Averaging is a convenient fiction of neuroscience

https://www.thetransmitter.org/neural-coding/averaging-is-a-convenient-fiction-of-neuroscience/

by domofutu • 2 years ago

110 104 2 years ago

glial2• 2 years ago

All models are convenient fictions. I heard a neuroscientist once describe averaging as a low-pass filter. People know it hides high-frequency dynamics. But unless you have a way to interpret the high-frequency signal, it looks an awful lot like noise.

sroussey• 2 years ago

I think of summaries as the text equivalent of averaging. Some high frequency stuff you don’t want to loose in that case are things like proper names, specific dates, etc. In the face of such signal, you don’t want to average it out to a “him” and a “Monday”.

datameta2• 2 years ago

That would be a median in your example no? A spurious average might be us thinking that the statistical mean word contains all vowels except for 'e', and that 'm' is twice as likely as the other most likely consonants.

bitwize• 2 years ago

That makes a lot of sense. Thank you for this analogy.

We use Conscrewence at work for internal documentation, and when I pull a page up it wants to recommend an AI-generated summary for me. Uh, no, Atlassian, I'm on this page because I want all the details!

heyitsguay• 2 years ago

My grad school research was with an NIH neuroscience lab studying low-level sensory processing that offered a fascinating perspective on what's really going on there! At least for the first few levels above the sense receptors in simpler animal models.

To oversimplify, you can interpret gamma-frequency activity as chunking up temporal sensory inputs into windows. The specific dynamics between excitatory and inhibitory populations in a region of the brain create a gating mechanism where only a fraction of the most stimulated excitatory neurons are able to fire, and therefore pass along a signal downstream, before broadly-tuned inhibitory feedback silences the whole population and the next gamma cycle begins. Information is transmitted deeper into the brain based on the population-level patterns of excitatory activity per brief gamma window, rather than being a simple rate encoding over longer periods of time.

Again, this is an oversimplification, not entirely correct, fails to take other activity into account etc etc, but I'm sharing it as an example of an extant model of brain activity that not only doesn't average out high-frequency dynamics, but explicitly relies on them in a complex nonlinear fashion to model neural activity at the population level at high temporal frequency in a natural way. And it's not completely abstract, you can relate it to observed population firing patterns in, e.g., insect olfactory processing, now the we have the hardware to make accurate high-frequency population recordings.

robwwilliams3• 2 years ago

By “low level” do you mean in the thalamus or cortex or something else. Live a citation. I initially thought that “low level” meant at the level of receptors and the first few synapses. But to the best of my knowledge gamma oscillations will not play a roll in the periphery.

It would be great if you had a citation. I have been reading Karl Friston’s work all day.

jtrueb• 2 years ago

It is a low-pass filter in the frequency domain with a roll-off that is not smooth. I quite like [1] as a quick reference.

https://www.analog.com/media/en/technical-documentation/dsp-...

etrautmann2• 2 years ago

Not the OP but we're talking about different things here. Much of the concern about averaging is about averaging across trials. Smoothing a spike train over time isn't really the issue that this thread is concerned with, since that's just averaging successive samples within some small window.

datameta2• 2 years ago

In physics the model we choose is based on the scale - as in the macro sense all quantum effects average out over the several sextillion atoms in, say, a wood screw.

ggm2• 2 years ago

But unless you have a way to interpret the high-frequency signal, it looks an awful lot like noise.

In other words, they're looking for their lost keys under the lamp-post because it's easier there. If there is a signal in the HF, it's not yet understood. This feels like "junk DNA" -which is I believe receiving more attention than the name suggests.

JumpCrisscross5• 2 years ago

they're looking for their lost keys under the lamp-post because it's easier there

This is a strange criticism. If you're looking for your keys in the dead of night, and there is a lamp post where they might be, you should start there.

The streelight effect criticises "only search[ing] for something where it is easiest to look" [1]. Not searching where it's easiest in all cases.

In this case, we know averaging destroys information. But we don't know to what significance. As the author says, "we now have the tools we need to find out if averaging is showing us something about the brain’s signals or is a misleading historical accident." That neither confirms nor damns the preceding research--it may be that averaging is perfectly fine, hides some of the truth that we can now uncover or is entirely misleading.

