Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session A: Tuesday, August 12, 1:30 – 4:30 pm, de Brug & E‑Hall

Sparks of cognitive flexibility: self-guided context inference for flexible stimulus-response mapping by attentional routing

Rowan P. Sommers¹, Sushrut Thorat², Daniel Anthes², Tim C Kietzmann³; ¹Institute of Cognitive Science, Osnabrück University, Universität Osnabrück, ²University of Osnabrück, ³Universität Osnabrück

Presenter: Rowan P. Sommers

Flexible cognition entails a rapid adaptation of stimulus-response mappings. Standard neural networks struggle in tasks requiring rapid remapping. Here, we propose the Wisconsin Neural Network (WiNN), which generalizes fast-and-slow learning to real-world tasks demanding flexible behavior, using adjustable context states that guide attention in a pretrained convolutional neural network. We evaluate WiNN on a variant of the Wisconsin Card Sorting Task, revealing several markers of cognitive flexibility: (i) WiNN autonomously infers underlying rules, (ii) requires fewer examples than control models reliant on large-scale parameter updates, and (iii) can perform rule inference solely via context-state adjustments. This approach offers a path toward context-sensitive models that retain knowledge while rapidly adapting to complex, rule-based tasks.

Topic Area: Visual Processing & Computational Vision

Extended Abstract: Full Text PDF