And here goes @petarmilin from @ooominds with "What can be used from learning?" with a wink to last week's "What can be learned from usage" 😂 Image
@petarmilin @ooominds Why learning? Poggio added 2 top-levels to Marr: the 4th is learning, the 5th is evolution. Looks like we agree with @haspelmath!
@petarmilin @ooominds @haspelmath This is where the magic happened: @petarmilin 's kitchen in Serbia where he and Harald Baayen started building a computational implementation of the Rescorla Wagner equations that would scale to deal with the huge amounts of data you need to study language @ooominds #History Image
@petarmilin @ooominds @haspelmath Learning is a significant force in shaping the system vs the system dictates what needs to be learned and what learning needs to be able to account for so linguists and psychologist can fruitfully collaborate. @dfdurdevic would agree 😍 Image
@petarmilin @ooominds @haspelmath @dfdurdevic Contrasting different types of learning helps us understand what happens: some algorithms are static and focus on storage & retrieval while others are dynamic. Disciplines have
preferences: usage-based linguists focus on classification and prefer exemplar models such as MBL. Image
@petarmilin @ooominds But 1/ MBL does not necessarily perform best for language and 2/ECL has more explanatory power: it explains why MBL does what it does. You can read the details in our 2016 paper in Cognitive Linguistics doi.org/10.1515/cog-20…
@petarmilin @ooominds Plurality, not religion! Image
@petarmilin @ooominds Can Rescorla Wagner handle more abstract types of input? It can: it learns elegantly from a top-down input layer that contains naive linguistic concepts as capture by Behaviour Profiles. Details here:
doi.org/10.1037/xlm000…
@petarmilin @ooominds Yesterday @dfdurdevic asked whether the individual learning rate that @Adnane_E1 demonstrated manifests outside language as well #StayTuned Image
@petarmilin @ooominds @dfdurdevic It does: different types of learners explore their environment in different ways. Some explore more, and at a more relevant level, which leaves them with more information to exploit later on.
Details here
doi.org/10.1111/cogs.1…
@petarmilin @ooominds Language is too complex for any one discipline to tackle on its own - but together we can thrive! Image
@petarmilin @ooominds BUT we have to tread carefully: let's not overstate our case but build it meticulously because we don't want to risk having learning exiled a second time in 75 years! Image

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Dagmar Divjak

Dagmar Divjak Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!