The second objection to the assumption that the trial-and-error method could have led to the establishment of magical and religious habits arises from the inability of animals to act towards unperceived objects as if they were actually present. A dog never welcomes by gambols or licks the hand of an absent friend, while Religion, and at times Magic, show primitive man in more or less systematic relations with powers he has never sensed. When the Shaman draws lines upon the sand, describes various curves with his arms, utters sundry incantations, he does not address a power he perceives, nor even one he has really seen, although he may believe that he, or some one else, has seen it. That animals are moved to action by memories of past perceptions, is, of course, not open to doubt. Their whole life is a long testimony to that ability. Any one will recall instances of chains of concerted actions indicating clearly, on the part of some one of the higher animals, domesticated or wild, the anticipation of a particular person, object, or event. What they never do, is to behave as if the remembered object was really present, though not sensed. H. Spencer, discussing adversely A. Comte's opinion that fetichistic conceptions are formed by the higher animals, relates the following observation concerning a retriever who had learned for herself to perform an 'act of propitiation.' She had associated the fetching of game with the pleasure of the person to whom she brought it, and so, 'after wagging her tail and grinning, she would perform this act of propitiation as nearly as practicable in the absence of a dead bird. Seeking about, she would pick up a dead leaf, a bit of paper, a twig, or other small object, and would bring it with renewed manifestations of friendliness. Some kindred state of mind it is which, I believe, prompts the savage to certain fetichistic observances.'[12] So far the dog could go, but she could not have imagined the presence of an unseen being and behaved towards him in the same manner. Another significant point is that the absent objects towards which animals may direct their actions are always, so far as one may judge, identical with those actually sensed by them at some time, _i.e._ their behaviour never shows that they have transformed, imaginatively, objects with which their senses have made them familiar. Whereas man can not only believe in the presence of unseen objects, but he can also imagine beings never actually sensed by him, and behave towards them according to the traits and capacities with which he has endowed them.