Can you explain why this technology/approach is so underfunded/neglected, when some implementations seem simple/benign, and the benefits seem large?
Great question, I think it’s largely because the implementation wouldn’t be as simple as it may first appear so relatively deep pockets are required. Also, the amount of researchers in this field is pretty low (low thousands?). It’s still much simpler than invasive stimulation (e.g. Neuralink), but not something that can be implemented overnight.
The easiest headset to initially implement would use electrical stimulation, and there are devices on the market that use electrical stimulation, for example, this one for depression: https://flowneuroscience.com/
The issue is that we all have different shaped heads, skull thickness, shapes of brain etc and this can lead to up to a 100% difference in the electric field in the brain https://www.sciencedirect.com/science/article/pii/S1935861X19304115. To phrase that differently, because our brains and heads are different giving two people the same stimulation can mean one has improved intelligence and the other does not. But luckily there is a way around this, namely taking an MRI scan of the user’s head, simulating brain stimulation, then personalising the stimulation to their head and brain. This essentially gets rid of much of this variability between people by accounting for the different shape of the head and brain. The issue of course is that we can’t go and have an MRI scan when we buy this headset, it’s expensive time consuming and doesn’t scale across the population. This is where the field has sat for a few years, have personalised stimulation at great expense or don’t and have it and get poor results. Most research groups cannot afford to put every participant through an MRI, so most research on this topic has poor results.
So a company would have to create these models, validate them against people that have gone through MRI scans. Create the physical hardware, ideally track the position of the stimulation electrodes relative to the head (which is an engineering challenge in itself). Package this all up nicely, run a large scale study to prove it enhances aspects of intelligence and improves productivity then ship it. Then on a rolling basis perform more studies on enhancing different aspects of intelligence then ship that as hardware/software updates. It’s a big undertaking.
Can you explain why this technology/approach is so underfunded/neglected, when some implementations seem simple/benign, and the benefits seem large?
Great question, I think it’s largely because the implementation wouldn’t be as simple as it may first appear so relatively deep pockets are required. Also, the amount of researchers in this field is pretty low (low thousands?). It’s still much simpler than invasive stimulation (e.g. Neuralink), but not something that can be implemented overnight.
The easiest headset to initially implement would use electrical stimulation, and there are devices on the market that use electrical stimulation, for example, this one for depression:
https://flowneuroscience.com/
The issue is that we all have different shaped heads, skull thickness, shapes of brain etc and this can lead to up to a 100% difference in the electric field in the brain https://www.sciencedirect.com/science/article/pii/S1935861X19304115. To phrase that differently, because our brains and heads are different giving two people the same stimulation can mean one has improved intelligence and the other does not. But luckily there is a way around this, namely taking an MRI scan of the user’s head, simulating brain stimulation, then personalising the stimulation to their head and brain. This essentially gets rid of much of this variability between people by accounting for the different shape of the head and brain. The issue of course is that we can’t go and have an MRI scan when we buy this headset, it’s expensive time consuming and doesn’t scale across the population. This is where the field has sat for a few years, have personalised stimulation at great expense or don’t and have it and get poor results. Most research groups cannot afford to put every participant through an MRI, so most research on this topic has poor results.
Instead, a prospective startup needs to find a way to personalise the stimulation without an MRI scan. One way is to use AI to generate an MRI scan based on the shape of the persons head, their demographics and maybe even their DNA (see https://developer.nvidia.com/blog/kings-college-london-accelerates-synthetic-brain-3d-image-creation-using-ai-models-powered-by-cambridge-1-supercomputer/?ncid=so-twit-448517#cid=ix11_so-twit_en-us). The other way is to create a model where you give it someones head shape, demographics and / or aspects of their DNA and it tells you what kind of stimulation would work for them given its training in simulation. Early versions of this already exist! For example, this paper takes head circumference and can tell you how much to stimulate reducing the inter-person variability by around 25%.
https://www.sciencedirect.com/science/article/pii/S1935861X21001352
So a company would have to create these models, validate them against people that have gone through MRI scans. Create the physical hardware, ideally track the position of the stimulation electrodes relative to the head (which is an engineering challenge in itself). Package this all up nicely, run a large scale study to prove it enhances aspects of intelligence and improves productivity then ship it. Then on a rolling basis perform more studies on enhancing different aspects of intelligence then ship that as hardware/software updates. It’s a big undertaking.