Animals Are Way Smarter Than You Think
How do birds know how to build nests?
How do spiders know how to build webs?
How do bees know to use hexagons to pack honeycombs tightly?
How does a bower bird stack gifts for prospective mates so elaborately?
Whenever people observe animals doing something magnificent, we mostly call it instinct. These instincts are thought to enable these animals to accomplish tasks that are far beyond their intelligence. A spider must be quite dumb because of its tiny brain. How could it make a complex web with multiple types of stickiness, and how would it know that this will help it get food?
Instincts are thought to enable animals to do things far beyond their intelligence.
I really think that this is mostly wrong. But to explain why, we first have to understand ourselves.
Human intelligence is extremely variable and multi-faceted. But its hallmark is that it is adaptable. The intelligence super-power of humans is that we can co-opt many of the different specialized calculators in our brains in order to accomplish totally new and adaptable thought tasks. In fact, one of the first articles on this blog, “The Simulation In Your Head,” is about exactly that. Presented with a totally novel problem, many animals are stumped. Humans, however, can usually solve problems that are completely different than anything they’ve encountered before. The “general-ness” of our intelligence really is astounding.
But we have taken that generality and defined intelligence around it. Many IQ tests are largely tests of the ability of the test-taker to abstract out or generalize information to a wider or different application. The “smartest” humans are often the ones that can best sense the important patterns in the noisy data stream of a life and use those patterns in new ways to be first or take some advantage.
But one thing that has become extremely clear in the last decade of AI research is that there are other useful types of intelligence. There are AI algorithms now that can play Go or Chess much better than any human that ever lived. There are AI algorithms that can detect the “cat-ness” of a portion of picture better than humans. And there will very soon be AI that looks at video from driving cars and apprehends the safest path forward much better than humans. Yet none of those algorithms or computers could instruct a robot to make a peanut butter sandwich. And because of this, we think of computers as fundamentally dumb. But they’re not. They are absolutely brilliant at some narrow forms of intelligence.
I truly believe the same is true of many, many animals. Let me give you an example.
Squirrels bury nuts to save for later. Many people consider this to be instinctual. But if the squirrel were simply following some genetically-encoded program, then deviating from that program would cause it to go haywire. Yet we see squirrels that are about to bury the acorn, then out of the corner of their eye they will see another rival squirrel watching them. That squirrel being watched will actually pretend to bury nuts while in the view of its rival, and then finally actually bury the acorns once he is out of sight.
If burying nuts were purely some program that genetics encoded into brain structure, then this major behavioral strategic adaptation wouldn’t occur on the fly. The squirrel is smart enough to presume the motives of the other watching squirrel and just pantomime the “instinct.” To me, this makes it very unlikely that burying nuts is an instinct at all. More likely, the squirrel finds itself with excess and realizes the crooks of trees is not a secure location from squirrels and birds that focus their existence in trees. So it goes somewhere where its rivals aren’t in order to hide the acorns. I see no reason to attribute this to instinct at all. The squirrel doesn’t have to “know” winter is coming. She just has extra food now and wants sole access to it later. Is it really so hard to believe that squirrels can plan ahead, presuming future hunger and food access problems?
But because squirrels can’t count to 4 or solve mazes like humans or push colored buttons as well as a chimpanzee, we assume their intelligence is low.
Squirrel intelligence is not as low as we think. It is, however, much narrower than our own.
Squirrels don’t have the huge brains with plastic, re-programmable neocortices that chimpanzees boast and humans have in spades, and this qualitative difference in intelligence makes us underestimate their quantitative intelligence.
Or think about a bird building a nest. This is not a learned ability, as nests building is almost by definition not taught from parent to child birds. But birds find a way to build a nest out of hundreds of sticks with their mouths (you try it!) in order to protect their eggs. Why should we rob birds of the inventiveness and creativity to craft a nest with intelligence? Again, you can see in the modifications of nests that the birds are not simply following some rote program of instinct.
