Friday, September 3, 2010

debunking chomsky's poverty of the stimulus

Melody, a researcher in cognitive science at Stanford, has posted a detailed discussion of the problems with Chomsky's famed povery of the stimulus argument from the perspective of the last 40 years of computational learning models. Hindsight is always 20-20 right?

Money quote:

....there are at least two goals of modeling in cognitive science : 1) to discover the best computational method of accounting for a given phenomena, 2) to discover the best account that is also psychologically plausible. The goal has never been to rule out a whole class of models on the basis of one ill-starred example.  Because — quite frankly — models don’t deal in ‘logical possibilities.’  They are not mathematical or logical proofs.  Step 3 in Miller and Chomsky’s paper is a pseudo-scientific non sequitur.

The whole post is well worth reading.

2 comments:

mike said...

In support of the poverty-of-stimulus argument, check out Legate and Yang’s “Empirical reassessment of stimulus poverty arguments,” The Linguistic Review 19: 151-162. Also see Charles Gallistel's chapter, Learning Organ, in The Chomsky Notebook.

Huuk said...

Using the term "debunking" in regard to a scientific debate as subtle and complex as advancing, refining, or editing the 'poverty of stimulus' hypothesis produces, in me at least, a strong bias against the competence and trustworthiness of the blog as a whole, as does the claim in the post "Why Linguists Should Study Math" that "Math is good training for the mind. It makes you a more rigorous thinker." This claim is, I think (as someone with a fair amount of math), as thoroughly discredited as the similar claim for learning Latin.

Nuts and Bolts of Applying Deep Learning (Andrew Ng)

I recently watched Andrew Ng's excellent lecture from 2016 Nuts and Bolts of Applying Deep Learning and took notes. I post them as a he...