Speculations on the evolution of thought

This article looks at the nature of human thought through the lenses of computer science and evolutionary biology and contemplates why it is important we do not lose sight of our own biology as technology leap forwards.

Let’s start by thinking about what our brains did way back in evolutionary history, something we can get a good idea for by looking at existing organisms with minimally evolved nervous systems. In these simple organisms such as c. elegans (a worm) the function of the nerves is to help the organism react to basic changes in its environment such as temperature and pressure, with simple results such as moving towards or away from the stimulus.

As organisms have evolved in complexity, and evolved more sophisticated sensory systems, the complexity of the reactions to those systems has also evolved. As the number of sensory systems and their complexity increases it becomes advantageous for the organism to be able to integrate all of these inputs, resulting in the slow emergence of consciousness.

Lets now think about consciousness, and how that ranges in complexity between organisms. In simpler conscious organisms (say something like a goldfish), the gamut of reactions displayed is fairly simple and can be amply described by feelings of attraction and repulsion. As we look at more complex organisms the number of emotions we need to describe their behaviour increases. For example, there is certainly more to the behaviour of dogs emotionally than there is for goldfish.

Now to bring in the lens of computer science. We can think of these emotional reactions as heuristics (algorithms for making approximations) that have developed to help the organism react in an appropriate manner to ensure its survival. As organisms have evolved the heuristics they use for reacting to environmental stimulus have increased in complexity. In the most complex organism (including ourselves as humans), rational thought has also emerged as a heuristic.

Wait a minute I hear some of you say, rational thought is not a heuristic, it is step by step logic! But is that really true? It is true we can think logically, but it is only in very limited circumstances (when proving mathematical theorem’s, or debugging computer programs for example) that we are thinking purely logically, despite what economists may tell you about rational markets. For the most part we are taking plenty of mental shortcuts, we have to otherwise we would never work things out in time to act on them. This can be easily seen in the long list of cognitive biases we are susceptible to that you can see on psychology sites (or in textbooks).   

I will assert that the evolved purpose of the brain is to ensure that the organism reacts to environmental stimulus in a way most likely to ensure its own survival (which isn’t to say it always gets it right). I will side step for now the question of free will, and if we are in control, I would like to speculate about that in another essay. For now lets think about how that might happen.

My speculation is that the way we think is a nested set of heuristics, each heuristic finds a better approximation of the “best reaction” based on the output of the heuristic before it. At the start of the chain is our senses which feed into our most basic emotional reactions such as Fear/Aggression (the fight of flight response). Above this rational thought gets a look in, predicting a reason for these emotions and producing higher level emotions such as embarrassment or disgust which we then reason about further to produce a final response.   

So, if this is true what are the implications? Applying the computer science lens again, our emotions, as the earliest developed heuristics are also the simplest, they make broad predictions that then frame the context in which future rationalisations are made. Effectively they are taking the role of narrowing the search space of possible reactions for later heuristics (such as reason) to search within. This works well as long as the solution space does not change significantly with time. For most of human evolution this has been true.

However if the solution space becomes markedly more complex, with many more possible solutions occupying local maxima in the space of best solutions, things can break down. It is possible that a simple heuristic evolved for a simpler solution space will pick an area of the solution space far from the global optimum, and that subsequent heuristics, searching in this local area, will find a much worse solution than they did in the past.

Translating this back to less abstract terms, over the past few generations as a species we have started modifying our own environment at an ever-increasing rate. As a consequence, the processes on which our thoughts are based are operating in increasingly foreign environments to that which they evolved as useful. As we move forward it becomes increasingly important that we take care when modifying our environment that we make things easier for ourselves, and not harder.

Looking at this practically, we need to ask questions like; are AI or algorithms built to maximise the profits of a corporation by taking advantage of our cognitive biases in our best interests as a species? Should we be weaponizing our environment against ourselves for individual profit or should we be trying to work together to create an environment where we make it easier, and not harder, for our evolved heuristics to make the right decisions?

Our emotions have evolved over millions of years, they are never going to change on the same timescales as our machinery. We need to make sure our machinery accounts for that. 

If you have an opinion on this article, please let me know in the comments.