Peace in Our Time? Integrative Cognitive Science

In the 21st century the paradigm wars in cognitive science seem to have greatly subsided, at least in most quarters, and especially amongst younger scholars (thanks to Alex Cristia and Jenny Culbertson for confirmation of this observation). I’d like to claim that we have entered an era of what we can call “Integrative Cognitive Science”. I talk about the end of the paradigm wars from the perspective of a generative phonologist to a relatively broad linguistics audience in the video clip linked here. My presentation goes from about 13 to 23 minutes of the recording. The other speakers, John Hale, Mark Liberman and Adrian Staub touch on related themes from their own perspectives. (January 9, 2016). 

“Integrative Cognitive Science” can be defined as research that integrates theories, goals and methods across the divides of the debates. Prince and Smolensky’s Optimality Theory is one example, as is Smolensky’s Integrated Connectionist Symbolic framework, and there are many more that go under the labels of “hybrid” theories or “two systems” theories. I’ll be looking for good examples, especially ones that have led to breakthroughs in our understanding of cognition, and to applications in AI and other domains (Jan. 10th 2016)

Alan Prince and Paul Smolensky collaborated on a 1991 Linguistic Institute course entitled “Connectionism and Harmony Theory in Linguistics”. Other “memorabilia” from their collaboration can be found here.

From MIT Press:

In The Algebraic Mind, Gary Marcus attempts to integrate two theories about how the mind works, one that says that the mind is a computer-like manipulator of symbols, and another that says that the mind is a large network of neurons working together in parallel.

The “integrative” characterization of Marcus’ work is challenged in this review.

The 1954/1965 book by Osgood and Sebeok that coined the term “Psycholinguistics” presents an agenda for integrative cognitive science on its back cover. This is pre-Chomskyan psycholinguistics – behaviorist learning theory, information theory, and structuralist linguistics. (Update: Not sure if psycholinguistics was coined there – wikipedia has some earlier citations).

AAAI Spring Symposium on Knowledge Representation and Reasoning: Integrating Symbolic and Neural Approaches, Stanford University, CA, March 23-25, 2015‎. A quote from Hinton’s abstract:

The fathers of AI believed that formal logic provided insight into how human reasoning must work.  For implications to travel from one sentence to the next, there had to be rules of inference containing variables that got bound to symbols in the first sentence and carried the implications to the second sentence.  I shall demonstrate that this belief is as incorrect as the belief that a lightwave can only travel through space by causing disturbances in the luminiferous aether.

Not really on integration per se, but this page summarizes a panel discussion by LeCun and others that touches on the question of whether there will be a third neural network winter.

More evidence that “integrative” is a buzzword these days, and also that this is an active research direction – a “Dear Colleague” letter from the NSF on “Stimulating Integrative Research in Computational Cognition”

A 2016 paper on Integrating neural nets with first order logic.