Zero-shot counting with a dual-stream neural network model

• DOI: 10.1016/j.neuron.2024.10.008
• Publication Date: 2024-11-01T14:38:06Z
• Authors: Jessica A.F. Thom...; Hannah Sheahan; Tsvetomira Dumbalska; Julian D. Sandbrink; Manuela Piazza; Christopher Summe...
• Keywords: Zero (linguistics); One shot

To understand a visual scene, observers need to both recognize objects and encode relational structure. For example, scene comprising three apples requires the observer concepts of "apple" "three." In primate brain, these functions rely on dual (ventral dorsal) processing streams. Object recognition in primates has been successfully modeled with deep neural networks, but how structure (including numerosity) is encoded remains poorly understood. Here, we built learning model, based dual-stream architecture which able count items "zero-shot"-even if themselves are unfamiliar. Our network forms spatial response fields lognormal number codes that resemble those observed macaque posterior parietal cortex. The also makes successful predictions about human counting behavior. results provide evidence for an enactive theory role cortex understanding.