When I was a child, I was fascinated by birds nests. I loved peeking to see the blue and brown speckled mockingbird eggs (a mockingbird once dive-bombed my ear). I never disturbed or touched the nests while the birds were awaiting the eggs or tending the eggs or their young. But well after the nests were abandoned, I did take them down and look at them. I was always astounded to find how soft these homes of sticks were on the top and inside, where the eggs had lain months earlier. There were downy feathers, or strings from palms, or even cotton strings taken from near clotheslines. Soft plastics were sometimes used, or even synthetic polyester tufts. The birds knew they wanted a soft bed and they scrounged for whatever would make that possible. This was a smart adaptation, not automated instinct.
But nests and spider webs and complex mating dances and whale songs bring up an even more important question that flies in the face of the instinct paradigm:
How can a string of DNA consistently program the grown layout of brain cells…such that a complex behavior which interacts and adapts to the animals’ environment can consistently be programmed as instinct?
There are lots of things about life that are truly astounding, especially as one studies microbiology; but this beggars belief. I simply do not think that nest-building and nut-hiding and spiderweb-crafting CAN be programmed into linear strings of DNA.
Typical animal genomes are a few billion letters long, but this type of programming, to create specified brain structures at just above the neuron level to produce complex, adaptable behavior…I am sorry, but the concept of genetically-passed-down complex instinctual behavior in an emperor wearing no clothes.
That said, of course there are some much simpler instincts that are feasibly encoded in animal genetics. Dogs, apes, and humans seem to share a reflexive jerk-back fright response to visual fields that contain a thin, smoothly-curving line. This anti-snake reflex is hilarious to try out with your dog and a hose on the ground. But this is a fairly simple structural encoding. It creates neurons looking for a type of visual pattern, and then sends an alert signal. That is all. It does not cause my dog to then drag a stick around the yard in a grid pattern to drive away all snakes. It is clearly a programmed response. Even human infants have different brain responses to frogs and snakes, described in this paper, which contains the figure below:
I do think that the instinct exists for birds to desire to protect their offspring, including their eggs. There is very likely a hormonal trigger that causes attachment to the eggs or perhaps even attachment to the perceived closely-future eggs about to come out of the female.
I also think that salmon are genetically programmed to seek out the smell of their birth stream when it’s time to spawn. But again, this compulsion is not complex. It is likely a yearning for something remembered, and so the fish is drawn to that familiar early scent near the end of its life.
These programmed instincts are incredibly impressive, but the mechanisms involved are fairly straightforward.
Simple instincts may indeed be genetically programmed, but more complex behaviors that include adaptive, complex actions…those are examples of animals being smart.
Animals’ brains deserve that credit.
But then why do mockingbird nests look so different from robin’s nests? Why do golden orb weaver spiderwebs all look pretty similar, but black widow webs all look very different from golden orb weavers, while similar to other black widow webs? All trapdoor spiders make no web at all. If complex behavior is not instinctual, then why can we tell the difference between bowerbird subspecies just by looking at their bowers?
Dr. Mark Blumberg of the University of Iowa provides an answer. He notes in a wonderful paper that DNA is not the only thing inherited from parent to child in a species.
“A developmental systems perspective, however, encourages a broader definition of inheritance to include all the biological and environmental factors that influence individual development, especially those that are reliably transmitted. By this view, DNA is certainly part of our inheritance, but so are all the species-specific cytoplasmic factors in the egg that are passed from mother to daughter. And so are the numerous environmental factors in which every biological system develops, including (but not limited to) temperature, oxygen, carbon dioxide, atmospheric pressure, and gravity.”
All the animals of a single species have a very similar toolkit to work with. For instance, golden orb weaver spiders have the ability to make extremely strong webbing (including a gorgeous thick silk they put in the center to rest upon). This allows them to make huge, 7-foot-wide spoked webs to catch passing flying insects. Black widow webbing is relatively weak, and so they make small volumetric webs that are meant to ensnare smaller insects passing by an enclosed corner or small interior space. The brains of the two spiders are likely quite different, and perhaps golden orb weavers have much better vision (for sending anchoring web spokes to far away branches via the wind). But even if it were programmed to do so, a black widow could not make a golden orb weaver web. These spiders and their children also have the same geographical area, local insect types, temperatures, and plant types. Blumberg even notes (correctly) that the children inherit the same gravitational field the parents used.
