Languages and Trees

[Sai Saurab]

Hi Ashish
A few questions that may seemingly be disconnected but have a common thread.
From here as well as other material you have written, I get the sense that the unambiguous representation of meaning is in trees. But all the natural languages that I know of use a linear representation and using context and background knowledge I re-construct the tree in my head to grasp the meaning. Of course, there is some structure that we use like books, chapters and paragraphs. Is there some fundamental limitation by which ideas cannot directly be communicated as trees?
In that article, you say that Sanskrit reduces ambiguity as compared to English but we see that even Sanskrit texts are often subjected to multiple interpretations by application of various grammar rules.
It is even more surprising when you consider that even all programming languages are written left to right in a flat fashion and are then converted to Abstract Syntax Trees. This can be quite misleading many times when you don’t realize the tree nature because of which a change of one line of code may have huge consequences elsewhere downstream in the tree structure. Would more direct/raw manipulation of ASTs be a better programming interface?
Even DNA, RNA etc seem to be information encoded into linear sequences.
Is there a canonical reason for this? Would really appreciate your insights regarding this.

[Ashish Dalela]

When I was still an undergraduate student, I had gotten interested in the Protein Folding Problem. The problem is that all proteins are linear sequences of amino acids. The sequence of amino acids in a protein is encoded in the DNA, but what is unique to the protein is how it folds. Some folding patterns are functional, and some of them are dysfunctional. If proteins in your body start folding in a different way, without a single change to the molecular structure, you might have many diseases. Nobody really knows why the proteins fold in a specific way, and the problem is ongoing for more than 60 years. For any large protein, the number of conformations (folding patterns) is so large that it is computationally impossible to decipher one of these patterns in the time that the universe has existed! Imagine what it takes to decipher the folding patterns for trillions of proteins in our body. As an undergraduate, I did not know enough, but I believed that it needs some fundamentally different approach than physics.

That approach, I can now say is due to the various types of combinations permitted by meanings. For example, all of chemistry today rests on the single idea that two opposite charges attract, and similar charges repel. Let’s call that idea “opposites attract” for simplicity. But in the real world of people, there are many other ways by which complex structures are created. For example, “complements attract” to form organizations and teams. Likewise, “similarity attracts” as people with similar goals come together for a common purpose. Then, “neutrality attracts” as we are drawn toward societies where there is equality, justice, meritoriousness. Then “individuality attracts”, not because it is an opposite to something, or because it is a complement of something, or because there is something shared, or neutral, but because it is inherently attractive.

The reality is so incredibly complicated that we can formulate some theory based on “opposites attract” but because we neglect so many other effects, we cannot explain the observation.

When you see some linearity in a language it is because our consciousness is linear–we read one thing at a time, and it goes linearly. Even if you produced a tree-like structure, a person will still read it linearly. A computer certainly parses a tree linearly (e.g. breadth-first or depth-first). So, linearity is the property of our observation of phenomena, and non-linearity is the reality.

The linearity of a book is misleading. While writing a text, we frequently refer to the previous sentences, the previous paragraph, previous chapter, things outside the book, even forward-looking statements, and then many kinds of back and forth. If you truly draw an information graph of a book, it will be an extremely complicated mind-map. But it is denoted linearly for our reading. Our minds, intellects, senses, etc. are non-linear, but our choices are linear.

When reality is non-linear, and our observation is linear, then for the same reality, there are potentially infinite linear sequences of observations. Just like you can express the meaning of the book using a different sequence of sentences, paragraphs, chapters, etc. Ultimately, it is a humongously complicated graph that can be linearly parsed–i.e. it is a tree. But just because you are parsing it linearly in one way doesn’t mean that is the only way. And the evidence for that is that you get probabilities in your prediction. Given X, the next event can be A, B, C, etc. with probabilities of a, b, c, etc. Basically, you don’t know where in the tree you are.

So, using these probabilities, some people try to do machine learning algorithms. What they don’t realize is that you change one small thing slightly and it has a totally different meaning, so you get lots of false positives and false negatives, but since the probability is better than nothing, people call that “progress”. It is as much progress as quantum theory giving us probabilities. It is just that we are able to extrapolate atomic probabilities to molecular probabilities. So, the linearity is misleading. It just seems linear, because we see one thing at a time.

[Sai Saurab]

Fascinating! It is interesting you should mention protein folding and machine learning. Any thoughts on Alpha Fold 2 being touted as a big breakthrough?
I just realized that the reason why even a computer linear traverses a tree is due to the underlying abstraction of a turing machine right? which is a head operating on a linear tape. That fundamentally limits the power of a computer to simulate reality as reality is non-linear as you point out.
Conscious thought, as you say, is linear but much of the heavy lifting in cognition is not done at the conscious level for eg., sensory perception or skills like walking or driving.
Also in the context of using ideas like complements attract or similarity attracts it makes sense when you explain it, but I feel silly in trying to think in that fashion which I suspect is due to the heavy conditioning of contemporary education. Reminds of a comment(clearly a non sequitur) on this blog post –  “revel in the delusional comfort of anthropocentric vanity about the way the world was imaged to be thousands of years ago”. My question regarding this is – For creating laws of how protein sequences (as meanings) form / interact, do you get the ideas from human / psychological interactions?

[Ashish Dalela]

There are many goggles through which we see the world. Given some goggles, the world is seen in a certain way. And there are many layers of such goggles. What you call “ideas” are very superficial goggles. Underlying these ideas is another goggle of intellect, which has beliefs, of what we consider true or false. Below that goggle is another goggle of ego, of what I think I am, or my picture of myself. And even below that goggle is another goggle of values, or what I consider valuable, which is called mahat. Even deeper than this goggle is my idea of happiness.

From this idea of happiness, comes the notion of what is valuable. From that value judgment comes some identity and goals. From that identity comes some beliefs. From those beliefs come some theories or interpretations of the world. The mind is the external-facing instrument for interpreting the world, but underlying that external interpretation is one’s idea of oneself.

Unless we change that inner idea about who we are, we cannot think differently. It is not that you analyze society and people and say that that’s how the atomic world must be. You must have a different type of goggle through which you see the world differently. Otherwise, there are materialistic theories of society. So, if you want new ideas, then you have to change your idea of yourself. As this deep-level idea is changed, the superficial ideas are automatically altered. Then you get new thinking. If you understand this process, then the question is not about what the world is. The question is: What is your idea of yourself from which all this is coming?

Therefore, we are not interested in the philosophical analysis of the world based on some set of goggles. There are infinite such goggles. We are interested in asking: How should I see myself, what should be my values, my idea of happiness, from which I create an identity? And spiritual life is meant to familiarize ourselves with alternative goggles. But we have to practice changing the goggles. This is called cheto darpan marjanam. The mirror of the heart is dirty, so whatever you see is dirty. If the mirror is cleaned, then a clear image of reality is created.

So, we have a different epistemology or the process by which we know. And the essence of that epistemology is that we don’t waste our time with arguments, observation, analysis, speculation. Our goal is to change the goggles, and then everything is changed. All these scientists and philosophers are speculating: maybe the world is this way or that way, and for us, that is a waste of time, because their goggles or their fundamental idea about themselves is faulty.

There is no need to ask a scientist if his theory is correct. Just ask: Who are you? And if he says, “I” don’t exist, it is a constructed identity, then we can say: “good luck”. The problem is that all these dissatisfied, disgruntled, unhappy intellectuals have become teachers in today’s society. In the Vedic society, such fools will never be considered intelligent, because someone will ask: “So, did you become perfectly happy by your philosophy?” Then the emperor will be naked.

[Author]

Posts