[1] https://en.wikipedia.org/wiki/Streetlight_effect

etrautmann2• 2 years ago

This is broadly speaking not correct. If you average together a bunch of trials with variable timing, then the result can tend to wash out higher frequency components (which you might not have realized were in the data), but trial averaging is not a low pass filter at all. There are some nice methods to recover temporal structure that changes across trials prior to averaging, like:

https://www.sciencedirect.com/science/article/pii/S089662731...

UniverseHacker17• 2 years ago

I've become increasingly convinced that the idea of averaging is one of the biggest obstacles to understanding things... it contains the insidious trap of feeling/sounding "rigorous" and "quantitative" while making huge assumptions that are extremely inappropriate for most real world situations.

Once I started noticing this I can't stop seeing this almost everywhere- almost every news article, scientific paper, etc. will make clearly inappropriate inferences about a phenomenon based on the exact same mistake of confusing the average for a complete description of a distribution, or a more nuanced context.

Just a simple common example, is the popular myth that ancient people died of old age in their 30s, based on an "average life span of ~33 years" or such. In reality the modal life expectancy of adults (most common age of death other than 0) has been pretty stable in the 70s-80s range for most of human history- the low average was almost entirely due to infant mortality.

The above example is a case where thinking in terms of averages causes you to grossly misunderstand simple things, in a way that would be impossible even with basic common sense in a person that had never encountered the idea of math... yet it is a mistake you can reliably expect people in modern times to make.

KK7NIL2• 2 years ago

I think you'd enjoy this video on different types of means and their applications: https://youtube.com/watch?v=V1_4nNm8a6w

mcmoor• 2 years ago

It's just trying to assume normal distribution when it's not normal. Modern science rely so much on that distribution that i actually whether they have overestimated its ubiquity just because it's so damn convenient to use.

Jweb_Guru2• 2 years ago

While it is true that not all distributions are normal, many distributions are approximately normal (or at least, normal in some sensible space that maps onto the actual collected data). IMO the amount of ink spilled on the idea that science is fundamentally flawed because distributions aren't always normal is probably too high (especially among non-statisticians), and frankly that's not where statistical analyses usually go wrong. A much larger problem (that has nothing to do with the ultimate shape of the distribution) is stuff like postselecting from a set of plausible models until you find one that finds significant results, and claiming that was what you intended to measure all along (this is why it's important to consider stuff like hyperpriors, much moreso than lack of normality).

UniverseHacker17• 2 years ago

Yes, this is the actual issue- assuming a normal distribution in cases that are not normally distributed. People also even leave off the standard deviation, and seem to be mentally thinking of the average as a point rather than considering even the “width” of the normal distribution- so the overemphasis of the average even causes people to misunderstand things that really are normally distributed.

hinkley2• 2 years ago

Averages are very bad in bimodal distributions.

And that includes issues of public policy, where going left sort of works, and going right sort of works, and going in the middle sucks for absolutely everyone.

JumpCrisscross5• 2 years ago

Averages are very bad in bimodal distributions

They're bad with multimodal distributions, generally, as well as any random process governed by a distribution with no mean, e.g. Cauchy [1].

(Neuronal firing appears to be non-Gaussian, possibly lognormal [2], which does have a mean [3], but it isn't equal to the simple average.)

[1] https://en.wikipedia.org/wiki/Cauchy_distribution#Properties

[2] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6633220/

[3] https://en.wikipedia.org/wiki/Log-normal_distribution

bosma• 2 years ago

In reality the modal life expectancy of adults (most common age of death other than 0) has been pretty stable in the 70s-80s range for most of human history- the low average was almost entirely due to infant mortality.

This isn't true: https://ourworldindata.org/its-not-just-about-child-mortalit...

Spooky233• 2 years ago

It mostly is. The biggest gains are in childhood. Aldo consider that you’re looking at figures for England and Wales, which isn’t necessarily representative.

The largest contribution to improving life expectancy is measured by reductions in child mortality. The factors that drove those improvements (infection control, improved hygienic practices, food quality, regulation of food and drug purity, medicine) benefited everyone, but had the biggest impact on the old and young. In the 1850s, 8,000 infants died annually from adulterated milk alone.

