damasio (Descartes´ Error – emotion, Reason and the Human Brain, Avon Books, 1994): are we only imagining an observer in the brain or is there truly one, a spectator who happen to watch the unfolding action / play on the stage of consciousness like a guest watching his own experiences? It turns out that our consciousness is not well understood as a stage, Descartes error is misleading and misrepresents what is happening due to contemporary cognitive neuroscience. But may Elon Musk heading into the theme with 3% of his time and some of his money applying neuro-tech make a similar one? Are we able to understand and later also manipulate our concepts by only looking at the firing pattern with electrodes? Is there a stage for that, is this technically accessible or do we have to go a dimension deeper and look at the roughly 7000 synaptic connection each nerve cell has? That would complicate the challenge significantly as it would mean to monitor a lot more: the estimates in numbers of synapses vary for an adult, ranging from 1014 to 5 x 1014 synapses (100 to 500 trillion). The chances that we accomplish to listen to the changing pattern of synaptic connections and their “weights” in a foreseeable future are some magnitudes lower – more in he realm of science fiction.

But, there is some recent evidence that we might have a good chance to decode and recode thinking through a paradigm called “thought identification” in a broader scale. Starting in 2011 (Nishimoto 2011 or later Haynes & Geraint 2014) ;researchers gathered results that Even if he does not say it openly, to start a company and raise millions of dollars to boldly go where no one has gone before is surely based on science and results, not just assumptions. There have been studies for more than 10 years indicating that reading concepts in different humans with fMRI studies and machine learning, so why not make this an applicable technology for inter-human communication and create robust brain-machine interfaces getting even closer to the brain and rid of big clunky machines? Marcel Just and colleagues can distinguish what people think just with analysing brain data, may it be objects (hammer, chair), food (apple, banana), emotions (fear, anger, love) or abstract concepts in physics. 

The representation of concepts in the brain is not an unsolved mystery anymore and should have stirred more applause, outcry or critique, but mostly went unnoticed outside of scientific circles. Memory studies strongly points to smaller structures than just a single neuron firing as the seat for long term concepts. But firing patterns in neuronal assemblies or cortical columns made of thousands of neurons may be the representation for experience and recognition of concepts and therefore knowledge. Our connections may be in constant re-construction but also specifically pruned and modelled through experience and learning. What the connection points of nerve cells, the so called synapses do is making a certain firing pattern or neurons more likely, distributed over the whole cortex and deeper buried brain structures. This distribution shows in the fMRI already in a quit low temporal resolution. The connectivity and the weight of synaptic connections might be the underlying structure for the “precise timing of spikes”. What can be predicted by machine learning already in relatively high spatial resolution over a longer time like with fMRI may be a statistical measure for the thought of a cow or a chair. It represents pushing the buttons in the right sequence so that right firing pattern emerges which we call an experience. See the quote of Miller in “The Expanse” in the last post. 

Electrodes directly applied over the whole brain and listening to as many neurons firing as possible, not just the blood flow like with fMRI as a stand-in for “neural activity” might give a much more nuanced window into thoughts. If thought identification can cracked with tiny, bluetooth enabled 10.000 electrode arrays in just a tiny part of the brains surface becoming incrementally better is to be seen. But, extrapolating and leaving ethical complications aside – given a big sample and a lot of human data acquired in a non-invasive everyday way, he might become the google of neuroscience through it, building a human colossus / golem, enabling human/machine hybrids with millions of high bandwidth brain interfaces. It may be a way how to tame silicon AI with mass collaboration of wetware – by starting a co-evolution. By increasing our output bandwidth.

We BECOME an interconnected human hive mind, BEFORE artificial intelligence becomes self-aware and potentially is side-lining us without blinking. An upgrade before we become a footnote in planetary history or a pet at the whim of AI overlords, just applying our optimisation hysteria/history or inventing even more optimal solutions which do not include us. The human cat version Musk actually considers a good outcome. Usually the more intelligent being in the food / resources chain takes it all and eliminates all the other competitors quickly. The main argument, that we are not fast enough to compete with machine learning algorithms, but having an organ which is quite capable of “computing” available in billions units world-wide is a technical approach of – ok, but what do we have? But so far all these units or brains/bodies live like in a vat, very much isolated from each other only reachable/changeable through language and low-bandwidth, slow media –  yes, the arts, literature, sex, films and poetry tries.

Understanding each other means after the upgrade, that we can hear/see/feel each other thinking, that we would be connected like Sensates, envisioned by the Wachowskis. We would need invitation from others to really “participate” in one´s experiences, hopefully, but economy and bad people (like Milton Bailey Brandt, “Whispers” or “the Cannibal” in the series) might wanna exploit that. And trolls. And governments, and military, and… it could become a nightmare or an evolutionary step…the biggest one in 50.000 years.


Haynes, JD; Geraint, R. (2014)  “Decoding mental states from brain activity in humans”Nature Reviews

Just, M.  (2012)  What brain imaging can tell us about embodied meaning Symbols and Embodiment: Debates On Meaning and Cognition. Oxford Scholarship Online DOI: 10.1093/acprof:oso/9780199217274.003.0005

Just M, Varma S. (2007) The organization of thinking: what functional brain imaging reveals about the neuroarchitecture of complex cognition. Cognitive, Affective & Behavioral Neuroscience. 7: 153-91. PMID 17993204 DOI: 10.3758/CABN.7.3.153

Nishimoto, S.; Vu, An T.; Naselaris, T.; Benjamini, Y.; Yu, B.; Gallant, JL (2011), “Reconstructing Visual Experiences from Brain Activity Evoked by Natural Movies”, Current Biology21 (19): 1641–1646

Wachowski, L, Wachowski. L, Straczynski, JM (2015-218) Sense8, Netflix  retrieved at 23.07.2019, 21.31


Scroll to Top