The integration of room views

Rooms cannot be experienced within a single view as humans cannot look backwards. Nevertheless, humans are able to form an understanding of the whole room. We were interested how and under which conditions integrated room representations are formed based on which long-term memory structure. Participants experienced views of a rectangular virtual room from its center through a head mounted display. Afterwards, they saw a room view and indicated the direction of a non-visible room object using the arrow keys of a keyboard. Participants responded quicker for the first view encountered than for later experienced room views. This pattern did not change when participants rotated physically during learning or only visually. These results indicate that participants did not update experienced room views during learning to memorize integrated room information and are therefore inconsistent with integrating cognitive map parts via path integration [1]. The results are consistent with memorizing separate room views and the transitions between them [2] as well as with memorizing an integrated room memory in a reference frame oriented along the first experienced room view [3]. Our data cannot clearly separate between the two possibilities suggesting that both strategies might have taken place to some degree. The model best fitting with the data suggests that integrating within a single reference frame most often occurred when participants could look around in a self-determined sequence as long as they wanted in continuously changing perspectives. Contrary, when the sequence of views was pre-determined participants most often relied on a sequence of stored views. In sum, results indicate, firstly, that humans do not necessarily integrate experienced room views during learning, even they know that they have to act on an integrated room representation afterwards. Secondly, the first experienced room view acts as an anchor later experienced views are related towards. Thirdly, spatial long-term memory formation seems rather independent from updating the spatial surrounding in working memory during learning