Honeybees and some other insects, in learning the sun's course, behave as if they can estimate the sun's position at times of day when they have never seen it, but there are competing ideas about the computational mechanisms underlying this ability. In an approach to this problem, we provided incubator-reared bees with opportunities to fly and see the sun only during the late afternoon. Then, on a cloudy day, we allowed bees to fly for the first time during the morning and early afternoon, and we observed how they oriented their waggle dances to indicate their direction offlight relative to the sun's position. The clouds denied the bees a direct view of celestial orientation cues and thus forced them to estimate the sun's position on the basis of their experience on previous evenings. During the test days, experience-restricted bees behaved during the entire morning as if they expected the sun to be in an approximately stationary position about 180°from the average solar azimuth that they had experienced on previous evenings; then from about local noon onward they used the evening azimuth. This pattern suggests that honeybees are innately informed of the general pattern of solar movement, such that they can generate an internal representation that incorporates spatial and temporal features of the sun's course that they have never directly seen.A wide variety of animals can use the sun's azimuth as a true compass, compensating for its daily movement relative to terrestrial features (1-3). Complicating this task is that the azimuth changes at a variable rate over the day, shifting relatively slowly as the sun rises in the morning or sets in the evening and rapidly as the sun crosses the local meridian at midday. Furthermore, the daily pattern of change, the ephemeris function, varies with season and with latitude. Animals are generally thought to learn the current local ephemeris function, rather than using the average rate of shift of the azimuth. What little is known about this learning process suggests that animals do not merely memorize a series of time-linked solar positions. Instead, in a wide range of invertebrate (4-12) and vertebrate (13-15) taxa, animals have been shown to estimate solar positions that they have never seen. For example, in a classic series of experiments, Lindauer (4, 5) showed that incubator-reared honeybees (Apis mellifera) that were allowed to see only the western half ofthe sun's course (from noon onward each day) could, when they flew for the first time in the morning, use the sun to search for food in a direction in which they had been trained. Thus, somehow they had correctly learned to expect the sun in the eastern half of the sky in the morning. In this paper we ask what integrative mechanism might underlie this apparent ability ofbees to compute solar positions that till gaps in their experience of the sun's course.Four distinct computational strategies have been proposed to account for the ability of insects to estimate unknown positions of the sun. Three of these assu...