I think of my own journey. In 1824, 200 year ago Spooky23 would have died 20 years ago, a half blind cripple. 2024 Spooky23 is healthy with no back issues and god willing a few more decades.

UniverseHacker17• 2 years ago

What I was saying is in reference to this study, which suggests a modal life span of about 72 years of age in the paleolithic: https://onlinelibrary.wiley.com/doi/10.1111/j.1728-4457.2007...

usednoise4sale• 2 years ago

Our world in data pushes a clear agenda, and it isn't really to be trusted.

Consider: https://ourworldindata.org/child-mortality-in-the-past

From this article: "Researchers also collected data about hunter-gatherer societies. The 17 different societies include paleolithic and modern-day hunter-gatherers and the mortality rate was high in all of them. On average, 49% of all children died.[5]"

This is cited as coming from: https://www.sciencedirect.com/science/article/abs/pii/S10905...

Which categorically states: "Unfortunately there simply is not enough direct paleodemographic archaeological data to make definite claims about the global patterns of infant and child mortality rates of our Paleolithic hunter–gatherer ancestors."

The author of the Our World in Data piece seemingly intentionally conflates the proxy with actual archaeological evidence of the actual child mortality rates. Given the clear warning in the cited article about making definite claims, I cannot read the deception any other way.

After seeing this error, I do not know how you could possibly trust anything they have to say on the matter.

llm_trw8• 2 years ago

In reality the modal life expectancy of adults (most common age of death other than 0) has been pretty stable in the 70s-80s range for most of human history- the low average was almost entirely due to infant mortality.

This is even wronger than what you critique.

For every period in history that we have good data for people had a half-life - a period in which you'd expect half of all people to die: https://batintheattic.blogspot.com/2011/07/medieval-populati...

For example in medieval Germany it looked something like:

    |   Age | Half-life |
    |  0-10 |        10 |
    | 10-20 |        40 |
    | 20-40 |        20 |
    | 40-80 |        10 |

It's called a population pyramid not a population column for a reason.

The exact age varies by location, but even if we ignore everyone under 10, half of all people left would still die before they are 40.

JumpCrisscross5• 2 years ago

if we ignore everyone under 10, half of all people left would still die before they are 40

Wouldn't it be 50 since the half-life is an interval?

UniverseHacker17• 2 years ago

What you said in no way conflicts with what I said. For example, if people have dangerous lives in a way that is unconnected to age they will tend to not live long, yet the modal life expectancy due to the additional mortality of actual old age can still be quite old.

What I was saying is in reference to this study I read long ago, which suggests a modal life span of about 72 years of age in the paleolithic: https://onlinelibrary.wiley.com/doi/10.1111/j.1728-4457.2007...

wodenokoto• 2 years ago

Parent is talking about mode, so the most common number. Let’s build a distribution that satisfy both.

Oldest age is set 80, mode is set to 75. We start by distribution one percentage point to each age. Then add an extra to the mode and all the remaining to zero.

We now have a distribution where the most common age of death, other than zero be in 70-80, and more than half the population die before they reach 40.

SkyBelow• 2 years ago

Is there an issue with how the data is grouped? At 19.9, you have a 50% chance of living to 59.9. But less than a season later at 20.1, you now only have a 50% chance o living to 40.1. How can the former be right if the latter is right?

The growth in life expectancy from surviving early childhood makes sense, but the decline in life expectancy for crossing 20 feels too sharp.

Mistletoe• 2 years ago

I was at the Denver museum’s mummy exhibit and disturbed to see that they said the lady died at 30, a “normal age for death in those times”. You would think a museum should know better.

https://aeon.co/ideas/think-everyone-died-young-in-ancient-s...

orwin• 2 years ago

Half the people died before 40 in the middle ages, discounting pre-5 mortality. I would assume it is lower for female, as first pregnancies had a 10% mortality rate (this rate increased after the middle age, until germ theory), and following ones carried a 2-5% mortality rates.

UniverseHacker17• 2 years ago

Great article that pretty well explains the situation with modern vs ancient life expectancy.

ddfs123• 2 years ago

In reality the modal life expectancy of adults (most common age of death other than 0) has been pretty stable in the 70s-80s range for most of human history-

I am pretty sure this is wrong. East Asian cultures celebrate 60th birthday as becoming very elderly, and if you live to the 70s it's almost as if you achieved Buddhahood.

