On the perceptual identity of depth vision in the owl

The key element to testing stereopsis in animals is the stereoscopic displaying of computer generated images that contain no monocular cues, even though an animal might try various strategies, such as closing one eye or making rapid head movements to obtain parallax information. A display devoid of these cues can be provided by Julesz random-dot stereograms (RDSs) in which the disparity information is camouflaged by a random matrix of thousands of minute dots. In chapter 2, a separate section is enclosed to provide information about the method used to construct stereograms as well as methods needed to study motion-parallax. Basically, these methods involved the use of static or dynamic random-dot patterns that could be transformed either autonomously, or with every head-movement by use of a head-tracking device. The experiments described in chapter 3 have provided new information about visual perception in the owl. First, a behaviour shaping technique was developed that prepared owls to be visually tested by images presented on a distant CRT-monitor. Second, it was demonstrated that owls can abstract the figure-ground organisation of computer generated random-dot images by testing two functions of vision that are thought to be involved in figure-ground segregation: (1) texture perception and (2) motion perception. Having established that random-dot patterns can be used to test visual capabilities in barn owls, used was the technique of time-multiplexed stereoscopic displaying to test whether binocular vision in owls enables stereopsis. Considerations in relation with sensory ecology have led to the claim that there is a high degree of similarity between the avian and mammalian neuronal algorithm essential for stereopsis. That is, the adopted strategy avoids feature extraction until information from both eyes has converged. The latter is known as global stereopsis and can be demonstrated by the use of random-dot stereograms. While comparative psychology has provided evidence for global stereopsis in monkeys, cats, horses and falcons, it has failed to specifically demonstrate relative depth based on binocular cues alone. This basic function of stereopsis was investigated (Chapter 4) by examining discrimination-transfer. It is shown that the barn owl possesses stereopsis because this animal is able to determine the relative position of surfaces in RDSs. In Chapter 5, it was hypothesized that when one specifically addresses those aspects of the owl visual system, which set it apart from binocularity in primates, it should be possible to determine the functional significance of stereopsis in the owl. The data show that stereopsis is a robust and highly sensitive function of the owl visual system. Moreover, it is argued that binocularity in the owl must be seen as an adaptation to nocturnal life. While stereopsis has been demonstrated in a wide range of different vertebrate species, strong evidence of animals exploiting motion parallax as an independent monocular depth cue has been found only in insects. It has been stated, however, that there are considerable similarities in the underlying computational theory of depth perception based on either stereopsis or motion-parallax. If this assumption is correct then this should be reflected in the results of empirical studies. The experiments described in Chapter 6 clearly show that owls trained to categorise RDSs can perceive similar depth impressions from motion parallax displays without the need of additional training. This finding provides the most direct evidence in animals that perceptual knowledge derived through either stereopsis or motion parallax is closely related. Finally, Chapter 7, the above described experiments are discussed by reviewing the main findings, and by proposing new experiments to further enhance our understanding of the perception of depth