Post your comments on the papers and presentations from 10/15 and 10/16 here.
2 thoughts on “Papers and presentations round 2”
Charisse
Comments related to: Marcus et al (1999) Rule Learning by 7-month-old Infants
Have the findings that infants have at least two types of tools (statistical and algebraic) to learn about their environment, specifically language acquisition been applied to language or communication disorders? Presumably understanding how typical development may occur will provide opportunity to better detect atypical development. In addition to early detection of possible language development disorders, I’m curious if or how statistical and algebraic learning compare to strategies of infants or children who may have a disorder, but are still successful at acquiring language. Lastly, I’m curious whether these tools are equally used for children growing up in a multilingual household?
Charisse: These are great questions. I suspect, but am not absolutely positive, that the field is still not at the point where we have a sufficient understanding of of “statistical” and “algebraic” learning that we can answer them, or that we even know that there are two distinct types of learning. That is, a lot of the explicit models that people construct to explain learning data have properties that are a mix of what Marcus is calling “algebraic” and “statistical”. For example, connectionist models are not incompatible with variables, and so connectionists have subsequently shown that their models can handle Marcus’ data. There’s very detailed discussion of this in Marcus’ book “The Algebraic Mind”, which I have a copy of and can lend anyone who is interested. I’ll try to talk about some related issues w.r.t. the models that are used for category learning on Monday.
Comments related to: Marcus et al (1999) Rule Learning by 7-month-old Infants
Have the findings that infants have at least two types of tools (statistical and algebraic) to learn about their environment, specifically language acquisition been applied to language or communication disorders? Presumably understanding how typical development may occur will provide opportunity to better detect atypical development. In addition to early detection of possible language development disorders, I’m curious if or how statistical and algebraic learning compare to strategies of infants or children who may have a disorder, but are still successful at acquiring language. Lastly, I’m curious whether these tools are equally used for children growing up in a multilingual household?
Charisse: These are great questions. I suspect, but am not absolutely positive, that the field is still not at the point where we have a sufficient understanding of of “statistical” and “algebraic” learning that we can answer them, or that we even know that there are two distinct types of learning. That is, a lot of the explicit models that people construct to explain learning data have properties that are a mix of what Marcus is calling “algebraic” and “statistical”. For example, connectionist models are not incompatible with variables, and so connectionists have subsequently shown that their models can handle Marcus’ data. There’s very detailed discussion of this in Marcus’ book “The Algebraic Mind”, which I have a copy of and can lend anyone who is interested. I’ll try to talk about some related issues w.r.t. the models that are used for category learning on Monday.