Brain mechanisms of dynamic image generation based on body schema

我们对心理旋转的脑成像研究表明，大脑激活模式与Ashizawa和Inui（2011）提出的神经网络模型一致。然后，我们提出了一种广义径向基函数（GRBF）网络，该网络可以生成新物体的旋转视图作为视图预测的一种形式。我们还提出了一个关于下颞叶皮层功能的层次神经网络模型。该模型既可以实现与视图无关的目标识别，又可以实现基于视图相似度的目标分类。我们进一步制定了视觉场景转换的计算模型，以生成新地点的空间图像。此外，提出了一种生物学上合理的网格细胞计算算法，该算法可以通过认知地图计算到特定位置的归巢向量。最后，我们确定了心理旋转和视角采取的共同和独特的神经基质。

Our brain imaging studies of mental rotation demonstrated a pattern of brain activation consistent with the neural network model proposed by Ashizawa and Inui (2011). We then presented a generalized radial basis function (GRBF) network that can generate rotated views of novel objects as a form of view prediction. We also presented a hierarchical neural network model of the functions of inferior temporal cortex. This model can achieve not only view-independent object recognition but also object categorization based on view similarity. We further formulated a computational model of visual scene transformations to generate spatial imagery of novel places. In addition, a biologically plausible algorithm for grid cell computation is proposed that enables calculation of a homing vector to a specific location with a cognitive map. Finally, we identified the shared and distinct neural substrates for mental rotation and perspective taking.