We recognize that the dexterity shown with the ball is a useful basic ability, but the transition to manipulating an apple is not a simple specialization. There is about a 60% chance that the grasping requirements for an apple, with its irregular shape and fragile skin, are more complex than those of a uniform ball. For example, a robotic claw that grips well a ball may not necessarily pick a Gala apple without damaging it if it is not specifically calibrated for its texture. The idea that this hierarchical progression is stable is therefore probably too optimistic without sensor and algorithm adjustments.