Concepts

Doing without concepts?

The project of the hermeneutic net is a kind of conceptual engineering. Ostensibly, it is about concepts.¹ But what are concepts?

In Doing Without Concepts (DWC), Edouard Machery argues that concepts don't form a natural kind and therefore that the notion of "concept" is of no use to cognitive psychology. His primary motivation there is to put an end to the pointless divagations running amok in cognitive psychology about "the nature of concepts".

Machery considers four candidate explication of the notion of a concept: (1) default bodies of knowledge, (2) temporary bodies of knowledge in working memory, (3) bodies of knowledge under organisms control, (4) concepts as constituents of thoughts,² (5) categorization devices. Machery argues against 2-5 and settles for 1. Then he argues for the concept heterogeneity hypothesis, according to which there are very different kinds of default bodies of knowledge that have very little (~no) relevant properties in common, beyond serving as, well, default bodies of knowledge (he focuses ond prototypes, exemplars, and theories, but admits that there are likely more of them).

Having concluded that concepts are not much of a Thing, Machery goes on using the word "concept" because the social habit of using the word "concept" is too sticky and it's not practical for him to try too much to change it. Had everybody (or at least a sufficient majority of philosophers) agreed with him about concept heterogeneity and decided to switch to a more precise language, he would stop using the word "concept".

But would he?

Even though I was strongly inclined to buy his argument, I still find myself talking about "concepts". This suggests there is some kind of Greenspun's tenth rule at play here. Even if you try to eliminate the concept of "concept" from your language/ontos, it finds its way back in because it serves some important role. What is this role?

At the time of reading DWC, I was like "Yeah, maybe he's right about taking concepts to be default bodies of knowledge… but maybe not quite? IDK, something feels off…".

What happens in a mind in situtations that we describe using the word "concept"?

Calling concepts "default bodies of knowledge" presumes that they are/contain knowledge and are used by default.

Knowledge: For a mental element to "count as knowledge", it has to satisfy some criteria. What these criteria are (or should be) exactly is debatable but for ~propositional knowledge, there's certainly truth and for knowledge-how (procedural knowledge), there's "ability to effectively accomplish some goals". When a concept is (judged as) "wrong", that's not necessarily because of the concept including some false assumptions (although this does happen). It's more often the case that a concept is (judged to be wrong) when we realize that it fails to do the work that we want it to do (e.g. reason about minds). Overall, it might be more appropriate to say that concepts are "bodies of information" or "bodies of mental structure".

Used by default: This seems to be missing something important. When I say "pass me a pillow", the recipient of the message retrieves from their memory the mental structure «pillow», uses it to locate the object and take an appropriate action. There is a default body of information in there (in memory?) that the receiver draws on to take an appropriate action but once retrieved, but it is retrieved in a selective way and what ends up being reconstructed in memory differs depending on the circumstances. E.g. depending on the context in which I'm prompted to construct a mental structure of the "dog" concept, I end up constructing a different structure.

So there seem to be 2 things: (1) the bodies of information/[mental structure] in memory; (2) the things in live/active thought that can draw selectively from these bodies and adjust on the fly to do stuff. Obviously, they are related but using the word "concept" for both is confusing.

OK but then, what's going on when we use a (lexical) concept in communication?

• In the case of asking to pass a pillow, I expect³ that they interpret the word "pillow" as a pointer to a body of information in their memory that is sufficiently similar to mine to make them do what I want them to do.
• If I'm trying to explain some idea to someone, I am using (i.a.) pointers to stuff in their memory to make them construct a mental structure sufficiently similar to what I want them to construct.
  • As a special case, I may want to give them a new concept. In my mind, this concept is related to other concepts and if the other person shares⁴ those concepts, I can skillfully point at them to make them construct this conceptual structure and solidify in their memory, associate it with a new lexical phrase.

Concepts as messages

My initial best alternative-to-Machery's guess was that concepts were better thought of as something like (1) constituents of (~language of) thought and (2) ~propositional compressed (sub-)messages used in communication. The two are clearly related: communication is not just inter-personal but also intra-personal, both diachronically, via memory, or synchronically, since compressed messages are easier to hold in awareness, facilitating thought. Now I see the supposed concept-messages more as pointers to structures in memory that elicit mental structures functioning as constituents of thought.

Mathematical concepts

Mathematical concepts (and other formal or very precise) concepts are outliers in this because they are relatively context-independent, the memory→awareness mapping is very stable. What a group is doesn't change depending on whether you're talking about GL₄ or the Rubik's cube group.