I have just read the best book of my life, as of yet at least. “Being You", by Anil Seth. No, not the waffling found in self-help books, or by therapists or “gurus”, about “finding yourself” and all that, but the scientific basis of why we even feel to be ourselves, or our selves.
I couldn’t put down the book, simply had to read on, and now I am going through it a second time. Better than any fictional novel. The first read was a breathless sprint, to get the big picture, now in the second read I get the details.
I am currently going through a difficult time. Some person, let’s call her MF, appears to have a completely different view on my experience with her. I feel deeply hurt by her behaviour, and have explained that accordingly, but MF simply says “I didn’t do that”, and that’s the end of it — any further dialogue is being refused, leaving me dangling without any possibility to reconcile. Now, this strict refusal appears to be basic indifference and carelessness, combined with limitations of the intellectual capability to even conceive the possibility to experience the same “reality” differently, but this difference in experience is striking. Because I don’t think MF just makes excuses — also, but not only —, she did (does) have a different experience.
Here’s where Seth’s book becomes relevant.
I have written a few times about our prediction-based perceptions, and I don’t need to substantially correct any of that after reading the book. Lots to add, though.
Seth beautifully develops his concepts and scientific findings,1 all integrated in a coherent explanation of what consciousness is, why we have our daily and continuous phenomenological experiences of the outside world, and of the inside in ourselves. And why they are the way they are.
Seth calls our conscious experience a controlled hallucination. A hallucination is the perception of something that does not exist in the “real world”. Is a creation in our mind. A controlled hallucination is the prediction of our mind of what is there, corrected by the sensory signals emanating from what we try to perceive. We experience our predictions, continuously corrected by the signals from our senses. As Kant, von Helmholtz, Hume, and others have said a long time ago, we cannot experience the world as-is, only our interpretation of it – that is, our predictions, based on generative models. We don’t perceive the world as it is, we perceive it as it is useful to us to do so. Hence, a continuous, controlled hallucination. As described here, this prediction-based perception is an energy-efficient and fast way to get a grip on our environment and our bodily functions. Our brains would not even have the processing power to analyse all our incoming sensory signals anew from one moment to the other.
The being you part comes into full play when considering the signals from within our bodies: it’s a controlled hallucination as well. Our brains have basically evolved to keep us alive, that is, to keep our physiological parameters within vital bounds. But our brains are not fully hard-wired to bodily sensory signals. For example, blind persons will use the brain region usually used for visual processing for other purposes. After we’re born, our brains learn to figure out what’s what inside our bodies,2 just like with the external world, and thusly, over time, develop, expand, and correct the generative models for internal sensing and regulation: best guesses, in the Bayesian sense – but tuned for control.
The perception of the outside world is tuned to recognise objects, hence the errors in prediction, as indicated by the incoming signals, will be used to correct the prediction to jibe with reality. For the inside, prediction errors are minimised differently. The prediction is what we need to survive — think, for example, about the correct glucose level in the blood —, and actions are used to make the sensory signals match the prediction, thusly making the prediction error go away. One possible action for a too high blood glucose level would be to release insulin into the blood stream. The controlled hallucination is extended with a controlling hallucination. Perception and control combined in an elaborate Bayesian network of sensing and actions.
It’s this comprehensive, continuous Bayesian best guessing about our bodies as well as the the world around us that makes us feel to be ourselves, both regarding sensory inputs as well as actions. At any time we have exactly one integrated experience of all that, consciously or not. This way, external experiences get connected with our emotional state. A similar external experience can then recall the emotions, or an emotion can recall the memory of an external experience.
Let me stop here. I can only scratch the surface anyway. Read the book. It’s only some 360 pages. It will open your mind about your experiences. Sometimes blow your mind, until you think it through. As said, not a self-help book, but understanding these fundamentals, and thinking them through for what they mean on a practical level, nonetheless give insights into, and maybe even advice for, daily life.3 It just needs some observation and reflection. For me as relentless observer of myself, other people, and life in general, it’s a goldmine.
The Context Revisited
So MF definitely experiences a different controlled hallucination than I do, and thus a different reality, even with joint experiences together. Refusing any reconciliation is indifferent and heartless, and adds another layer of pains, but then again, also this belongs to the overall perception and thus experience in consciousness, which includes emotions, which in turn are based on our minds best guess about bodily sensations. We two clearly have different generative models that shape our perceptions and experience.
I am not the most emotional person, and I know others who are not either. But in my experience we can compensate that by being (self-) aware of this, and then using intelligence and imagination to put ourselves into the shoes of the other, and make conscious efforts to try to come to a common understanding nonetheless. MF evidently is not able to do that, alas. I have never experienced a person with her limitations. In my life, dialogues have always been helpful for both sides, and were never refused. And why should they? Dialogues are the only tool we humans have to resolve conflicts and differences.4 And from a humane point of view, should we not seek reconciliation, or at least understanding? MF’s behaviour is baffling and saddening, sometimes infuriating, an inconceivable out-of-bounds experience for me. But also – or maybe even especially – interesting to observe and study in the light of all the above.
