Nea Artaki, on the island of Evia, Greece  October 2018

I’ve been having discussion with folks about empathy, and why I’ve chosen the model I have to pin my work on.  Obviously, my understanding has increased as I’ve read more, and thought more about this, so I thought I’d share some thoughts on how things have transpired.

First off, as with most of my work, I strive for a functional/constructivist/evolutionary perspective.  What does that mean?

1.  I read a ton of other people’s work.  Folks have been pondering a lot of these questions for a long time, and it surely helps to understand others’ thoughts.
2. When reading, I look for work that tells me something about the base biological function of connection, and then theorizes from there.  I love Stephen Porges (the Polyvagal Theory) Franz de Waal (The Age of Empathy) Daniel Siegel (the Neurobiology of We) and attempt to make myself aware of other great researchers and thinkers who also tie their work to biology (Carl Rogers, Anthony Damasio, V.S. Ramachandran.)
3. That said, I don’t ‘tow the party line’ about anything.  These people are all awesome, but at some level they also are somewhat unaware that the knowledge they generate is intrinsically tied to the social structure they come from.  They lack that key insight, which (I think) would cause them to publish larger systemic interpretations like I do.  I’d love to get them in a room and have a freewheeling discussion.
4. I take my knowledge of system construction (I’m a design prof in engineering), how a given function has to work, and what might be an evolutionary design path would have to be if I had sensors, and processors in more of a mechatronic perspective, as well as an Minimum Viable Product (MVP) perspective, and think about how these things progress over time.  This is, as far as I can tell, a completely foreign mode for really thinking about this stuff in the empathy researcher community, who are largely tied to empiricism, observations and experiments because of their own academic practice.
5. The short version is that I am constantly asking two questions:  “How would that work,” and “How would I evolve a product that would do that?”  These turn out to be powerful questions that really cut through the bullshit.

So we know that we have the Empathy Pyramid below, and, as sentient (or semi-sentient) beings, which really includes almost all multicellular living creatures, intra-organism communication matters.  Even if it’s only for reproductive purposes, animals have to do some coordination.  I’m not going to pick nits at the bottom of the scale (how sentient ARE sponges?) but if you want to argue at the dinner table, have at it!  I’d actually love to listen!

OK — here’s the pyramid for your reference.

If one were evolving an empathetic system, the first sensor level one has is ON/OFF. As I’ve said earlier, this likely first showed up in a pronounced form in the Silurian Period, which was where bony fishes showed up.  Heck, it may have even happened in the Cambrian (500 million years ago!) with trilobites.  Maybe I’m just being vertebrate-ist!  But we do know that bony fishes swam in schools, and one can start to produce that layer of mirroring empathy with an on/off sensor.  If the fish in front of you can be seen by you, that little sensor goes off, and you head toward it.  Couple that with a couple of food sensors, and you now have a collective organism that can move in concert.

Short version — one now has the beginnings/evolutionary seeds/MVP of mirroring empathy.  Evolution and natural selection can now take off and make this more sophisticated.  The basic action is laid in.

Next up is what I call State Behavior.  If you have a sensor that first can determine ON/OFF, the next natural progression is to fine-tune that sensor’s ability to distinguish between a statistically accurate reading (in hypothesis testing, we call this a ‘correct detection’) vs. one where the sensor goes off, but no little fishy is swimming in front of you.  (In the world of hypothesis testing, we call this a ‘false alarm’.)  There is a balance between these two, where you hone your threshold.  Modern radar systems do this — this is called a Classic Detection Problem, and the curve that characterizes our little fishy’s detector is called an ROC curve.  For those with engineering interest, you can look all this up!  Believe it or not, it all started with a Presbyterian minister back in the 18th Century called Thomas Bayes, though I’d be remiss to not point you to the work of Claude Shannon and Norbert Wiener.

