The paper presents the results of the MPEG-7 Core Experiment CE-Shape-1, which evaluated the performance of shape descriptors for non-rigid shapes with a single closed contour. The experiment compared various shape descriptors under conditions of scaling, rotation, and non-rigid motion. The shapes used were simple pre-segmented shapes defined by their outer closed contours. The main goal was to assess the robustness of shape descriptors to small non-rigid deformations. The shape descriptors were categorized into three types: contour-based, image-based, and skeleton-based. The contour-based descriptors included P320, P567, and P298. The image-based descriptors included P687 and P517. The skeleton-based descriptor was DAG. The results showed that P298 and P320 performed the best in terms of robustness to scaling and rotation. P298 uses a simplified contour derived from a novel process called discrete curve evolution, while P320 uses curvature scale-space. Both descriptors are based on the best possible correspondence of maximal convex/concave arcs in simplified contours. P687, based on Zernike moments, performed well in terms of theoretical support and experimental verification. P517, based on multilayer eigenvectors, was the only descriptor without existing references in the literature. The results also showed that DAG, based on skeleton-based descriptors, was not robust to scaling or rotation. The best possible performance for part A was about 93%, with all descriptors except DAG performing nearly optimally. In part B, the similarity-based retrieval showed that P298 and P320 significantly outperformed the other descriptors. In part C, the performance of the descriptors was tested under non-rigid motion, with P298 and P320 again showing the best results. The paper concludes that P298 and P320 are the most useful for searching for similar objects obtained by non-rigid transformations. Both descriptors are cognitively motivated, as they are based on the best possible correspondence of maximal convex/concave arcs in simplified contours. The results also show that the cognitive motivation is essential for the computation of similarity values, as verified by human visual perception.The paper presents the results of the MPEG-7 Core Experiment CE-Shape-1, which evaluated the performance of shape descriptors for non-rigid shapes with a single closed contour. The experiment compared various shape descriptors under conditions of scaling, rotation, and non-rigid motion. The shapes used were simple pre-segmented shapes defined by their outer closed contours. The main goal was to assess the robustness of shape descriptors to small non-rigid deformations. The shape descriptors were categorized into three types: contour-based, image-based, and skeleton-based. The contour-based descriptors included P320, P567, and P298. The image-based descriptors included P687 and P517. The skeleton-based descriptor was DAG. The results showed that P298 and P320 performed the best in terms of robustness to scaling and rotation. P298 uses a simplified contour derived from a novel process called discrete curve evolution, while P320 uses curvature scale-space. Both descriptors are based on the best possible correspondence of maximal convex/concave arcs in simplified contours. P687, based on Zernike moments, performed well in terms of theoretical support and experimental verification. P517, based on multilayer eigenvectors, was the only descriptor without existing references in the literature. The results also showed that DAG, based on skeleton-based descriptors, was not robust to scaling or rotation. The best possible performance for part A was about 93%, with all descriptors except DAG performing nearly optimally. In part B, the similarity-based retrieval showed that P298 and P320 significantly outperformed the other descriptors. In part C, the performance of the descriptors was tested under non-rigid motion, with P298 and P320 again showing the best results. The paper concludes that P298 and P320 are the most useful for searching for similar objects obtained by non-rigid transformations. Both descriptors are cognitively motivated, as they are based on the best possible correspondence of maximal convex/concave arcs in simplified contours. The results also show that the cognitive motivation is essential for the computation of similarity values, as verified by human visual perception.