Sunday SciKu | Machine Dreaming

Photo by Jr Korpa


This week’s sciku is inspired by a new hypothesis about the purpose of dreaming. No one really knows why we do it, but dreaming must have some beneficial function. Though it’s only 2% of our body mass, our brains consume 20% of our energy. If we just turned them off at night, we’d save 50-100 calories, and that’s a large price for any animal scrambling for food as our ancestors have throughout evolutionary history. If it wasn’t worth the resources, we wouldn’t do it. And yet we do.

The new explanation, proposed by a neuroscientist at Tufts University, was inspired by a problem that arises in machine learning. If you give AI a task to learn, like navigating a car from one point to another, it does a much better job of solving the problem specifically, rather than generalizing what it learns to other environments. This “overfitting” has always been a challenge with AI, and is often solved by adding extra noise into the system to prevent the AI from settling onto a precise solution.

It’s only a speculative idea for now, but maybe that’s the purpose of dreaming. We learn about our environment during the day, but then play with what we learned mentally at night, so that our concepts of what we’ve experienced don’t become too rigid to apply to future situations.

I was re-reading Iain Mcgilchrist’s excellent book “The Master and His Emissary” the other day, and I can’t help but think the bicameral nature of the brain might be playing role here, too, as the right hemisphere is the “generalizer,” harvesting information from the left’s narrow and concrete view of the world to make broader associations it can assimilate into a bigger picture that the left hemisphere is never able to see.

This is the stuff of poetry and all art, of course—the associative, inarticulate understandings of the right brain. The stuff of dreams. So this new idea is interesting to me, though whether it can ever be confirmed remains to be seen.

 

strange dream:
satellites skimming the surface
of thought

 

Sunday SciKu | Storing CO2

This week’s sciku is a reminder that getting serious about reducing atmospheric carbon has to focus on nuclear energy. There’s so much we can do, including turning carbon into useful materials like silicon carbide—but they require large amounts of power, and renewables like wind and solar create too much waste at this scale. It’s not a technological or behavioral problem, it’s a physics problem. The energy found in sunlight and wind and waves is just too diffuse.

This week, researchers at the Salk Institute published a promising analysis on the production of silicon carbide from plant matter. SiC is a valuable superconductor useful both for electronics and structural ceramics. We can turn CO2 into materials like this, locking up 14% of what the plants have pulled from the air, but the process requires heating the material to 1,600 C in a furnace—and we still mostly do that by burning carbon. Put a compact fusion reactor like SPARC next to the factory, though, and we’d actually be getting somewhere.

 

all that carbon
to capture carbon
carousel

 

Sunday SciKu | Martian Microbes

Everywhere we look on Earth, there’s life, including deep beneath the surface, where microbes cut off from our biosphere for a billion years are still feeding on the energy released by the radioactive decay in rocks.

This week, researchers at Brown University published evidence that the same energy source also exists in Martian rocks, providing the fuel for potential subterranean life on the Red Planet.

We already know that oxygen and methane levels in the Martian atmosphere fluctuate seasonally in a way that might suggest the presence of microbial life there now. Proving that there is life on Mars would tell us a lot about the abundance of life in the universe and our place within it.

 

our loneliness
buried deep under the dust
on Mars

 

Sunday SciKu | Bat Scat

It’s always difficult to peer into the past. All the proxy records we use to estimate the paleoclimate come with serious flaws and are limited to individual points on a very large planet. Many temperature reconstructions are actually proxies of proxies, for which the error bounds don’t just add but multiply.

Now we can throw bat shit onto the tapestry of tree rings and ice cores and lake sediments that climatologists have to squint at. In research published this week in Biogeosciences, a team from Canada dug through two meters of guano to construct a 4,300-year climate history for the area surrounding a Jamaican cave. Because plants and animals produce different types of sterols (like cholesterol), we can look back through these accumulated layers of bat scat to see whether the colonies were eating more fruits or insects at any given time. Insects thrive in wet conditions, so a bug-based diet is a sign of a wet period, whereas more fruit would mean drought.

Using that assumption, the team was able to find evidence that the Minoan and Medieval warm periods extended into the Caribbean. Whether these warm periods were regional or global has been a contentious question in climatology, and the more data sources we can find the better.

The assumption that the bat diet is only responding to climate conditions seems to me more than a stretch, though. Currently, there are five species of bat living in this cave, according to the paper, and there are many forces beyond climate that could punctuate the equilibrium of that local ecological balance. What if a virus or fungal infection wipes out one of the bat species that prefers insects? What if an earthquake diverts a river and changes the local hydrology? Maybe volcanic ash from a nearby eruption changes the soil pH making a new plant species thrive. Maybe human migration introduces a species of rat competing for a common fruit.

Those are just some examples off the top of my head that were likely to have happened at some point over the course of those 4,300 years, and that would each would look like climactic shifts using this methodology.

The point is that it’s impossible to accept any of these lines of evidence as proof—which is why we need so many of them.

And the point isn’t limited to paleoclimate reconstructions—it applies to everything we’re trying to call historical truth—including current events. Even in science, a very large percentage of published papers are false. And that’s using the scientific method to eliminate every sliver of bias that we can—imagine how often a news story is false, composed by a journalist on a deadline writing for clicks. And if we in the information age can’t determine what is true today, how much truth are we gleaning from the compiled historical accounts of the past?

And like the proxies of proxies, all the fake news compounds on itself. That’s why almost every post on your Facebook timeline is false. Truth is the greatest myth of the modern age—we seem to take it on faith. And it’s not because reality is subjective; it’s because we have so little access to anything objective. Everything we consume or perceive is filtered by politics and perspective and expectation.

So if you care about anything, you have to question everything. Never gather all your data from one source. That is your PSA for the day, and here’s the Sunday sciku:

 

found in guano
what we really know
of history

 

Sunday SciKu | Addition by Subtraction

Cognitive biases are always interesting because understanding them is such an important aspect of critical thinking. We can only see the world through layers of filters, and it’s impossible to understand anything without adjusting for them.

This week, researchers at the University of Virginia published work on a bias I’d never heard of before—I don’t think there’s even a name for it yet, but maybe it will be called “additional bias”? When problem solving, it turns out we’re much more drawn to solutions utilizing addition rather than subtraction.

For example, training wheels have been added to bicycles as a way to teach kids how to balance. It took decades for us to realize that a better solution is just to take the peddles off, and let them practice on a simple “balance bike.”

In the study, participants were given LEGO problems that could be solved either by adding more blocks or removing some of them. Almost invariably, people default to adding more blocks even in scenarios where blocks cost money to buy. It seems as if our minds generate additive solutions more readily than solutions through subtraction—and because we tend to stop at the first working solution we find, that’s what we end up going with.

The consequences of additional bias are wide-ranging in engineering and ecology, but also apply to daily life. Just ask Marie Kondo.

 

priceless dust
on an empty shelf
yard sale