Now, here is the mystery for me. Whereas my computational brain model is able to explain to me literally everything I wanted to know about both the human and animal brains, including how emotions and cognition work, there is one exception – animal dreaming.
Early on I was excited to have discovered a way to make my computational brain either act as an ‘animal brain’ or a ‘human brain’. It happened when I extended my early bare-bones Xzistor brain model to also contain an inductive inference algorithm – and suddenly it very much resembled the human brain resulting in humanlike behaviors in robots: not really faster to learn, but faster to guess what to do in new environments by using past experience and ‘generalising’ knowledge across domains.
This was amazing – as I could see the robot ‘think’ (infer) and then ‘try’ things from past learning to solve problems in novel domains. It also made my task easier of teaching the robot to navigate through its learning confine to a reward source. For the animal brain instantiation (with the inductive inference switched off), I really had to teach the robot every little step of the way. It still got there in the end – but the bot never showed any initiative, no intelligent guessing…it knew how to navigate to the reward source, or it did not.
This ability to start up a virtual agent or robot in either an animal brain mode or human brain mode seemed to me to be a really cool thing, but I never thought it would become important in my brain modeling discussions with others in the field. Just understanding the human brain was enough of a challenge for most of them!
Then I serendipitously met Dr. Karolina Westlund (PhD) online. Karolina has been an associate professor of ethology (looking at animal behaviour from the evolutionary perspective), but with wider interests in scientific disciplines like affective neuroscience and applied behaviour analysis. She is an animal welfare consultant now with a passion for anything and everything to do with captive animal behaviour and improving the lives of animals. She also helps others understand animal emotions (https://illis.se/en/).
And it was the modelling of emotions that led to our paths crossing.
I was writing a critique of the “Theory of Constructed Emotions“ by Lisa Feldman Barrett, when I became curious to know if anyone else had critically reviewed this theory of how emotions work in the brain.
And there it was, written by none other than Karolina Westlund: “My problems with the Constructed Theory of Emotions”.
I was so impressed by the problems she had identified around this theory and the way she had expertly addressed these issues, that I provided links in my own critique to her work. After having published my own Critical Review of the Theory of Constructed Emotion, I noticed that Mark Solms had also pointed others in the direction of Karolina’s blog post and was in the process of writing a critique of the Theory of Constructed Emotion himself. I will come back later to Mark as he is an internationally renowned fundi on dreams!
Reading about Karolina’s work around animal emotions and animal welfare on her website, opened up a whole new world for me. As I read the comments people had posted on her critique of the Theory of Constructed Emotion, I realised there were MANY people who were truly interested in what goes on inside the heads of animals!
So, I thought this is the chance to talk to those who might also be interested in how we can model animal emotions – and what a computational model of the brain can teach us about the unknown world of ‘animal psychology’.
As if to read my mind, Karolina reached out to me after having looked at my Critical Review of the Theory of Constructed Emotion, and in our discussions, she wanted to know one key thing from me:
When it comes to the animal brain, what is your ‘burning topic’.
And yes – it of course was ‘animal dreaming’…
So, I promised Karolina I will write a blog post to explain why animal dreaming is such a perplexing thing to me.
I will now divert to my computational brain model, but don’t worry, I will stay far away from the mathematics and keep things really simple. So do not become filled with abhorrence if I tell you the brain model is called the “Xzistor Mathematical Model of Mind”!
Here is my dilemma around animal dreaming – straight from the model:
1) Animals do not experience ‘mind wandering’ like humans. See abstract below:
“Here, psychologist Herbert Terrace of Columbia University teaches sign language to a chimpanzee, but no researcher suggests that these animals can communicate about anything but the present. Consequently, perhaps only humans experience mind wandering…” from Science
2) Without a functional mechanism of mind wandering, animals cannot ‘daydream’. In human brain mode, my Xzistor robots will ‘thread’ through past memories (associations) akin to mind wandering and effectively daydream, explained in my short (free) book “Understanding Intelligence”.
3) In my book “Understanding Intelligence” above, I also offer my basic explanation of how cognition (“thinking”) works to create intelligence in the brain by ‘directing’ mind wandering towards contextually helpful past thoughts (associations) when the brain needs to find a solution to a problem (i.e. generalising past learning). So logically, “thinking” cannot take place in the absence of a mind wandering mechanism.
4) My model says that ‘sleep dreaming’ is effectively just a drug-induced version of ‘daydreaming’ where the brain switches off volitional limb movements and shut the eyes (avoiding distractions and unnecessary/unsafe actions). My Xzistor robots go to sleep when I push a button (“S”-key) and then they start performing this threading process with effector motions disabled – recalling associations from their association database based on a protocol based on similarities.
Now here is the dilemma!
If I say an animal cannot perform ‘mind wandering’, I am by implications saying the animal can also not ‘think’ (I am not talking about normal cognition e.g. running to fetch a ball, I mean inductive inference – “reasoning” – to seek for solutions to new problems by generalising past experience). So, the model suggests animals do not think reflectively like humans, they just operate in the moment. (Shuuut! I cannot possibly tell my animal lover friends that animals do not think!!!)
But then, if I say animals cannot mentally ‘thread’ or ‘daydream’, it means they also cannot ‘sleep dream’.
Does this make logical sense though?
I think so. When we watch our pooch lying on the veranda watching the road, we don’t observe emotions flit across its face (or wagging of the tail) while it is recalling happy/sad events from the past. We see the dog very much expressionless, only responding to external stimuli (the pigeon, the post man, the barking dog down the street).
But then one of my Xzistor LAB collaborators from the US, Carlos Alvarez (PhD neuroscientist), alerted me to a study where researchers have measured the neural activity of a rat and compared the sleep activation patterns with the awake activity – and it seems like the same neuronal populations were activated during sleep – repeatedly – exactly the same as when it performed its main activity when awake! I am simplifying here, but here is a link to the MIT News article: “Rats dream about their tasks during slow wave sleep” https://news.mit.edu/2002/dreams.
This very much looks like the sleeping rat was reliving its often-repeated activities in its cage in its sleep state with limb movements switched off.
Dilemma! My computational model has never been wrong in the past!
If animals can dream by means of the neural process of ‘mind wandering’…it means they can retrieve past memories in a contextually sequential (structured) manner. Why would they only do this when sleeping and not when thinking? If they were able to switch this ‘threading’ mechanism on when thinking, it would enable them to perform inductive inference. This would have given them a huge evolutionary advantage – the ability to literally ‘reason’.
Maybe this would have been too much of an advantage! They would have been more like humans in their cognitive abilities.
Wow! A planet full of mammals capable of inductive inference (“reasoning”) – this would have made for a completely different evolutionary trajectory and would have been fascinating to watch, except it could have brought the project on this planet to a quick catastrophic ending – as too many ‘clever’ life forms would have resulted in all types of ‘technologies’ being developed, including…weapons.
Somethig tells me this would have resulted in a cataclysmic conclusion.
We were only saved by the savvy grace of the random mutations that is building all life as we know it – for those who still believe they are random…
PS: Note to self – nobody better to ask about this than Mark Solms, whom I once heard talking about dreams, and it was solid science, not folklore.