We’re still struggling to figure out exactly what Consciousness is. One theory that I kinda like (although I have no sense of how correct it is) is that consciousness is a way of approximating the base reality. There is too much data coming in through our senses, and it is impossible for every part of the brain to process everything at the same time. So it creates an approximation, which is perceived in the form of consciousness. We can refine this statement slightly by saying that the brain is approximately a neural net, or some form of a predictive machine. The inputs to the machine come from the senses, the emotional and physical state of the body, the stored memories, and the stored habit patterns, and thoughts, and other higher order stuff that is not accessible to the consciousness and hence cannot be described in words, and the output is prediction of the future.

It is not clear why out of this process consciousness emerges, but it does. Somewhere in the prediction process, it becomes necessary, or simply efficient, to have awareness. My guess is that through the approximation that is consciousness, parts of the brain are better able to process the world. A computer is a somewhat good analogy. At the core of every computer is a constant stream of binary data being processed by the CPU. But we see none of that, instead, there are layers upon layers of abstraction, each layer making it easier to understand and interact with the layer below. Nevertheless, the only “reality” within a computer is in the binary. In the same way, the consciousness is a layer of abstraction that makes it easier to access reality, which is otherwise too vast to process continuously.

This model helps us to also formalize Mindfulness. If we agree that the brain is a constantly evolving neural net that is building models to better predict the future, then one of the inputs in this neural net should be the state of the system itself. The operating system in your laptop needs to constantly check its own state to make sure things are working fine. This is what Mindfulness is. Its intent is to create a feedback look for the predictive algorithm. The more input a neural net has, the better model of the world it builds (supposedly). Note that the consciousness is not the entire state of the brain. It is just the output of its input-world-processing center. So we are still unable to create a perfect feedback loop, but nevertheless having this partial feedback loop is better than having none.

Here’s a concrete example. Say every morning you get a craving for a cigarette. This is because a memory system gets triggered when you wake up and remembers that a cigarette in the morning forces your body to release dopamine, which is a desirable state to be in. This then triggers the motor functions necessary to smoke. But after you smoke, chances are your body feels like shit because of the trash it needs to process. If this goes unnoticed and does not form a memory, a brain system that would have gotten triggered to prevent the smoking never comes into existence. By being mindful of the post-cigarette state, you create a competing time-triggered brain system that counteracts the initial urge for smoking.

Another concrete example comes from GPT-3. GPT-3 has been trained upon the entire internet corpus, and so it produces amazingly real sentences. However, it does not take its own output as input and as such small errors start compounding and very soon its sentences become gibberish. It is unaware of its own mistakes and as such is unable to correct them. GPT-3 is the epitome of a non-mindfulness.

To summarize, Mindfulness is a method/paradigm for creating (partial) feedback loops within the neural network of the mind to improve its predictive ability.