Reality and our Brains. Part 2

This is the second installment of my LinkedIn articles on the relation (and interplay) between our brains and reality.

The first article discussed two main points: first, the famous neuroscientific mantra that “we do not see reality as it is but as we are”, and second, the editorial function of the brain by which it fulfills perceptual gaps to create a fluid story of what we call reality, regardless of the fragmented nature of the data it records through our senses.

This second article, as promised, focuses on the predictive nature of our brains and its impact on perceiving reality. As discussed- in some length- below, the brain does not just passively receive signals from the external world ready to react in the best possible way. No. It actively creates assumptions of what will happen next (meaning, what it will perceive from the external world now), with these assumptions forming the basis of what we see as reality in our everyday lives.

Is it possible that our experienced reality is, in fact, our brains’ best assumption of what is happening and not what is actually out there?

Buckle up. It’s going to get weird.

Which one is it: Brainstorm or Green Needle?

This is one of the viral videos that seem to confuse us about what we are perceiving (like, the color of the dress, Laurel/Yanny and others). What makes it unique though is that what you hear is completely based on your expectations. If you expect to hear Brainstorm then this is what will happen, while if you expect to hear Green Needle, again, this is what will happen.

How is this possible? Mind you, this is not like the rabbit/duck or young lady/old lady drawings from the first article, where artists designed these drawings carefully and in a clever way to portray both. Linguistically, brainstorm and green needle are not as similar. Yet, we can hear the one or the other just by expecting it!

This is because, we do not only see reality as we are… we see reality as we expect it to be! And our lives depend on how accurate our expectations are.

How Bayesed are we?

It is remarkable that our brains’ prediction function (of what is about to happen) can be best explained through the ideas of a 18th century English statistician, philosopher and Presbyterian minister named Thomas Bayes.

Bayesian Inference, as it’s called, takes into account probabilities based on the previous understanding of a situation which are updated when incorporating observed data to produce new subsequent probabilities.

Here are the three main components of Bayesian Inference when it is applied to decision science:?

Priors. The beliefs we hold before an event occurs. These personal/subjective beliefs form the basis for our brains’ predictions. They represent previous knowledge built through experience. This is what our brain expects to happen; its starting probabilities.

Evidence. The new data that we receive as well as the likelihood of these data repeating in the near future.

Posteriors. The updated beliefs that we develop by combining our priors and the actual outcome (new data). Posterior beliefs go on to form possibly new and revised priors for the future.

It was the famed neuroscientist Karl Friston, Neuroscientist and Theoretician at University College London, that popularized the idea of a brain that continuously uses previous knowledge to develop probabilistic scenarios for what is happening, only to update these probabilistic scenarios on-the-go by incorporating new information.

But, evolutionary speaking, why does the brain prioritize internal predictions over external data?

The answer is… surprises.

Minimizing surprises is key both to more efficient power management in our nervous system and to our survival. Surprise, in this context, has a fancy name: prediction error. If the internal model producing the scenario of the highest probability to occur is proven right, and by “right” I mean smaller difference between prediction (internal data) and sensory input (external data), then prediction error is minimized. The opposite means a high prediction error, produced by a very wrong/inaccurate internal model. ?It is obvious that the former leads to safer outcomes of decisions, since we predicted them, than the latter, and thus better chances for survival. The lower the (negative) surprise the better. Positive surprises are discussed further down.

Predictions, a thousand times every second

This is not a theoretical discussion. This is how the brain works on a daily basis. More accurately, on a second-by-second basis. Actually, multiple times per second. Even when we are reaching to grab our cup of coffee our brains predict how the cup will feel when we grab it, for each finger, and how much strength we need to apply (and where, and how exactly) to pick it up appropriately to drink from it. ?Without even us consciously thinking of it!

As the model from Jeff Hawkins (author of the book “A Thousand Brains”) shows below, specific points on the cup (locations) are mapped in the brain, together with specific points in our fingers (sensory) that are expected to match the location points when we grab the cup. The mental model that combines the two (locations vs. sensory) in the best way will be chosen for the action. If successful, then prediction error is minimized. If not, and we drop the cup, then prediction error is maximized.

Hawkins provides the following steps in explaining how predictions work in building up a decision:

Step 1. REFERENCE POINTS. Through perceptual inputs and movement in space the brain recognizes elements of the world and places them as reference points in relation to us.

Step 2. SPATIAL/MENTAL MODELS. Based on that, the brain constructs models of key properties of these elements.

Step 3. PREDICTIVE PROBABILITIES. Then, the brain predicts what would happen to those elements if something acted upon them giving probabilities to different outcomes.

Step 4. DECISION MAKING. The action that MATCHES better internal intentions to external probabilities is chosen.

But how the brain creates reference points? Thousands of neural columns (millimeter-long parts in the neocortex responsible for processing info from a tiny part of our sensory organs) are constantly “voting” independently on what they perceive. The decision with the most "votes" becomes the dominant perception of reality and it's used to develop models for further decisions. This is why Hawkins called it “A Thousand Brains”.

