Illusions of the Night

Why do we dream?

‘When Dreams Come True’ by Leonid Afrenov

One of the many things my wife Charanya does is to create board and card games around Indic themes. You have fun playing, but you are also left with a little more appreciation of Indian Art, Hinduism, regional histories, and so on. Early yesterday morning, in a 2-hour creative rush, she built a new Tamil wordplay game for us to play as a family, with cards made out of chart-paper and detailed gameplay rules written down. In the 30-minutes that we played yesterday evening, I ended up learning a few new words (plus, since I am not formally schooled in Tamil, I started to get a better understanding of when to use ‘N’ and when to use ‘n’, for instance).

But what fascinated me most was the creative process itself. This was not an idea that she was developing over a period of time. It came to her in her dream the previous night. She woke up with the game - in all its detail - completely formed in her mind, and she pushed through immediately to create it. This reminded me of a few articles I read recently about a new theory for why we dream.

Why do we dream? Why do we go on these magical and mysterious nocturnal journeys? It has spawned great cinema and art, and almost all of literature is indebted to it; but why does this phenomenon exist in the first place? Science does not have a consensus answer, but there are hypotheses. One very interesting theory was put forward recently by Tufts University researcher Eric Hoel. You can read about it here (and here, here, etc.), but let me provide the gist of it. (Eric Hoel, by the way, is one of the most interesting writers I have recently discovered - his elegant essays manage to bring together the science of consciousness, AI, Art, ethics and pop-culture. Before you go any further, please subscribe to his fantastic Substack here. You can thank me later.)

Hoel observes how, after a dream-filled night of sleep, we often get answers to questions that eluded us the previous day. A breakthrough happens when we dream, clarity seems to have emerged. Exactly like Charanya’s clearly formed idea for the card game. Why does this happen? For this Hoel goes back to a more fundamental question - is there any evolutionary purpose served by dreams, or (as many scientists claim) is dreaming just an artefact of our minds that doesn’t have any other survival benefit? Hoel’s proposal is that dreams do serve an evolutionary purpose. But to understand this one has to take a small detour to Machine Learning and the concept of ‘overfitting’.

In Deep Learning, overfitting happens when a model gets trained too well to the learning data sets it has been provided. Such a model works extremely well in most standard scenarios, but fails when there are curveballs thrown at it - the model doesn’t know how to handle ‘strangeness’ in its worldview. Quoting from the Tufts article:

When such machine learning programs do the same task again and again, they can become “overfit”—able to do that one thing really well, but not to learn lessons and create general knowledge that can be applied to different tasks. To prevent that, programmers often introduce random variables, or noise in the data.

Hoel believes that dreams serve the purpose that these ‘noise’ variables do in machine learning. Because one day follows another in similar patterns, our brains get ‘overfitted’ to these daily rhythms; we perform our regular tasks very well, but might be caught completely unawares when something very unexpected happens. In the wild, being overfitted to regularity could be fatal - you need to prime your brain to respond immediately when even the hint of a man-eating tiger is seen from the corner of your eye; the strange imaginary creatures that jump at you in your dreams might be just the net-practice that your brain needs.

Similarly, creativity requires you to break free from the conditioning of regularity. If your mind is overfitted to your day job, your chores and so on, you will find it harder to be creative. Dreams help shake things up, they exercise your mind by taking it through these strange journeys, sometimes thrilling and often scary, serving as random noise data that push your mind out of its overfitted conditioning. You wake up, then, with a mind more fit to solve hard problems that you were stuck with, more supple - and unshackled - to make that leap of creativity.

As I read Hoel’s hypothesis, I couldn’t help but think about Acharya Gaudapada’s philosophical writings. It struck me that there is a parallel to be drawn between this 21st century scientific theory and Gaudapada’s metaphysical reflections from 1700 years ago. Let me try and explain.

Gaudapada was one of the early proponents of Advaita Vedanta, and perhaps the most important pre-Shankara Advaitin. His most important work is the Mandukya Karika, which is a commentary on the Mandukya Upanishad. It is a lean, simple, classic philosophical gem. The Karika’s subject matter is the nature of ultimate reality itself, and the device he uses is the phenomenon of dreams. I don’t know if there is a more systematic philosophical analysis of dreaming than what we find in the Karika (and in Shankara’s commentary on the Karika).

Gaudapada urges us to examine our dream experiences, and then to switch and look back at the waking world very closely. Is there really a difference between these two states, he asks us. While we are dreaming, everything we see and hear and feel seems extremely real to us; it is only on waking up that we tell ourselves it was ‘just’ a dream. But wait, you might say, all sorts of fantastical things happen in our dreams (things that are obviously ‘unreal’), which never happen in our waking life; so obviously there is a difference between the waking and dreaming states. To which he responds that those dream experiences seem fantastical only from the perspective of the waking state; while you were dreaming, they seemed real to you, very much so - your dream self got scared or thrilled.

Gaudapada takes on every possible objection to this line of thought, and calmly knocks them down. You might rap on the surface of the table in front of you now and say that this is a real, solid, object, that causes real pain when you smash against it - so obviously this is different to the shadowy dream state, right? Not really, says Gaudapada. When you are inside the dream, the dream-table seems as solid to you as this ‘waking-state’ table; while you are dreaming, you hear the sound as your knuckle raps on the dream table, you feel the pain when the dream-chair lands hard on your dream-foot. There is no clean way of saying that one state is more real than the other.

Where is Gaudapada going with this? He isn’t so much lifting up dreams to the same level as waking reality, but pulling down the waking state to be on par (from an ultimate perspective) with dreams. He is doing this to weaken our conviction about the reality of the world we see around is. He is pulling the ground off from underneath our feet. But why is he doing this? Because it is only when we are shaken out of our smug assumptions about this empirical world that we will begin to probe deeper.

By blurring the lines between dream and waking states, Gaudapada is pushing us to ask ourselves ‘What is reality? If the world around me right now is no more real than a dream, is there something indisputably real?’ Gaudapada’s answer is resoundingly positive. The witness of the three states (waking, dreaming, deep dreamless sleep) is real. The dreamer is real, though the dream itself is not; the waker is real, though the waking world itself is not.

In other words, by hacking away everything that we consider to be real across the three states, he draws our attention to what is left over - the substratum of Reality that is manifesting as these three states, the pure awareness in which the dreaming world, the waking world, and the blankness of deep sleep rise and fall, like waves in a limitless ocean.

It struck me, as I read Hoel’s essays, that if we expand the question ‘Why do we dream?’ from the scientific to the metaphysical plane, then Gaudapada too is suggesting an ‘overfitted mind’ hypothesis. We become overfitted to the waking world around us, and take it to be real. If not for dreams, we might continue forever in this state of ignorance. It is often only when we dream, and more importantly when we reflect back on our dream experience after we wake up, that we start to doubt the solidity of the ground we are standing on. And this feeling of freefall leaves us with no option but to try and better understand the nature of reality, to find out what is really real. Dreams, then, in the Vedantic framework, are the noise signals fed into the learning model, to prevent us from being overfitted to the world we see around us right now; dreams open up the path of inquiry.

So then, why do we dream? So that we may Awaken.