Abstract
A laboratory-based setup was developed that allowed us to train bumblebees to a rewarding object (local cue) in the context of a panorama and simultaneously to record extracellularly the neural activity of units interpreted as mushroom body extrinsic neurons (MBENs). This setting simulates conditions flower visiting bees face when foraging excluding the compass related conditions. The bumblebees navigated between the colony and the test/feeding arena on their own motivation, starting from a fully functional colony with a queen. They learned by sucrose reward a specific feeding place marked by the rewarding object local cue (blue pyramid cylinder) placed in a fixed spatial relation to the panorama (matching panorama). Spontaneous choice tests and extinction tests revealed enhanced search behavior for the learned matching condition. Single or multiple (up to three) simultaneously recorded neurons referred to here as units revealed learning-related differences either by increased, decreased or unchanged neural activity. We focused on tests in which the local cue was presented in the same area as the panorama (learned condition: matched) or in separate areas (mismatched condition). Neural activity differed in these two test conditions. This learning effect did not depend on the areas tested, indicating that the whole arena had been learned as an environment in which the local cue and the panorama were spatially related. No changes in neural activity were found for head direction toward the goal (local cue, panorama, matched or mismatched), distance to the goals or the gateway. No place-cell-like activity patterns were found. Significant neural effects were also seen in some units during multiple walks between local cue and panorama in mismatch tests.