An interesting turn now takes place on the evolutionary path.  Better discrimination hones first those beginning seeds of mirroring behavior.  On/Off gets better and better.  At the same time, in order to determine a better On/Off, we also have to evolve a better signal estimator.  Now emergence starts playing a role.  That On/Off detector leads to a better estimator, and that estimator starts to evolutionarily seek advantage of its own.  Fish that feel compelled to stay in a school have an evolutionary advantage over fish that do not feel compelled.  The level of compulsion also likely leads to tightness in formation, or even looseness, as there is now a dynamic balance between size of the fish, and tightness of the school.  We start to see the advantage to different states- like anger, fear, passivity, and so on.

This was made possible by first solving (with evolution) the beginnings of collective movement.  But as with all acts of emergent sentience, it ends up having purposes beyond the original evolutionary adaptation.  The evolutionary winners find new uses for their new hardware/software combo.  The evolutionary seed is sown for emotions in the state differentiation and estimation problem.

Here is a key takeaway — evolution does not follow a pre-planned route to increased evolution, or sophistication.  Cockroaches have been in their same form for 140 million years!  Yet when you create a composite organism (schools of fish, as well as bands of humans!) where some level of sentience and shared information processing plays a major factor in their survival, we see how our more evolved displays of root-biological empathy (attachment behavior, prosody, development of the vagus nerve) come to the fore.  Think of it this way — it’s hardware that’s evolved, that’s just waiting to be used!

OK.  Now we have the basis for both mirroring behavior (ON/OFF) and emotional empathy (STATE processing.) What happens as we refine state processing?  Well, let’s think about how state processing would have to work.

In order to evaluate and make a decision on a particular state, we would need to evolve a probabilistic detector/estimator combo that would take in data, and then guess at a given state AFTER a certain amount of data is received.  That means State estimation grows out of time-based averaging/dichotomous decision making back in original problem — whether we can see another fishy or not.

Now our State estimator grows in sophistication.  It reads and crunches more and more specific data, now gives several different State recognitions, and pops out a decision.  It may do this more quickly, but over time, quick decisions might not pay off, especially as task complexity increases.

And maybe, as time scales lengthen/increase, averaging for STATE estimation starts yielding NO evolutionary benefit.  Things change around the organism, and so some level of temporal windowing starts playing a role in evolution.  Taking in the data, and matching it to your own experiences, which inherently takes a larger processor and is hugely computationally expensive (you’re matching more details, over different scenes!), and more time, starts yielding some degree of evolutionary advantage.

And it may turn out that the state information also starts confusing your estimation of what’s happening with the individual you’re attempting to coordinate with.  If you’re angry/sad/etc., you’re mixing up the signals that are preventing you from making a good decision, for you specifically, or the collective!  Now we can start seeing for an animal navigating a ton of different environments (remember that humans are spread across the globe!) becoming data-driven makes more and more sense.  Rational empathy starts becoming emergent and an advantage.

AND… finally, we end up where we want to consciously improve our own ability to estimate different changing circumstances, with different changing individuals.  And we realize maybe that WE’RE getting in the way.  So we evolve more profound, differentiated ‘being’ inside the collective.  As well as the ability to watch ourself.  Now we have the seeds of larger self-awareness.  Which leads to more emergent behavior.

What’s the bottom line?  The Empathy Pyramid makes sense.  It generates itself, and takes us far away from the warm Silurian seas our ancestors swam in so long ago.

There’s a couple of points I’d like to leave you with.

1.  None of this is tied explicitly to a triune brain.  Any systems person can tell you that if you have one computer, and you have to hack a system together, you can do it with a number of different processor architectures.
2. That said, one can see the advantages in optimality IN a triune brain structure.  Different parts work together, as well as differentiate function along the lines of instantaneous action, state, and experiential data.

So don’t tell me crows can’t love, or be smart.  Or that octopi can’t coordinate.  But you can also see the evolutionary advantages of having a triune brain as far as accelerating evolution and sophistication.

There’s also a relatively clear moral lesson that comes out of this.  We got better, and more sophisticated and evolved, because we took care of each other, in groups with increasing size and diversity.  We might remember this during this particularly mean political season.  We all get to cross the finish line together.