A Minimal Task Reveals Emergent Path Integration and Object-Location Binding in a Predictive Sequence Model
This provides a mechanistic account of sequential world modeling relevant to cognitive science, though it is incremental as it builds on existing hypotheses in a minimal setting.
The study tackled the problem of understanding how predictive neural networks form structured internal world models by testing if action-conditioned sequential prediction suffices, and found that a recurrent neural network learned path integration and dynamic object-location binding in novel scenes, with prediction accuracy improving across sequences.
Adaptive cognition requires structured internal models representing objects and their relations. Predictive neural networks are often proposed to form such "world models", yet their underlying mechanisms remain unclear. One hypothesis is that action-conditioned sequential prediction suffices for learning such world models. In this work, we investigate this possibility in a minimal in-silico setting. Sequentially sampling tokens from 2D continuous token scenes, a recurrent neural network is trained to predict the upcoming token from current input and a saccade-like displacement. On novel scenes, prediction accuracy improves across the sequence, indicating in-context learning. Decoding analyses reveal path integration and dynamic binding of token identity to position. Interventional analyses show that new bindings can be learned late in sequence and that out-of-distribution bindings can be learned. Together, these results demonstrate how structured representations that rely on flexible binding emerge to support prediction, offering a mechanistic account of sequential world modeling relevant to cognitive science.