Needless to say, and I’m sure you already know this, but I’ll say it anyway because it’s worth emphasizing: “…whether targeting ultrasound to a given brain area increases or decreases neural activity in that area…” is a very narrow slice of the more important question of “…whether targeting ultrasound to a given brain area changes what it’s doing, and if so, what is the nature of that change…”.
Separately, it’s annoying that Martin et al. did a whole project on stimulating LGN without asking volunteers if they noticed anything funny about their vision. Or if they did ask the volunteers that question, then it’s annoying that they didn’t say anything about it in their biorxiv paper. Unless I missed it. One might cynically assume that the volunteers didn’t notice anything unusual, and the researchers left it out of the paper because it would make their experiment seem less exciting.
Speaking of which, anecdotal hearsay datapoint: I have a friend of a friend with access to a tFUS research setup. They stimulated their own nucleus accumbens, and didn’t feel anything. (I don’t know what parameters they used or anything.)
right. figuring out what's really going on in the targeted tissue is more of an animal and brain slice problem, and from a science standpoint I'd want to sort that out thoroughly before even messing around with humans, but some people want "results" in a "timely manner" so here we are.
I just meant, like, there’s a high-dimensional space of everything that a brain area could possibly be doing between its inputs and its outputs, and “the activity of this area is high vs low” is just one axis of that high-dimensional space of possibilities. For example, one type of neuron could get less active while another type gets more active, which changes something about what it computes, or whatever, I dunno.
Positive results, negative results, no results, increased blood flow, decreased blood flow, no changes in blood flow. Is it reasonable to postulate that what’s being picked up on fMRI is not changes mediated by ultrasound but instead just changes in an undulating system? Seems likely to me.
Needless to say, and I’m sure you already know this, but I’ll say it anyway because it’s worth emphasizing: “…whether targeting ultrasound to a given brain area increases or decreases neural activity in that area…” is a very narrow slice of the more important question of “…whether targeting ultrasound to a given brain area changes what it’s doing, and if so, what is the nature of that change…”.
Separately, it’s annoying that Martin et al. did a whole project on stimulating LGN without asking volunteers if they noticed anything funny about their vision. Or if they did ask the volunteers that question, then it’s annoying that they didn’t say anything about it in their biorxiv paper. Unless I missed it. One might cynically assume that the volunteers didn’t notice anything unusual, and the researchers left it out of the paper because it would make their experiment seem less exciting.
Speaking of which, anecdotal hearsay datapoint: I have a friend of a friend with access to a tFUS research setup. They stimulated their own nucleus accumbens, and didn’t feel anything. (I don’t know what parameters they used or anything.)
right. figuring out what's really going on in the targeted tissue is more of an animal and brain slice problem, and from a science standpoint I'd want to sort that out thoroughly before even messing around with humans, but some people want "results" in a "timely manner" so here we are.
Do you mind explaining what you mean on whether ultrasound changes what a brain area is doing?
Love your AI Alignment work btw! Read through a ton of it a few months ago during a 3 hour car ride.
Thanks!
I just meant, like, there’s a high-dimensional space of everything that a brain area could possibly be doing between its inputs and its outputs, and “the activity of this area is high vs low” is just one axis of that high-dimensional space of possibilities. For example, one type of neuron could get less active while another type gets more active, which changes something about what it computes, or whatever, I dunno.
Positive results, negative results, no results, increased blood flow, decreased blood flow, no changes in blood flow. Is it reasonable to postulate that what’s being picked up on fMRI is not changes mediated by ultrasound but instead just changes in an undulating system? Seems likely to me.
when the comparison is pre/post stimulation, maybe; when it's sham vs real stimulation, the idea is to rule that out
This was a very helpful post, thanks for making it!