arielzj

Even in the Netherlands, where assisted dying has been legal and normalised for over 20 years, only about 6% of people die this way - and just one in nine cancer patients, the group it's most available to. The piece digs into why so few use it, how the trends are increasing, and how the rise may not be monotonic indefinitely.

Connectomics — mapping every neural connection in a brain — is advancing fast enough that a complete human brain map may be feasible within a decade or two. But there's a problem nobody talks about: human brains donated to science are preserved hours or days after death, by which point the tissue is too degraded for the long-range neuron tracing a full connectome requires. The unexpected fix may come from preservation/biostasis/cryonics organizations preserving brains of assisted-dying patients within minutes of death.

A piece outlining how if the recent claims that Eon systems have uploaded a fly should count as valid, then it would already be the case that these previous papers had uploaded a fly as well:

-Shiu et al,

-Lappalainen et al.,

-the NeuroMechFly groups

There's a long way to go still before a true upload

...I'm not sure what this is supposed to mean. Asking about ideal conditions is a reasonable starting point for establishing a baseline, and high-quality preservation is definitely something that can be achieved under laboratory conditions with animal models (e.g. https://www.sciencedirect.com/science/article/pii/S001122401...)

These were physicians who were being paid to complete surveys - no reason to think they were more likely to select/reject this one over any of the others available to them at the time.

First author here to provide context:

1. We had both general practitioners and various specialists. We set a minimum quota for each so that we'd get some sense of what all the various groups thought. We aimed for and obtained 180 specialists along with the 150 general practitioners. See figure 1 c.

1a. We absolutely did not change the protocol after the survey was designed, or selectively exclude sections of the population. The inclusion/exclusion criteria are as per the methods, and this was a one-shot survey (no piloting beyond initial validation on colleagues). I'm all for open and reproducible science!

2. The introduction does not claim an equivalence between induced hypothermia and preservation, just that induced hypothermia does provide some precedence that preservation may work.

"Some surgical operations, such as repairs of aneurysms of the aortic arch, are performed using deep hypothermic circulatory arrest (Connoly et al., 2010). In this procedure, a patient’s body is cooled to ~20°C, experiencing both circulatory arrest and complete electrocerebral silence on EEG. Patients can survive 30+ minutes in this state with >90% survival rates and intact post-surgical cognition (Percy et al., 2009).

This demonstrates that long-term memory and personality can persist through prolonged cessation of brain activity - a finding which builds upon other neuroscience data showing that these depend on structural properties such as neuronal connectivity, rather than requiring uninterruptible electrochemical dynamics (Stecker et al., 2001).

Some terminally ill patients, based on this medical evidence and their personal beliefs about future medical advances, have requested procedures to preserve their brain structure after legal death..."

3. I was also surprised the probability estimates were as high as they were! I think doctors just think this has a greater chance of working than is commonly perceived.

As a neuroscientist, I became fascinated by the supposed phenomenon of 'terminal lucidity', where people with severe dementia suddenly become lucid again shortly before death. It sounds impossible given how catastrophically damaged their brains are, but it keeps being reported across cultures and centuries. If it's real, it could suggest new treatment methods for Alzheimer's (not just another ineffective amyloid-clearing antibody), yet it's shockingly understudied.

"The treatment works like this: doctors use a modified teleporter that targets just one cubic centimeter of brain tissue at a time. That tiny chunk gets scanned, disintegrated, and instantly rebuilt in the exact same spot - minus any disease proteins."

tl;dr: neuroscientists think there's a 40% current brain preservation techniques work well enough (under ideal circumstances) to preserve someone's brain in a way that's potentially compatible with uploading them

