top of page

How do we think?

Updated: Aug 2


image of people nearby in a crowded train
Crowding on the Central Line in London: how do we think about this experience?

An interesting question. Some people would answer along the lines of logic and rationality, while others would look at the way in which the brain's own neural network reconfigures itself, and yet others might answer in terms of imagination and creativity. But each of these is only a part of the answer. The more experiments we carry out at PEARL, the more all-encompassing the answers to this question become. We are able to show the brain creating a model of the world and how it deals with the breakdown of this model as we change the world. This in turn drives how we perceive and therefore how we act. We are beginning to see that behaviour is starting from that preconscious modelling of the world, based on lived experience and incoming multisensorial data and how the brain deals with the blending of these two clouds of information. We have to remember that the brain does not sense anything - it neither sees nor hears: it just lives in a black box and responds to electrical signals generated by a basic electrochemical shift. But what causes that 'shift', and how does this apply to how we feel about being in a crowded train or how we enjoy art ?


Karl Friston has developed a model of Active Inference, which works on the basis of what he calls “predictive coding”. This supposes that people do not try to understand what their senses are telling them about the environment, but instead that they predict how the environment will be, based on the combination of their lived experience and the incoming environmental data. The brain then responds to this prediction, for example, by initiating the release of a hormone to prepare them for a reaction. The body then reacts accordingly. When the reality matches the prediction we are prepared for the appropriate reaction. This forward-planning process keeps us alive as we are prepared for the next moment in the future. This inference creation part of the process is called Active Inference.


Friston sees Active Inference as a characteristic of living organisms - they minimise surprise through this predictive process. This is what living organisms must do to face their challenges of fundamental existence, and why they do it (to minimise the surprise of their sensory observations). Surprise is crucially important to survival. The surprising sound of a snapping twig in the quietness of the Savanna could mean the presence of something to eat or something that could eat you. Making the right call in that situation is a matter of life and death. Communicating surprise is therefore vital. In the language of Shannon’s Theory of Communication (which is often called Information Theory), the ‘amount’ of information contained in surprise is high. The background quietness of the savanna continuing as it does conveys no information at all. The snapping twig, however, carries a lot of information, so communication systems want to maximise surprise in order to be more informative, and this is the basis of all communications systems (and why Shannon’s theory was so important). Shannon likened this ‘surprise maximisation’ to entropy maximisation as encapsulated in the Second Law of Thermodynamics so that he could calculate how much information a communications channel (he was working for a telephone company at the time) needed to be able to carry. The communications systems in living organisms want to aid survival and therefore want to reduce, rather than increase, surprise. Minimising surprise therefore equates to entropy minimisation. How living organisms achieve this is by predicting what the world will be like and then optimising the probabilistic representations of their sensory inputs against how the world turns out to be. In other words they seek to make sure that the world really does match their prediction as they have predicted it on the basis of their sensory inputs.


Take the example of a fish. The world in which a fish lives is water. While it is alive, a fish seeks to remain as a fish so it seeks a water environment by trying to make sure that its sensed perception of the world is that it is wet. In Friston's terms, by seeking water, a fish is seeking to minimise entropy to remain being a fish. It infers that it is in a water environment and actively seeks to increase the probability that its sense of the world is that it is wet and that the probability of the world actually being wet is increased. That is Active Inference. In this way it minimises the surprise of finding itself in a dry world, such as air. If its senses indicate that the reality is that it is in air, the fish will struggle to seek water again in order to survive. This combination of entropy minimisation and probability maximisation is distinctive to living organisms. Once the fish dies it reverts to the Second Law of Thermodynamics and eventually maximises entropy as it turns to dust. Human beings, as long as we are living beings, also minimise entropy: we seek to maintain the living form, so we create a model of the environment to which we react. This is where Friston’s model points: by using concepts of entropy minimisation and updating probabilities we can estimate the world as predicted by our brain in order to respond and react with the best chance of surviving in the next moment.


Does this all sound improbable?


