,
25 tweets,
5 min read
Read on Twitter
Thread for the interested (or masochistic) neurotwitter verse...
This is a great little essay by @RomainBrette because it truly does get at the heart (pun intended) of our disagreement:
This is a great little essay by @RomainBrette because it truly does get at the heart (pun intended) of our disagreement:
IMO, Brette, and others with his perspective, take an approach to the concept of metaphor that is misleading, and which serves as a constant tool for obfuscation in these arguments.
Example: Brett says quite clearly here that the heart is not actually a pump, but it's a metaphor. The idea behind this claim is that a metaphor is a "morphism", a structure preserving mapping.
According to this perspective, there is a class of "true pumps" in this world (presumably the pumps we manufacture), and there is a structure preserving mapping between the pieces of a heart (valve, liquid, etc.) and the pieces of a "true pump".
This entire perspective is completely wrong from the ground up, IMO. It attempts to treat language as if it forms pictures of the world for us, mapping our linguistic structures to the structures of the world, which is a dead end path.
When is something a literal truth? E.g., when is something literally a pump? This ultimately comes down to a question of how our words derive their meaning.
Is there a class of "true pumps"? Can we divide the world up into "pump" and "not pump" cleanly? Is the literal meaning (as opposed to the metaphorical meaning) of the word "pump", "an object from the class of true pumps?"
If so, then any object, like the heart, that is not in the set of "true pumps", but which we nonetheless are inclined to refer to as a "pump" because there is a morphism is a metaphorical pump. We use this morphism for explanatory power, not because the heart is actually a pump.
This is @RomainBrette's perspective, and it is just so wrong, IMO. He is falling into a classical philosophical trap, which stems from the desire to treat language like a formal system.
Per Wittgenstein's ideas in Phil. Investigations (static1.squarespace.com/static/54889e7…), meaning emerges from use, from the manner in which we play games with our words. Accordingly, something is literally true if it aligns with our usage, if it follows the script of our language game.
So, how do we use the word "pump"? Generally, we use it to refer to objects that move liquids via force and valves. Thus, the heart is literally a pump, according to our language games.
When is something a metaphor then? The correct answer, IMO, is not Brette's answer: it is not "when there is a morphism". A metaphor refers to a situation where we knowingly break the rules of our language game.
So, for example, I normally use the word "snake" to refer to certain types of animal that are long and without arms, etc., but if I say, "oh watch out for him, he's a real snake", there is no morphism. It's a metaphor because I'm knowingly violating the game for artistry's sake.
When I say "the heart is a pump", that is not what I'm doing. Rather, it is perfectly in line with my typical rules of play with the word "pump", i.e. to describe objects that move liquid with force and valves. It is not metaphorical.
It is the same with "code" and "computer" in neuroscience. @RomainBrette (and many others) think that these words are used metaphorically in neuroscience. They are not.
The claim that these words are metaphors here rests on the idea that there is a class of "true code" and "true computer", of which our brains are not members, and that we use these words in order to utilize the morphisms between the true class and our brains.
I call BS. According to the standard games we play, there is "code" whenever information is transformed for the purposes of communication. Thus, an automatic door uses voltages from an IR sensor to encode the presence of an object and communicate it to the opening mechanism.
An automatic door does not metaphorically encode the presence of an object, it literally encodes it. A thermostat does not metaphorically encode temperature, it literally encodes it. Why? Because these usages do not violate our usual language games.
It is the same with the brain. Within neuroscience, we use the word "encode" to refer to situations where information is transformed into spikes for the purposes of communication. Nothing about this violates typical usage. It's not metaphorical.
Likewise, within computer science, we use the word "computer" to refer to things that perform Turing-computable operations, and this applies to the brain with no violation of usage. Hence, the brain is literally a computer.
All of this points to a difficulty that smart people like @RomainBrette and many others have: they want to fit everything in science into a nicely packaged formal system of language, and hope that we can develop a formal system that is universally applicable.
But, this is not possible, and it is not a goal we should have. The only question we should ever have in front of us is: what is the utility of this word? Do we gain something by playing this game with these words?
In the case of "code" and "computer" in neuroscience, my answer is a firm "yes". We open up a vast field of mathematical analysis that has proven quite fruitful, and which has served as the basis for systems neuroscience for decades.
The desire to limit our language, to say "here are the true computers/codes, you are only using a metaphor", is both a misunderstanding of how language works, and a tactic that will only impair progress in neuroscience.
Those of us who actually want to get on with continued progress in our field should resist such calls to set unnecessary limits on our language games. Don't kid yourself kid: the secret sauce ain't your new paradigm, it's just getting into the mud and getting work done.
