Cognitive and Neural Mechanism for the Recognition of Fragmented Visual Information

视觉识别系统有一种被称为视觉插值的能力，可以识别某些物体背后的物体或通过某些物体的孔看到的物体。我们研究了碎片字母视觉插值能力的计算模型和神经基础。提出并验证了最小距离连接模型，该模型假设碎片化视觉信息是基于信息之间的距离来识别的。脑功能成像显示，与完整字母识别相比，碎片字母识别在双侧背外侧、下侧和侧额叶皮质中伴随着更大的激活。实验结果表明，双侧白质变的程度，无论多么小，都与视觉插值能力的下降显著相关。结果表明，我们的方法在早期发现认知障碍方面是可行的。

The visual recognition system has a capability, called visual interpolation ability, to identify objects that exist in part behind some bodies or are seen through apertures in something. We have studied the computational model and the neural basis of the visual interpolation ability using fragmented letters. A least-distance connection model, which hypothesize that the fragmented visual information is recognized based on the distances between the information, was proposed and validated. Functional imaging of the brain showed that that fragmented-letter identification was accompanied by greater activation in the bilateral dorsolateral, inferior, and lateral frontal cortices, compared to complete-letter identification. Experimental results illustrated that the bilateral extent of leukoaraiosis, however small it may be, was significantly associated with the decline of the visual interpolation ability. The results showed the useful possibility of our method for early detection of cognitive impairment.