1/n) A small thread on argument structure... I've been thinking about this bc @TheFrontalLobe_ recently posted about Tim Van Gelder's famous paper, and I turned into a complete jerk in response (sorry again André). I was asking myself, why does that paper make me so irritable?
2/n) I realized why: it's the structure of the argument in the paper. I also realized that many other papers that get under my skin share the same structure. I need to learn to be less of a jerk on Twitter, yes, but I also want to highlight why this structure bothers me so.
3/n) Here's the structure I find so irritating:
1. Concept A is central to current theories of the brain/mind
2. Define A as implying X, Y, Z, claim concept B does not
3. Argue that X, Y, Z are surely not how brains/minds work
4. Conclude that we should abandon A in favour of B
1. Concept A is central to current theories of the brain/mind
2. Define A as implying X, Y, Z, claim concept B does not
3. Argue that X, Y, Z are surely not how brains/minds work
4. Conclude that we should abandon A in favour of B
4/n) Why do I find this structure irritating? It's that step 2... What many authors seem to do is implicitly or explicitly define the concept they're trying to take down themselves, while ignoring what many other researchers actually mean when they use that word.
5/n) In other words, in these papers they don't provide an analysis of how the field *uses* the word. Rather, they impose a definition on the field, then try to argue against this definition. As someone who believes strongly that "use == meaning" this drives me nuts.
6/n) The concept of "representation" is a perfect example. I would argue that an honest account of how we use this word in neuroscience (and ML) is this:
If signal X carries information about variable Y, and this information is used to control behaviour, then X represents Y.
If signal X carries information about variable Y, and this information is used to control behaviour, then X represents Y.
7/n) So, for example, we say that neurons in auditory ctx "represent" pitch, an engram "represents" a past experience, and DA neurons in VTA "represent" reward pred errors. When we say this, I don't think we mean anything more than "X has info on Y, and this is used downstream".
8/n) Now, let's say for sake of argument, that one thinks there's a problem with how a word is used in the field. Well, then the argument should be,
"We use word A in this way, but here's why this usage may naturally also imply X,Y,Z for some people, so A confuses the issue".
"We use word A in this way, but here's why this usage may naturally also imply X,Y,Z for some people, so A confuses the issue".
9/n) That's the argument that many ppl make to me when I argue why brains are computers. Though I disagree with the claim, I think it is a valid critique of my stance, and it doesn't make me even slightly irritable to hear it. It's a sound argument.
9/n) It's also a different argument structure than the one Van Gelder employed in his paper, which was implicitly more like the structure at the top. The key difference is that the first attacks a strawman, whereas the second calls for greater clarity in language.
10/n) So, for this reason, I get irritable when I read these papers, because it feels almost like an intentional rhetorical sleight of hand used to promote a new concept without attending to the actual state of the scientific field. It just feels misleading to me...
Fin) In conclusion, I'm going to continue working on my Twitter manners, but I encourage ppl to ask themselves when they write a paper attacking a concept:
Am I employing a definition of this concept I made up myself, or am I actually considering how it is *used* by researchers?
Am I employing a definition of this concept I made up myself, or am I actually considering how it is *used* by researchers?
