Converging approaches to extracting structure from motion: Psychophysical and computational investigations of recovering connectivity from moving point‑light displays