In PEARL we found that people inside a simulated train carriage respond physically on hearing the sound of a train decelerating, for example by leaning forward to resist the inertial forces of the deceleration even though the train was not moving at all. We have also observed shifts in neurological and physiological activity in response to the same stimulus, and that this response is fundamentally different if the train is crowded. But in neither of these cases is the train actually moving. So why the reaction? 


Well, this is Active Inference in action. In the first case, the brain is responding to the stimulus of the combination of lived experience and incoming multisensorial data and preparing the body to resist the inertial forces in order to protect itself before the forces come into play. In the train crowding scenario, the cognitive effort shifts from a small increase in the prefrontal cortex in the right hemisphere in the uncrowded situation, to a large increase in the prefrontal cortex in the left hemisphere in the crowded situation - and yet the train is not moving.


Active Inference is the result of a combination of this entropy-minimising niche-constructing creation of a model of the world and a Bayes-type predictive coding and updating by the brain to create new perceptions. Friston uses this model to create generative models to predict behaviour for machine learning and artificial intelligence models. At PEARL we observe that behaviour in action, and we can do so knowing precisely what the environmental stimuli are in each case, so we use it to inform the way we design the environment.


Of course, every individual has a different lived experience, and every individual's sensory systems will be slightly different, so every individual will have different models of the world even when they are experiencing the same environmental stimuli. Which brings us to the arts.


Art uses the Active Inference process to create worlds to which we respond. Performing arts like music or dance do this through sound and movement. They induce in each performer and in each audience member the inferred next moment, which is then 'tested' by the reality of the performance in the next moment. When that test shows that the inference was correct, we are happy, but perhaps a little bored. When the test presents a surprise, we have some form of rethink, which could turn into a protective sense of shock or an emotional sense of pleasure.

The musicians and dancers join with each individual audience member to create the performance, which is always about the next moment. That is what makes the arts so fantastic.


Every performance is original for everyone. This is how composers like Beethoven were so masterful: he challenged the lived experiences of performers and audiences by producing challenging sensorial events to which the brain responds. Whether this was in complex harmony, rhythm, combinations of sounds or loudness, he challenged the Active Inferences being made by all the participants. He developed the world from the classicism of composers like Haydn and prepared the ground for composers like Stravinsky, just as Galileo built on the work of Copernicus to stretch the learning of the current orthodoxy. And Beethoven created beautiful music with no idea of how his music worked in people, but he had a great idea of what he wanted to express in his own terms and how to do that. In some ways Beethoven was a kind of Galileo for music.


The same applies to other arts of course - drama, poetry, sculpture, painting are all about invoking the active inference process in each individual so that each can create the new world in order to respond to it.


By feeling the world through art we can figure out the science. I do feel that trying to understand the world through science without the perspectives of art leads to a sterile and unhelpful, possibly even unhealthy, existence. And actually, how much of the world do you understand if you view it in that way? The beauty of art is [how it plays with the brain so that] we enjoy it. Réjouissance follows Renaissance – or is it the reverse? The ‘rebirth’ in Europe was a coming-together of science and art resulting in a phenomenal flourishing of arts and sciences. The shift into the Enlightenment was an attempt to narrow this, understand it, and, arguably, to control it so that the opportunities were constrained to what was logical, rational and useful. What was lost in that transition was the sense of creativity, enjoyment, reasoning (different from rationality), and openness to the new. The attempt in Europe to rationalise knowledge and wisdom that started in the seventeenth century closed off much of human thought and enclosed it in a western European context. This was narrowed further during the 19th and 20th centuries as first engineering and then science succumbed to the desire for creating ever narrower fields of knowledge. The 21st century is the time when we need to open minds again.


This is why it is important to fuse the arts and sciences and destroy the artificial barriers between them. The sense of understanding and not understanding at the same time, and the realisation that time is both an instant and an eternity, neither of which can be measured, which is so redolent of the arts, is also applicable to sciences. We, as human beings, need to understand both and PEARL stands as the laboratory that is exploring this fuzzy frontier.


162 views0 comments

Comments


bottom of page