"Despite the last decade’s development of optogenetic methods for artificially manipulating engrams, and subsequent claims that there is a consensus that memories are stored in ensembles of synaptic connections, it remains unclear to what degree there truly is unanimity within the neuroscientific community about the neurophysiological basis of long-term memory. We surveyed 312 neuroscientists, comprising one cohort of experts on engram research and another of general neuroscientists, to assess this community’s views on how memories are stored. While 70.5% of participants agreed that long-term memories are primarily maintained by neuronal connectivity patterns and synaptic strengths, there was no clear consensus on which specific neurophysiological features or scales are critical for memory storage. Despite this, the median probability estimate that any long-term memories could potentially be extracted from a static snapshot of brain structure was around 40%, which was also the estimate for whether a successful whole brain emulation could theoretically be created from the structure of a preserved brain. When predicting the future feasibility of whole brain emulation, the median participant estimated this would be achieved for C. elegans around 2045, mice around 2065, and humans around 2125. Notably, neither research background nor expertise level significantly influenced views on whether memories could be extracted from brain structure alone. Our findings suggest that while most neuroscientists believe memories are stored in structural features of the brain, fundamental questions about the precise physical basis of memory storage remain unresolved. These findings have important implications for both theoretical neuroscience and the development of technologies aimed at preserving or extracting memory-related information."

I don't think that's true at all. We routinely provide people with surgery by highly trained specialists and ongoing indefinite care while they're still biologically active. Outside of the US, this can even be done reasonably affordable. As far as medical procedures go, the one described here is fairly simple, and cold storage is also not that expensive. This should be accessible to anyone in a developed country.

This is akin to saying that because we don't know how to unify general relativity and quantum mechanics, that we know nothing about physics. One can admit there's still a lot left to learn about a field while acknowledging the fundamentals might be reasonably well understood.

That's what's outlined in the article!

(Btw, I (OP) am not the writer of this piece, I'm just quoting it. But I am also a neuroscientist who's interested in brain preservation, for what it's worth.)

The idea of cryopreservation is simple and beautiful: When a person is beyond saving by modern methods, we can give them another lifeline. First, perfuse their bloodstream with antifreeze. Once administered, it’s possible to safely cool the body to incredibly low temperatures without ice forming inside organs and crushing delicate cells. Cold slows chemical reactions, all of them: According to the Arrhenius equation, every 10 °C decrease in temperature halves chemical reaction rates. A coolant like liquid nitrogen halves the ordinary rate of chemical reactions 22 times over. That stretches a single second of life into 48 days.

...

As you might expect, this simple idea turns out to be monstrously complicated and exacting in practice. The human body is much more finicky than the field’s founders in the 1960s had hoped. But the techniques exist. I’ve spent my life helping to develop them. The cryopreservation available today is far removed from the ideal — fussier, less elegant, and limited in what it can offer. There is much room for improvement and much work to be done. Still, it works, at least for a particular definition of working. We can’t yet warm up a frozen person and revive them, and it’s not certain we ever will. What we can do is reliably preserve memory in an information-theoretic sense. If our current understanding of neuroscience is correct, then we have the techniques to preserve all the information that makes a person who they are — albeit in a form that's impossible to extract with today's technology...

I mean, yes - that's why they're also giving out $25,000 annual research awards for milestones towards the target (some previous examples of which are described in the article)

Also a link to buy my book, mentioned in the article: https://www.amazon.co.uk/Future-Loves-You-Should-Abolish/dp/...

Link the preprint and data from survey the article is based on: https://osf.io/preprints/psyarxiv/keq7w

The consciousness research community absolutely cannot agree on a clear, principled line to draw between behaviours definitively associated with consciousness and those that are not indicative of consciousness. Candidate behaviours range from anything between verbal indications of self-awareness on the high end (only possible by humans already around a year old) through to any sign of environmental awareness on the low end (demonstrable even by bacteria).

Think it should rely on having brains? Well, jellyfish have neurons but not brains, so do they count? But hold on, even fungi have action potentials in their cells. Or on the other end of things, humans can clearly perform some complicated behaviours unconsciously, such as 'blind' people being able to see (google 'blindsight').

The hard problem of consciousness isn't going to be solved or dissolved by experimental science, because it's fundamentally a philosophical problem to be resolved through first-principles. Science can give hints, but that's all. See https://doi.org/10.1093/nc/niab001 for a proof of that statement.

Sure, it could be the case that AGI is developed progressively and slowly, the way we're seeing Waymo build autonomous vehicles. But that's just one way amongst many, and you could see it arrive suddenly through very different means, as is possible to imagine with scaled up LLMs.

I really wish people would consider all the possibilities, and assign their relative probability weights to them. Is Karpathy 100% sure it will be like self-driving cars? 50%?