JumpCrisscross5• 2 years ago

Your beef appears to be with simple averages, not averaging per se.

The average for life expectancy is the mean of the Gompertz distribution [1]. Specifically, one that is "left skewed and with a flattened slope for ages under 50 years for men and 60 years for women," which proceeds to become "more right skewed and leptokurtic" as age increases [2].

So a simple average in the <55 domain would underestimate the mean while in the >55 domain it would overestimate it. Which is almost comically problematic when comparing ancient societies that had a median age below that level to modern ones above it.

the modal life expectancy of adults (most common age of death other than 0) has been pretty stable in the 70s-80s range for most of human history

Not quite. 63 in 1900 to 83 in 2000 (in Sweden). Bigger differences when you go further back.

[1] https://en.wikipedia.org/wiki/Gompertz_distribution

[2] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2652977/

[3] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3000019/ Figure 1

UniverseHacker17• 2 years ago

What I was saying is in reference to this study, which suggests a modal life span of about 72 years of age in the paleolithic: https://onlinelibrary.wiley.com/doi/10.1111/j.1728-4457.2007...

Life spans were abnormally short in medieval Europe, even compared to ancient humans.

dalmo3• 2 years ago

I came to the same realisation about a decade ago, after being a "science" enthusiast growing up. As you said once you see it you can't unsee it. Most of science is just a scam. The exceptions are those fields backed up by real world engineering. All of social and most of biological sciences are worse than useless, they are outright dangerous.

hifromwork• 2 years ago

Most of science is just a scam

Most of the scientists are not scammers. So if you believe that all science disciplines other than engineering are wrong[1], you should use another word that doesn't suggest researcher malice.

[1]Which is a very strong statement, because you claim to be an expert in all science disciplines at once.

UniverseHacker17• 2 years ago

I think you are overreaching with that. Science at its core is curious and open minded problem solving combined with an anti-authoritarian skepticism- you accept things once you can understand and confirm them for yourself. Ideally, it is also Bayesian- e.g. considering all of the evidence, even weak evidence like personal anecdotes, while correctly keeping track of how strong or weak each bit of evidence really is.

You have the courage to trust what you understand deeply even if other powerful authorities disagree- and take the responsibility to make sure you do actually understand deeply (the conspiracy theory crowd is missing the second bit). As a practicing academic scientist, I feel like the vast majority of my colleagues with research focused academic careers also see it this way.

Nearly everything called "science" in popular culture (including school science classes, pop science books, etc.) is actually a sort of dogmatic religion that idolizes science. One of the weirdest things for me, is coming on here and when I share the type of creative thinking that actually leads scientists to new hypotheses, I am insulted and accused of being an "idiot" or "anti-science" by mostly career programmers for whom science is a religion, and not a creative process. When they talk to a scientist, they see someone that doesn't align with their fixed dogmatic views, and label it as basically the opposite of what they are actually encountering- real science looks like "pseudoscience" to them. These people would even attack someone like Isaac Newton as an idiot for daring to discuss something new that seems "weird" and idolize, e.g. a medical doctor repeating some old official stance from an institution that has been thoroughly disproven by new research.

ants_everywhere3• 2 years ago

Yeah it's a common mistake, but this is like intro to stats stuff. It's not some big secret that if you summarize a distribution with a single number then you've lost information.

I've become increasingly convinced that the idea of averaging is one of the biggest obstacles to understanding things.

I'd counter that it's easily one of the biggest assistants in understanding things. The Central Limit Theorem in particular has been enormously influential. Without averaging statistical mechanics and thermodynamics would have been impossible and with them would go the industrial revolution.

What you're noticing is one kind of mistake caused by lack of literacy in science. There are many many more similar mistakes. The solution isn't less literacy.

ithkuil5• 2 years ago

The solution isn't less literacy.

This.

It's interesting how a common instinct to seeing problems is way too often to "destroy and rebuild" instead of "correct and improve". "Institutions have failed us?" A) "Let's burn them down" vs B) "Let's have better institutions, keep what's good, fix what's bad"

I get it that sometimes, rarely, it's impossible to improve something because it's soo rotten and entrenched that you literally have to destroy and rebuild from scratch, but it's almost never the case.