As many of us, trying to put myself into the shoes of others is something I do all the time.5 It helps me to understand them, as far as I can even assume their positions in their lives. Which is limited, obviously. I guess this means to apply other generative models than my own personal “default” ones to create a shade of another controlled hallucination than my own.6 I try to do this for MF as well, but I have to admit that I am still meeting my limits to conjure up the corresponding troubled mind to explain and understand her behaviour. This haunts me. As said, an out-of-bounds experience. I dislike if I cannot make sense of a situation. And even more so if reason and reasoning cannot be used to achieve more clarity. I guess I really need some updated and expanded generative models for this edge case. And I think they do emerge, as I start to experience certain situations differently today than three months ago. It appears Seth’s theory works for me.
Let’s conclude on a positive note: see the controlled hallucination at work. Behold that painting of Camille Pissaro. Go full-screen. Make sure you keep some distance, depending on your screen size. On a phone, keep it at arm’s length. On a 27” monitor, take a step or two back.
Take your time, and let the scenery come to life. Pay attention to the buildings all the way down into the background, the chimneys and structures on the roofs, the horse carriages, the people. I see a vivid scenery, very clear, detailed. The picture does comes to life.
Now zoom in, or check out the crop below, and see what Pissaro actually has put on the canvas! Amazing. Maybe baffling, until you think about it.
Like other impressionist artists,7 Pissaro wants to make the beholder add his or her own share to “finish” the painting. He probably wasn’t aware of controlled hallucinations, Bayesian inference, and all that. But he clearly understood that we are not able to perceive the world as it is, but that our minds create what we perceive, and he masterfully reduced the sensory signals to an essential and meaningful minimum to make the scenery come to life in our minds. It’s like he reverse-engineered the generative process of perception.
The picture viewed from a distance clearly demonstrates that we are not experiencing the world as being “really out there”, that is, it’s not a passive revealing of an objective reality, simply based on the oncoming sensory signals, but a vivid and present projection — a reaching out to the world from our brain.8
You and I probably don’t even see the same picture, as we have different generative models, based on our genetics, life experiences, imagination, and whatnot.
Go back to the full view from a distance, and you’ll realise that it has become somewhat harder to see the initial impression, now that you are aware of the painting technique.
Pay attention to the details of the buildings from afar, and you can “see through” the technique applied. Your mind now makes a different prediction, as your generative model has been updated. Also, paying attention increases the precision9 of the values in the Bayesian inference process, so the posteriors — what you actually perceive — change.
As a real scientist, Seth adheres to the principles of explain, predict, and control. That is, a theory must explain, not just describe, based thereupon have the power to make testable predictions, and then allow to control the parameters to run specific tests on the predictions. His theory is based on his original work, but of course also of many others. The list of references is long. Also, again as real scientist, he points out the boundaries, as well as yet to explore and test many of the aspects. And there are many aspects, and facets, to his theory of consciousness, as one would expect. Comparable to what we know today what life entails. ↩︎
Called interoception, the signals from within are purely for control purposes, as the brain only is concerned with internal regulation, and couldn’t care less about objects in the body, for example where the liver is, and what shape it has. Proprioception is the perception of the position of our body parts in space. As with the outside world, objects matter for proprioception. ↩︎
I am sure you have listened to the toy voice that you can either hear saying “green needle” or “brainstorm”. I can easily hear both, depending on what I want and expect to hear. Every time, completely reliably. (I have asked others, and some cannot.) So can I not also shape my daily experience by my intentions and expectations? I have experimented with that, and it works, to a degree. I am aware that here I talk about a conscious, psychological level, not the unconscious and relentless Bayesian inference in the milliseconds range of the controlled hallucination. But Seth also describes experiments for how conscious expectations influence unconscious perception. He also explains that expectations and predictions are hierarchical, so it seems my conscious expectations can percolate down and shape the unconscious predictions of the controlled and controlling hallucination. ↩︎
Apart from raw power and violence, which I don’t consider to ever result in actual resolution and reconciliation. Hence, for example, military service is not conducive to my well-being, as hierachical raw power supersedes reason. Another example would be the power of the police force in a dictatorial system. Or a prison. Or a big corporation. Or Gilead. ↩︎
Or I think through scenarios where I mentally remove a person, or replace him or her with another, in order to clarify and delineate dependencies and influences, and their potential impact on my perception of specific situations. ↩︎
Yes, I am aware that I am again bring together conscious predictions and expectations with the unconscious, automatic ones of the controlled hallucination. See the above footnote. ↩︎
Such as Cézanne and Monet. But also think of the earlier non finito and sfumato techniques of Michelangelo and da Vinci, respectively. ↩︎
I am paraphrasing Seth. ↩︎
Decreases the the variance of the probability distribution of the values. ↩︎