Jerry Fodor; Proof-Theoretic Semantics

[gusl]

I like Jerry Fodor!

Would he be a pioneer of the so-called "computational turn"?

... a few link hops away, two Wikipedia articles: Proof-Theoretic Semantics (where the meaning is the use), and Logical Harmony

Btw, like Fodor, I am not interested in connectionist models... While I accept that intelligence must ultimately be implemented in terms of neural networks, I think it's more useful to think about phenomena at a higher level of abstraction, where logic plays a role. It's also more fun. It should also come as no surprise that I'm a fan of ACT-R.

What have we ever gained from NN modeling?

Comments

[bondage-and-tea.livejournal.com]

With connectionism you get learning, which is nice.

[gustavolacerda.livejournal.com]

What kinds of learning can you explain through connectionism? I can imagine that it explains things like reinforcement learning and psychomotor learning.

[bondage-and-tea.livejournal.com]

That's a bit like asking, "what kinds of things can you describe using language?" I don't think I have enough time to answer :-)

What kinds of things would you like to explain and for you what constitutes an explanation?

[xpaerimtlslaekv.livejournal.com]

It is reminiscient of William Calvin's postulate of serial order 'learning/planning' which according to him, eventually enabled the process of evolving symbolic thought.

I would agree that Neural Networks are limited, I find the method used in the implementation of darwinci to be useful, cross modal/cross knowledge base isomorphisms and connections using a hybrid of neural networks, [evolving rule sets in expert systems, and metarulesets across isomorphic groups in differing levels of so -called 'concepts'. Darwinci uses a database of images and sounds to reinforce linguistic 'concepts' across each..
Unfortunately, I believe Darwinci also uses genetic algorithms, which perform quite appropriately, but do not model complex conscious 'concepts/intelligence' [of this form]


Darwinci- [creativity] (http://www.hybridsociety.net/pdf_files/darwinci.pdf)

[I am devising an algorithm for evolving temporal reasoning.]

[xpaerimtlslaekv.livejournal.com]

An individual had said that many implementations of artificial intelligence lack the process of 'double-entry' or reassociative 'concept-rule sets', and Darwinci mitigates this.

[fub.livejournal.com]

Unfortunately, there is no real way to know what your connectionist model is learning. Horror-stories abound.

When I was busy with my graduation project (I studied Cognitive Science in Nijmegen), my professor always told me that there were three ways to make intelligent systems:
- If you know how humans reason about the domain, then just implement that in software, and you're done.
- If you have a good heuristic, just implement that in your program -- it will work most of the time.
- If you have no clue whatsoever how to tackle the problem, just use a connectionist model! Train it a bit, and you'll always get some output...

Also, with the right knowledge representation, declarative systems might 'learn' as well, if you allow the knowledge base to be updated.

[bondage-and-tea.livejournal.com]

Nice hierarchy of understanding :-)

Not sure I agree with the "if you have no clue whatsoever how to tackle the problem, just use a connectionist model" -- don't you have to have a vague idea of how to represent the problem in order to get a decent net, for instance what features are salient for the problem.

I'd like to learn more about symbolic learning algorithms. I know of the likes of ILP (http://www-ai.ijs.si/SasoDzeroski/ILPBook/) and ADATE (http://www-ia.hiof.no/~rolando/), but have just played at the software level, not tried to understand the algorithms.

Also interesting is the hybrid connectionist-symbolic stuff, e.g. Pinkas's translation of his penalty of logic into hopfield networks. There's some interesting more recent stuff of this nature going on.

[bondage-and-tea.livejournal.com]

And I HATE using the phrase "neural" networks with these very high-level models of distributed computation.

[gustavolacerda.livejournal.com]

hm... so "connectionist" can also refer to distributed computation at a level higher than the neuronal level?

[bondage-and-tea.livejournal.com]

Well, neurons do lots of crazy stuff involving protein synthesis, ion channels, chemistry. It's not clear that they do linear summing of stuff coming into the dendrite, apply a non-linear function, and then spew out the result to an axon. There's a large amount of variation on how neurons communicate with each other. It seems that dendrites perform complicated computations, something which isn't mentioned at all by "neural network" theories to the best of my knowledge. Real neural networks are a huge mess!

[xpaerimtlslaekv.livejournal.com]

This article may interest you..
temporal implications[navigation] of self-sustaining control in semiotic systems... (http://www.sdu.dk/Hum/bkw/Meaning.pdf)
Organized criticality.

I have not read any articles yet on ACT-R. I will this evening. [Be well]

[xpaerimtlslaekv.livejournal.com]

http://66.102.7.104/search?q=cache:9IpnNYvKBOQJ:www.sdu.dk/Hum/bkw/Meaning.pdf+%22emergent+semantic%22+frege&hl=en

Link removed, here is the non-pdf version of the article

[gustavolacerda.livejournal.com]

it's good to see that you're a real person!