UniverseHacker17• 2 years ago

this is like intro to stats stuff

Amazing how the replies to my initial comment range from angrily calling me an idiot for saying something so obviously and egregiously wrong, to claiming it is so trivial and obvious that I should not have needed to mention it.

The solution isn't less literacy.

Indeed, I can see how what I wrote could be taken that way, I also don't think so. The problem is combining an inadequate amount of literacy, with an almost religious belief in its absolute correctness, e.g. scientism.

domofutu• 2 years ago

Averages can definitely oversimplify things, especially in neuroscience where outliers often tell the real story. Taleb touches on this in Antifragile—focusing too much on the average can make us miss what’s happening at the edges, where the most interesting things are. Instead of leaning on averages, we might get more insight by paying attention to the extremes, where the real nuances are hiding.

UniverseHacker17• 2 years ago

Yes, I like Taleb's suggestion of considering a "fat tailed" Cauchy distribution- as to not assume extreme outliers can't exist

UniverseHacker17• 2 years ago

@ flagged aaron695:

The brutal reality is you don't have the IQ to understand averages or statistics like most people.
Most of our ancestors lucky enough to made it to 45 years old in human history did not make it to 70.
Using mode is misleading with "age", and you quickly showed you didn't understand 'the con' when you accidently tried to apply it.
This is a political, the "past was wonderful fantasy" which is anti-science. It's used by the Woke for instance."

In this case I am using an equally "wrong" model on purpose while being well aware of its limitations, just to make a specific point. It highlights a point where people are doing exactly what you are accusing me of- romanticizing the present, and not understanding the reality of how it actually differs from the past. E.g. what are the actual reasons people had shorter life spans in the past, and what their lives were actually like. One should not forget the limitations of such a simple model, which was basically my point in the first place.

I am fascinated by "evolutionary medicine" and using such ideas as a hypothesis generator to figure out ways to treat "modern diseases of civilization." I am in no way romanticizing the past, but trying to understand the specifics, to better figure out how to develop more effective modern day medical treatments, not to return to the past. In truth I despise political thinking altogether, and like to look at mechanisms and biological details.

Your post smacks of "scientism" which is incompatible with actual practice of science. The very idea that a certain line of thinking or theorizing is "anti science" or should be taboo for political reasons is itself incompatible with creative open minded problem solving.

I can see it was a mistake to use this specific example for discussing the problem with averages- ironically because it is so accurate. Since so many people on here hold the exact misunderstanding I was criticizing, they are getting angry and insulting me instead of my intent, which was to explain a phenomenon and have this click as a simple example of it. A less charged example, where people don't have strong opinions already would have been better.

throw_pm23• 2 years ago

I've heard this argument a million times, but I am very skeptical: where would the reliable data on infant mortality in ancient times come from? (so that it would allow us to compute precise values of average lifespan). All we have from those times are a few bone samples and a few anecdotes preserved in fragments.

I'm not saying anything for or against the ~33 years claim, just that I doubt that it comes from a precise estimate of expected lifespan at birth.

achierius• 2 years ago

All we have from those times are a few bone samples and a few anecdotes preserved in fragments.

This is the bit that you're missing: antiquity is generally considered to extend all the way into the 500-600s AD, with the Roman empire dominating its center and 'late antiquity' covering the whole early period of the Byzantine empire. This means that even just in Europe we have extensive documents (e.g. in the Christian era, baptism and burial records), as well as a wealth of burial evidence from which to extrapolate lifespans. If you look into the sources for Rome you'll find Egypt in particular come up a lot, which isn't an accident: the dry climate and extensive use of papyrus means that the region preserved far more records (and bodies) than any other part of the empire.

nosianu2• 2 years ago

In a Gresham College video lecture about microbial disease (if only I remembered the title...) the lecturing professor briefly mentions that in Victorian Britain life expectancy was actually "higher than today" when you exclude children and women (who had a great risk of death during pregnancy and birth).

https://www.gresham.ac.uk/watch-now/return-microbes-how-infe...

The video has a text with the words.

Look for

So, your life expectancy at five, in England, as a male, in 1870 was slightly longer than it is now, which is an extraordinary statistic, slightly shorter then if you were a female.

