Revisiting Associationism: Bridging Empiricism and Artificial Intelligence in Mind Representation

Description: The Association Association is a group of philosophers working on the history and current applications of associationist ideas about the nature of the mind, including the origins of the mind’s representations, the structure of inference, and appropriate standards of rationality. This cluster of views is most often attributed to philosophers like Locke and Hume, but includes also a wide range of other broadly empiricist figures, such as Aristotle, Aquinas, Ibn Sina, Hobbes, Berkeley, Bentham, Hartley, Bain, Reid, Darwin, De Grouchy, James and John Stuart Mill, Freud, James, and many others. More recently, associationist theses have been endorsed by prominent figures in cognitive science like Robert Rescorla, Linda Smith, and Cecilia Heyes, and especially in some of the highest profile contemporary debates over the prospects for artificial-neural-network-based artificial intelligence, by figures like the Churchlands, Sejnowski, Smolensky, Elman, McClelland, and key figures at marquee R&D companies like DeepMind, AnthropicAI, and OpenAI.

Despite these high-profile defenders, nuanced expositions of associationism remain rare. Associationism has often been used as a foil against which to develop alternative views, and as a result, most discussions of associationism focus on simplified or radical versions of the position. In particular, the more complex or moderate positions of major historical philosophers and contemporary scientists are often conflated with those of the most radical and extreme empiricist views, especially the radical behaviorism of American psychologists such as Watson and Skinner. Nearly all associationists tend to be united by anti-nativism about the mind’s contents or representations–in particular, they tend to hold there are no innate ideas, concepts, or propositional structures. However, this shared dedication leaves wide room for holding that other parts of the mind’s architecture are not learned through associative principles, especially:  a basic set of associationist learning principles themselves (such as contiguity, similarity, and cause-and-effect), architectural elements like faculties (such as memory, imagination, and attention), other general learning processes (such as methods of abstraction), passions (or affects/emotions), salience biases in perceptual input (such as towards certain colors, sounds, or tastes), and forms of internal reflection (involving elements like epistemic feelings and inner speech). A variety of new architectural developments in artificial neural network design also recommend re-appraisal of traditional arguments about the in-principle limits of statistical or associative learning models. The Association Association is united by the conviction that the subtlety and individual variation amongst the views of both historical and current associationists merits further careful scholarship and that the charitable application of associationist views can usefully inform our appraisal of these new technological developments. 

The first public meeting of the Association Association will be held at Gainesville, FL Jan 23-24. Research talks will be delivered by Cameron Buckner (UF), Hayley Clatterbuck (Rethink Priorities/UCLA), Mike Dacey (Bates), Tyler Delmore (York), Tamas Demeter (Hungarian Academy of Sciences), Trip Glazer (Dayton), Chuck Goldhaber (UF), Javier Gomez-Lavin (Purdue), Michael Jacovides (Purdue), and Talia Morag (Australian Catholic University). Submissions are invited for additional research presentation slots. We welcome submissions from faculty and graduate students, on thematically-relevant historical figures or current topics in psychology and artificial intelligence (1000 word abstracts or 3000 word papers, with a slight preference for papers). Submissions will be reviewed and travel bursaries will be available for accepted abstracts/papers (amount contingent on availability of funding). Please indicate in your submission whether you would like to be considered for a travel bursary.

A limited number of (free) registrations will also be available for non-presenters. Please contact [email protected] if you would like to attend.