Everything We Know About Genetic Traits Is Wrong
There’s a better way to think about traits
It’s common to hear that “XYZ trait must be genetic” because it runs in families, or because identical twins (who inherit the same variants of the 20k genes we all share) exhibit the trait more often than fraternal twins (born to the same family but inheriting slightly different variants of those same 20k human genes).
It’s also common to hear the opposite, that “no human trait can possibly be genetic” because to be genetic it’s assumed that a trait must be “automatic, irresistible, triggered by the environment, occur in every member of the species, be unmodifiable, and not require training.” In other words, deterministic.
I disagree with both of these interpretations, and hopefully I can convince you to think differently about what it means to be a genetic trait.
First of all, “genetic” is technically the wrong word. Genes represent only the 8% of our DNA involved in protein expression and regulation. The other 92% of non-genetic DNA is the interesting part that I describe below. (To prevent confusion, I’ll retain the common usage of the word “genetic” for this type of DNA as well.)
DNA is an instruction book consisting of 3 billion letters — “A” “T” “G” and “C” — arranged into paragraphs, and bundled across 23 pairs of chromosomes (chapters). This four-letter alphabet is similar to our 26-letter English alphabet, and can be used to define anything from protein sequences to stories, rules, patterns… and even genetic traits.
Let’s consider, for example, a specific type of human trait: a phobia. Some people are afraid of small holes, a phobia called Trypophobia. Certainly not everyone has this phobia, and even then, it’s situational, triggered by specific contexts, not active all the time.
Is a phobia genetic, written in our DNA? Yes. I propose that DNA acts as a set of instructions for implementing phobias, and by extension, all human traits, motivations, passions and drives.
Here’s how I propose it works. Each trait has 3 genetic components:
- Context detector: DNA provides the set of instructions to develop (in the brain, using neurons) an ability to detect specific environmental contexts (in the case of Trypophobia, to detect the presence of small holes). To conserve the amount of DNA needed to encode the context (or pattern) detector, the detector bootstraps itself with a minimal capability from DNA, along with an ability to learn or fine-tune its context detection ability with environmental (sensory) data inputs. Every member of the species develops this context detection circuitry.
- Trigger: DNA acts as a set of instructions to develop (in the brain, using neurons, neurotransmitters and hormones) a trigger mechanism. Once the context is detected, the trigger determines whether to act on the context (enact the phobia response) or ignore it. The trigger can be either genetic (e.g., having a genetic SNP variation that sets the trigger threshold) or assigned randomly at birth by spinning a genetically constructed roulette wheel (e.g., “if random.randint(0,10) < 2: trypophobia = True”). Yes, by my theory, DNA contains computer code!
- Response: DNA acts as a set of instructions to develop (in the brain, using neurons) a response function, that acts once the trigger has been pulled. In the case of Trypophobia, the response is an avoidance function, to run away and flee from small holes.
Each human trait (or fear, motivation, drive, passion, etc.) consists of these three genetically defined components: a context detector, trigger, and response function. Each component comes with an optional learning function, also specified in our DNA, to improve and tune its performance. So traits can be simultaneously genetic and learned, because DNA puts in place a learning function for each.
As mentioned above, DNA code can act like a computer program or algorithm. It’s not Python code, obviously, but conceptually all languages are the same. Logic and rules can be encoded in DNA using code, for example: IF-THEN-ELSE conditional logic. Loop FOR x IN y. Assign variables x=3.14, DEFine functions, etc.
Neurons interpret and execute this raw DNA code, either via their connections, or literally by reading code directly from DNA tape (like a Turing machine) and executing it. (This is theoretical, but I believe it to be true.)
As code, DNA can also store variables such as sensory patterns or initial training weights (learned by evolution) for neural networks in the brain to detect contexts in the environment, e.g., weights = [0.13, 0.97,0.36, etc.]. The learning function, defined in DNA as well, can then modify these initial weights as it learns.
I also theorize that DNA code references existing human skills and learned capabilities using previously agreed function names. For example, a phobia can exploit a learned ability to “flee” — e.g., run away, walk away, drive away, etc.— because each learned human skill is tagged with a prior (genetically defined) name, to allow our ancient DNA traits to call them as functions or subroutines. As we learn to walk and run, these skills and abilities are tagged so our ancient DNA code can reference them by name: e.g., self.do(“flee!”)
We’re obviously not born with the ability to walk or run, but our DNA code can safely assume those abilities will be in place later, when needed. Of course the urge to learn to “navigate the world” is itself genetic, and walking and running are the usual skillsets adopted, but someone without legs might learn a different way to “flee”. As long as a learned skill is associated with the well-known tag, the trait can leverage it.
It’s wrong to assume that, to be genetic, traits must be “automatic, irresistible, occur in every member of the species, unmodifiable, and not require training.” DNA-coded traits have context detectors, triggers, and responses with modulated outcomes. These genetic constructs have methods for learnability and tune-ability that can also leverage other learned human skills.
While traits are genetically defined, differences in traits are not necessarily caused by genetic differences. As mentioned, trait differences might be assigned randomly at birth by a “genetically-constructed, trait assignment roulette wheel” function. For this reason, any genome-wide association study (GWAS) of human trait differences may only discover weak correlations between gene variants and traits.
Hopefully, I’ve convinced you that our current way of thinking about genetic traits is wrong. There’s a better way to think about genetic traits.
