How formal theories of cognition are wrong, yet useful: Lessons learned from two academic spin- offs focussing on computational memory models
Abstract
Translating formal cognitive models into real-world applications requires bridging the gap between theoretical idealisations and real-world constraints. This talk examines that translation process through the lens of two spin-offs grounded in computational memory theory: MemoryLab, an adaptive learning platform used by 2 million students, and Precision Cognition Labs, focused on early detection of cognitive decline. Where MemoryLab uses memory models to improve education by better memorization and abolishing unproductive examination, PCL uses the same memory models to much more reliably estimate memory functioning - quantifying differences in cognitive performance in populations that traditionally where described as having “subjective complaints”. Drawing on experiences from both ventures, I argue that navigating this gap yields not only practical tools but genuine theoretical insight into memory, individual differences, and the scope conditions of computational cognitive accounts.