1870s Britain is not "ancient", but as far as medicine is concerned much closer to that than to modern times. What medicine there was wasn't even available to most people. They were pretty good with "physical" things like broken bones by then, but anything concerning microbes probably wasn't much better than what was done a thousand years earlier.

Obviously, as you point out, we don't have statistics from ancient times, only some hints, but that is the same for people making the claim about the short average life spans (without the little detail about infant and women mortality).

There also is the question of how meaningful it is to average over vastly different sections of the population (what you did and how you lived and what you ate, access to good water, hygiene), or times of famine or war.

Some related discussion, reddit, but the heavily moderated AskHistorians subreddit: https://www.reddit.com/r/AskHistorians/comments/12o4py/what_...

There we see the lower life expectancy of early agricultural societies vs. hunter-gatherers (ignoring their high number of deaths due to violence).

I think all of this also serves to illustrate OPs point about the problem of averages :-) There's just too much important information that is lost.

bell-cot• 2 years ago

In reality the modal life expectancy of adults (most common age of death other than 0) has been pretty stable in the 70s-80s range for most of human history- the low average was almost entirely due to infant mortality.

Until the last few-ish generations, pregnancy and childbirth have been leading causes of death for women in those in-between decades of their lives.

(And obviously War and Famine, too, for both genders.)

KK7NIL2• 2 years ago

Very interesting how measurement limitations drive scientific consensus.

The author portrays this as a major flaw in neuroscience, but it seems like a natural consequence of Newton's flaming laser sword; why theorize about something that you can't directly measure?

robwwilliams3• 2 years ago

Great note Mark. I agree. Action potentials are noisy beasts but much may be hidden in spike time coding that is obscured by averaging.

There is an even lower level problem that deserves more thought. What timebase do we use to average, or not. There is no handy oscillator or clock embedded in the cortex or thalamus that allows a neuron or module or us to declare “these events are synchronous and in phase”.

Our notions of external wall-clock time have been reified and then causally imposed on brain activity. Since most higher order cognitive decisions take more than 20 to 200 milliseconds of wall clock time it is presumptuous to assume any neuron is necessarily working in a single network or module. There could be dozens or hundreds of temporally semi-independent modules spread out over wall clock-time that still manage to produce the right motor output.

RaftPeople2• 2 years ago

There is no handy oscillator or clock embedded in the cortex or thalamus that allows a neuron or module or us to declare “these events are synchronous and in phase”.

Brains waves drive synchronization of groups of neurons, lower frequencies broader, higher frequencies more localized.

robwwilliams3• 2 years ago

That is uncertain. They must be a product of underlying processes, but the mechanisms are still opaque.

Gamma oscillations only run at about 40 Hz. That is not fast enough to clock neuronal computations or integrations in the 1 to 10 msec range.

Oscillations may have role in binding at larger scales. And when we use the word “synchronize” we generally seem to mean “given wall-clock time”.

Two neural events separated by 20 msec can be functionally coherent but may neither be in a particular phase relation or concurrent from an observer’s wall-clock perspective. Neuronal activity may not care about the observer’s timebase.

hinkley2• 2 years ago

There's an old case study from aerospace that shows up sometimes in UX discussions, where the US military tried to design an airplane that fit the 'average' pilot and found that they made a plane that was not comfortable for any pilots. They had to go back in and add margins to a bunch of things, so they were adjustable within some number of std deviations of 'average'.

stonethrowaway• 2 years ago

They used those original average measurements to design seats for passengers instead.

robertclaus• 2 years ago

As a computer scientist, I was blown away the first time my friend explained to me that his research focused on the timing of neuron spikes, not their magnitude. After talking about it for a while I realized that machine learning neural networks are much closer to simple early models of how neuron's work (averages and all), not how neuron's actually signal. Makes sense when you consider how the latest LLM models have almost as many parameters as we have neurons, but we still seem pretty far from AGI.

dilawar• 2 years ago

In many, perhaps most, signalling pathways, amplitude doesn't matter much (it does at log-scale). Given how well we control temperature and therefore rate of the reaction, it makes sense to use timing to fight off the noise.

llm_trw8• 2 years ago

Put another way, there is a reason why FM radio sounds much better than AM radio.

https://www.youtube.com/watch?v=AzvxefRDT84

lamename2• 2 years ago

Yes and no. An alternate perspective is that the output of each neuron in an artificial neural net is analogous to an F-I curve in a real neuron (spike frequency-input DC current curve). In this way, different neurons have different slopes and intercepts in their FI curves, just as a network of ANN neurons effectively have their activation functions tweaked after applying weights.