So if most of the animals of a species have all this in common and additionally have very similar bodies to each other, then it is quite reasonable that they will calculate very similar solutions, such as the same types of web designs or bird nests. The wild success of both species indicates that they make excellent use of their toolkits. A human raised in Arizona will run on two legs just like a human raised in Zimbabwe precisely because they both live in very similar gravity fields and have the same joints and limb lengths. Neither will prefer running backwards or on all fours.
Similarly, look at how humpback whales use long, sonorous songs, clicks, bubble blurbs, slaps, and leaping bellyflops to convey information to each other. They are using everything at their disposal to communicate with each other in low-visibility oceans. This isn’t instinct. It’s creatively and intelligently using what they have. And while blue whales cannot generalize enough to create Morse Code types of communication to describe krill density in kilograms per cubic meter, they do use what they have for hauntingly beautiful entreaties and warnings and yes, alerts of good hunting grounds.
Another key insight from Blumberg’s work is the case of baby rats. These surprisingly cute animals have a righting behavior to orient themselves feet down when starting upside down that is evident even just after birth. If you release a baby rat (a “rat pup”) upside down into water, it will right itself as it sinks and land at the bottom on all four paws, just like a dropped cat:
But someone suspected that this behavior is actually learned, not solely instinctual. So of course they invoked the Space Shuttle.
Pregnant rats were taken up on a Space Shuttle mission (STS-66) so that their fetuses would develop in zero-g. Then the rats gave birth after the mission landed. The rat pups were then subjected to the same drop test in water. There was a big difference in capability. The space rat pups did not right themselves when they fell through the water:
You can actually watch the summary video here (link downloads a .mov file).
It’s not that the space baby rats couldn’t have righted itself. It’s that they developed in an environment that didn’t have an up or down, and so they did not learn a preference.
This is a beautiful (and unbelievably cool and expensive) example of how even some very basic “instincts” are learned, and learned by very, very young animals.
In case you were wondering, after a few weeks back on earth the space rat pups gained the behavior to right themselves when released upside down in water.
So, I don’t think most complex instincts exists. I think animals are just extremely smart in their own narrow but still astounding ways.
Let me leave you with my favorite example.
Portia spiders are spiders that hunt web building spiders. But they are extremely small. Portia spiders are usually just ¼ of an inch across. But what they do is quite remarkable.
Upon spotting a target, the Portia will study its victim and the surrounding environment, looking for a path that will get Portia above and preferably behind the other spider so that Portia can descend on a web rope down and upon that spider, grabbing and immobilizing it.
But in order to do this, Portia must do the following:
Figure out the 3D environment surrounding its victim.
Figure out what paths will get it above the other spider.
Figure out what the other spider can see(!)
Figure out which paths will get it there without being seen.
In order to do this, Portia must hold a model of the other spider’s abilities and viewpoints in its head. This David-Attenborough-narrated video of Portia doing precisely this may be the best 4 minutes of your day (and the end will really impress you):
In order to plan and carry all this out, Portia will stare at the scene for 1 or 2 hours straight, and then move. Why so long? Because Portia has a brain about a millimeter wide, with only about 600,000 neurons in total.1
This behavior is clearly not instinctual. It involves calculation. It stretches the narrow intelligence of this spider to an almost literal breaking point: the areas inside young Portia spiders are literally up to 75% brain, with the brain actually extending into its legs at some points of development. The spider needs as much brain as it can cram into itself because the spider is using its brain to calculate — to think!
If such a small brain can accomplish complex hunting behavior, which requires extensive use of narrow intelligence, then surely we can grant large animals with millions of times more brain cells the ability to think…just maybe not exactly like us.
One of the reasons that we want to find alien life is that we will learn all new ways of thinking. But we have really smart life forms all around us. We just have to learn to appreciate their narrow ways.
We can start by not attributing all of their genius to instinct.
Proportionally compared to other spiders, Portia has a huge brain (it is only 6mm wide, including legs).