Registration to Departmental Seminar, Department of Philosophy (16 Nov 2026)

Title: Can We Develop a Scientific Theory That Is Both Precise and Broad? Yes, and the Cost Is Far Lower Than Expected
Speaker: Prof. Huang Liqiang
Professor, Department of Psychology, CUHK

Date: 16 Nov 2026 (Mon)
Time: 4:30pm-6:30pm HK Time
Venue: Room 220, Fung King Hey Building
(Limited seats. Registrations will be handled on a first come, first served basis.)

Abstract:
Scientists often face a frustrating trade-off between precision and breadth: a model that is realistic and precise usually works only in narrow contexts, whereas a more general model tends to lose accuracy or detail. Here, I argue that, at least in some cases, an optimal balance can be achieved. Specifically, I advocate a Comprehensive Exploration (CE) approach, which relies on large-scale benchmark datasets and on the search for an optimally, moderately complex model through exploratory and iterative model building. In a case study on visual working memory, a CE model with only 57 parameters outperformed a large neural network with more than 30,000 parameters. This finding suggests that it is possible to achieve both precision and breadth with only moderate complexity.

Delivered in English.
All are welcome.

Enquiries
Tel: 3943 7135
Email: philosophy@cuhk.edu.hk