I usually only say this to other neuroscientists who have a background in electrophysiology. The analogy isn't perfect, and is unnecessary to understand what ANNs are doing, but the analogy still stands.

Nahtnah• 2 years ago

Eh, the analogy is farrr from perfect. You're basically assuming you can reduce neurons to LTI systems. Which you obviously very much cannot.

ithkuil5• 2 years ago

An elephant brain has 3 times as many neurons as a human.

They are pretty smart animals but so are dogs who have way less neurons. The point here being that the number of neurons is just one of the many factors that determines intelligence (general or not)

smokel• 2 years ago

Even among humans with roughly similar neuron counts, there are notable examples of individuals displaying extreme stupidity.

Ey7NFZ3P0nzAe2• 2 years ago

80% of human neurons are actually in the cerebellum and are not related to consciousness at all

yberreby• 2 years ago

the latest LLM models have almost as many parameters as we have neurons

I often see this take, but the apt comparison is between parameter and synapse count, not neuron count. You should be counting hidden units rather than weights if you want to compare to neuron counts.

markhahn• 2 years ago

I think you should compare synapse and weight counts if you want a measure of the network's state/capacity. If you want something closer to its compute power, compare neurons and hidden units.

Bjartr• 2 years ago

To expand on that, as a point of comparison, a single neuron can have thousands of synaptic connections. So we're still a few orders of magnitude out from modeling NNs that have a degree of connectivity similar to the brain, even though the synapse counts are comparable.

a_c• 2 years ago

Human neural network build and trim connections constantly [1]. I imagine we will get much closer to AGI if we can update models dynamically, instead of just adding more neurons and more training. After all human didn't need reading billions of articles before writing an average one.

[1] https://en.wikipedia.org/wiki/Neuroplasticity

lamename2• 2 years ago

There's even an artist that made this point: Cartoon Neuron

https://www.redbubble.com/shop/ap/6229477

https://x.com/Cartoon_Neuron

personjerry• 2 years ago

Reminds me of "When U.S. air force discovered the flaw of averages" [0]

[0] https://www.thestar.com/news/insight/when-u-s-air-force-disc...

richrichie2• 2 years ago

There are even bigger problems. For example, the common “this region lights up more if this is done” type of fMRI studies are suspect because what the fMRI tool does may have no bearing to actual brain function. I read a book by a neuroscientist lamenting the abuses of fMRI in papers a while ago. Unfortunately, unable to locate the reference.

bashtable• 2 years ago

Is the book titled "Brainwashed: The Seductive Appeal of Mindless Neuroscience"? The writers are apparently not neuroscientists.

richrichie2• 2 years ago

No this was written by a neuroscientist.

1659447091• 2 years ago

Somewhat related book on how the concept of average can be misleading and/or detrimental, The End of Average

https://search.worldcat.org/title/The-end-of-average-:-how-w...

ithkuil5• 2 years ago

There are three kinds of lies: lies, damned lies, and fake quotes.

fat_cantor• 2 years ago

Another convenient fiction is that neuronal communication is all spikes, 1's and 0's. In this fiction, neuromodulators are ignored. Glial cells are ignored. The immune system is ignored. The first neurons in the brain that select for different auditory frequencies are ignored - auditory hair cells release a steady stream of vesicles packed with glutamate, and postsynaptic glutamate receptors compute something like a moving average of the glutamate concentration in the synaptic cleft. But it's much, much more complicated than that. Sounds like something is holding back the field, but averaging is a pretty lousy description of what it might be.

Log_out_• 2 years ago

The universe wants you to save energy so every intelligent life form allover runs flat copies of others, of which avg is just the scientific version thereof.

Thus neuro science is bad everywhere in the Universe.

Averaging is a convenient fiction of neuroscience

Related Stories