Gary Marcus

Just read the Berent & @GaryMarcus response. Summarizes familiar arguments against 1980s-style connectionism, but it would have been useful to see a discussion of more recent work, e.g.https://twitter.com/Joe_Pater/status/1048569045438541826 …

Recent neural architectures with structured inductive biases (syntactic, relational, compositional, etc), which I assume would not be taken to follow the "associationist hypothesis" (from @redpony, @jacobandreas, and others),

Which work are you thinking of here?

This review from DeepMind has a lot of references (incidentally we're reading it this week in my seminar): https://arxiv.org/abs/1806.01261 

Also the Stanford Tree RNNs stuff from @RichardSocher, @sleepinyourhat et al

Thanks for pointing me to this. It’s an interesting response, and in most ways a more current version of arguments I’ve usually heard critiqued in a more dated form.

(Of course, I don’t agree with everything, especially the rosy impression they give of the way innateness issues are treated in linguistic theory.)