This is how we learn and grow

Was my prediction correct or wrong? Did I receive a reward or a punishment? Did I drink coffee from my cup normally or did I accidentally, and unexpectedly, break it and spilled coffee all over… while having an important online meeting?

Outcomes, in terms of reward versus pain, determine the updating of the starting probabilities. These processes are not cold and mechanistic. They are crucial for the brain which means they have a huge impact on its physiology… and on our emotions.

When predictions are successful the brain continues as usual, learning that these predictions worked (the neuronal connections that produced them are useful) so it should apply them again when needed.

When predictions are unsuccessful then the brain needs to change in order to lower the pain it received from this failure, in the future. The neural connections that produced the failed prediction are not useful and need to be “punished” and re-arranged.

Indeed, Wolfram Schultz (Dialogues Clin Neurosci., 2016) found that we receive baseline dopamine amounts when a prediction is confirmed (“all goes to plan”), while our dopamine neurons show depressed activity when predictions fail (negative prediction error). When we receive more reward than expected (positive prediction error), then our dopamine neurons show more activation. So, our brains feel OK when predictions are confirmed, they feel depressed when predictions fail, and they feel euphoric when we get a stronger positive outcome than expected.

It pays off to predict reality correctly and it’s even better if we underestimate its rewards here and there.

Anxiety, Trauma, Fanaticism: ignoring reality

But what about brains that do not change? What happens and some people seem immune to new data, refusing to update their internal models of reality? Why some people refuse to see reality altogether, sticking to their distorted/outdated/problematic views? Here are three reasons for these unfortunate cases.

Anxiety. In their seminal paper on anxiety, uncertainty and the brain, Grupe and Nitschke (2013) describe the typical symptoms of anxiety and their corresponding brain systems. One of these symptoms is the inability of brains in (mostly chronic) anxiety to accept data that challenge their internal models. These brains prefer their version of reality regardless of new information. External data that go against their view of reality are ignored or actively attacked. This means that prolonged anxiety decreases the ability of the brain to naturally learn, change and improve its outcomes. “Reality” is only there to support and feed anxiety.

Trauma. Psychiatrist Bessel Van Der Kolk, in his best-selling book “The Body Keeps the Score”, dedicates a few chapters to explaining how the brains of people in trauma function. One of the ways that the brain deals with trauma is to freeze when being reminded of the traumatic events. Freezing means not experiencing the pain again (as defense) but it also means not learning and improving. It is typical for traumatized brains to ignore new information and react in a locked-in, repetitive way, being frozen or being hot. In such cases, the brain considers reality as not important, since its top priority is to seal itself from more pain when triggered to believe that the traumatic event is not over yet. “Reality” is only there to be traumatic again. ??

Fanaticism. Michael Shermer, author of the book “The Believing Brain”, in his column in Scientific American in 2008, explained that people believing strongly in ideas and theories will see a different reality than the rest of us. In his own words: “UFOlogists see a face on Mars. Religionists see the Virgin Mary on the side of a building. Paranormalists hear dead people speaking to them through a radio receiver”. This happens because “our brain and senses are prepared to interpret stimuli according to an expected model”. People invested emotionally in specific explanations of reality are ready to exclusively see these explanations present whenever they look around. “Reality” is only there to confirm the strong belief or theory.

Addiction is another condition that does not allow learning, or updating internal models, in any way that does not promote/support the addiction itself. But this is the topic of another article. ?Same goes for psychedelics. ??

Reality: Controlled Hallucination???

Anil Seth, author of the book “Being You: A New Science of Consciousness” and Professor of Cognitive and Computational Neuroscience at the University of Sussex, often talks about the brain generating its perception of reality as controlled hallucination. Why hallucination? Because the brain sees reality more based on its internal prediction models than as external data. And why controlled? Because the brain does eventually use external data to make sure it is not too far off. Indeed, the brain spends more energy to construct its probabilistic scenarios and its internal prediction models than in contrasting them with external sensory data. Prediction is super-important evolutionary speaking for our brains, and this is why more energy is dedicated to it. Without predictions we would just be reactive creatures, at risk of harm whenever something happened that could be avoided (or taken better advantage of) if it was accurately predicted. So, living with an internal world of constant predictions is not far from saying we live with a world of constant controlled (effective, useful, beneficial) hallucinations!

It is a messy world out there. The brain does not just create meaning from this mess to fit its needs, as explained in the first article... it expects meaning from it! Actually, it projects meaning into the messiness of the world in order to minimize surprises and optimize its chances to survive and to grow.

Because these two are the main goals of our brains: survival and growth. And they are both better served with prediction than with simple reaction. So, what we see and hear might be what our brains believe to be the more accurate scenario of what we should be seeing and hearing… and not what is really out there.

Imagine, or better, predict that!