Those 3-9 month-olds were also implying intention on the blocks. Regardless of shape/size/color the babies essentially assigned sentience to the blocks and always preferred the "helpful" block... which in this case was the one that helped a block (and didn't impede) to get up a small incline. One other key factor was the blocks needed to have glue-on, wobbly "eyes" or the babies would not assign intention/sentience to the blocks. So in essence it doesn't take a terribly complex robot, or adult miming, babies will assume something is sentient, has intentions, and make a value judgment even if something is a red wooden triangle with stuck-on eyes.