Friday, October 12, 2012

Analogy connecting Proteins to Words in a language

While analogies can be dangerously inaccurate at times, they can also be invaluable in other cases. Therefore, they are worth considering (with open-minded skepticism). So, here is an analogy of molecular biology research...

There are 26 alphabets in the English language and 20 amino acids (maybe more) in biological systems. Combination of these basic alphabets can be used to constructs hundreds of thousands of words, and similarly, combinations of amino acids can be used to create numerous different protein molecules. However, both are retrained in some way. For example, when constructing a new word, we will probably not create something like "qxtrpeeoo", because it is not speakable. Similarly, when making protein molecules, stitching together random amino acids would not produce a stable molecule in its environment. So, there is some similarities.

Now, lets try performing some "experiments" on words in a similar way that we do experiments in biological systems. Lets try a "kockout" experiment. We will remove the word "grocery" from all text in a city. We might see that traffic patterns have changes because people are trying to find grocery stores. After a few weeks, people adjust to the situation, so the traffic patterns settle. So, from this observation, what might we conclude about the role of "grocery" in society? Consider a different experiment in which part of a word, e.g. prefix or suffix, is removed or a different part is added. What type of conclusions would we reach by observing the consequence of such changes?

So what is a different approach to learning the meaning of words? I do not have an obvious answer, but following the above analogy, one might look at how babies learn language. First, they usually learn words that refer to real objects, such as "light" or "fan" or "mom". Applying this analogy to proteins, it might be reasonable to start with proteins that directly interact with the environment, e.g. binds metals or other environmental signals (i.e. objects outside the internal language of the cell). Moving from the word "light" to the meaning of a sentence such as "the light is on" or something more intricate such as "please turn the light on" is a much larger step because here the word "on" is entirely context dependent. The same word would mean something entirely different in a different sentence. It is possible that some proteins, specifically those involved in internal signal processing of the cell, are entirely context dependent. Consider the difference in the response when someone asks "is the light on" vs "light is on" vs "turn the light on". All of these involve "light" and "on", but the response of the listener is quite different in each case.

No comments:

Post a Comment