ReGraph: A Computational Account of Emergent Generalization in the "what" and "where" Dual Visual Streams
Where generalization capacity--the ability to extract context-invariant relational structures--first emerges remains a central question in AI and neuroscience.
ProofPaper ↗
Key points
- The foundation for this capacity lies upstream of the hippocampus, within the entorhinal cortex, where parallel pathways dissociate relational structure in the medial entorhinal cortex (MEC) from sensory content in the lateral entorhinal cortex.
- To investigate this in silico, we developed ReGraph, a recurrent dual-stream graph model with biological inductive biases, including retina-driven stream-specialized encoding, dorsal-to-ventral modulation, and dynamic lateral connectivity.
- Trained on the action benchmark Something-Something V2, ReGraph revealed a pathway-specific emergence of relational mapping: context-invariant codes and grid-like spatial bases uniquely co-emerged along the extended dorsal stream.
- Together, our findings provide a computational account that generalization may not be a faculty that emerges abruptly within a dedicated region, but a property that already takes shape as sensory information is parsed into factorized streams of hierarchical visual processing.
Sources (1)
- [1]ReGraph: A Computational Account of Emergent Generalization in the "what" and "where" Dual Visual StreamsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 08:30 AM
Where generalization capacity--the ability to extract context-invariant relational structures--first emerges remains a central question in AI and neuroscience.
The foundation for this capacity lies upstream of the hippocampus, within the entorhinal cortex, where parallel pathways dissociate relational structure in the medial entorhinal cortex (MEC) from sensory content in the lateral entorhinal cortex.
Extractive summary: sentences quoted from the sources.