Here’s a summary (from wiki + this video) of word learning biases that I think are relevant
The Whole Object assumption (just what it sounds like). “rabbit” doesn’t refer to the nose of a rabbit.
Type/Taxonomic assumption (discussed in previous comment). I’ll add that the Wikipedia page cites this paper[1] that some evidence suggests that shape plays a big role. This makes intuitive sense to me.[2]
Basic Level assumption. Rabbits are a basic level category while mammals or animals are a higher level category.
To me, 1-3 are pointing at very similar things. My guess: single objects are primitives in the mind, e.g. we chunk a marker as a single thing. The “basic level” is categories on objects, primarily using shape—so “spoon” refers to any object like this spoon. Above that are the higher level ones like utensils (which include forks) or animals, which have less shape similarity and are sometimes thought of as classes of basic level classes (utensils are things like forks, spoons, knives, and chopsticks; though animal feels different to me).
I see this as weak evidence that you can get further without having a principled way to distinguish “levels” via simple heuristics, or that “levels” are easier to work out.
Less relevant: the mutual exclusivity assumption. If today someone calls a rabbit a “rabbit” and later a friend calls his a “babbit”, then “babbit” must mean something like “fluffy” or maybe it’s a name for his pet. Likewise if there’s a tree and a winged slug and I say “Zib!” then “zib” probably refers to the winged slug.
Here’s a summary (from wiki + this video) of word learning biases that I think are relevant
The Whole Object assumption (just what it sounds like). “rabbit” doesn’t refer to the nose of a rabbit.
Type/Taxonomic assumption (discussed in previous comment). I’ll add that the Wikipedia page cites this paper[1] that some evidence suggests that shape plays a big role. This makes intuitive sense to me.[2]
Basic Level assumption. Rabbits are a basic level category while mammals or animals are a higher level category.
To me, 1-3 are pointing at very similar things. My guess: single objects are primitives in the mind, e.g. we chunk a marker as a single thing. The “basic level” is categories on objects, primarily using shape—so “spoon” refers to any object like this spoon. Above that are the higher level ones like utensils (which include forks) or animals, which have less shape similarity and are sometimes thought of as classes of basic level classes (utensils are things like forks, spoons, knives, and chopsticks; though animal feels different to me).
I see this as weak evidence that you can get further without having a principled way to distinguish “levels” via simple heuristics, or that “levels” are easier to work out.
Note: small sample size study—but the lit review in it is neat.
Interestingly, it’s not the substance the object is made of. Markman, E. M. (1991) cites a study using a plastic vs brass pyramid.
Less relevant: the mutual exclusivity assumption. If today someone calls a rabbit a “rabbit” and later a friend calls his a “babbit”, then “babbit” must mean something like “fluffy” or maybe it’s a name for his pet. Likewise if there’s a tree and a winged slug and I say “Zib!” then “zib” probably refers to the winged slug.