UC Berkeley

Many species negotiate complex spatial environments alongside other individuals and displaythe capacity to recall rewarding and costly locations. Prodigious navigators like the Egyptian fruit bats forage for food in convoluted aerial spaces and must account for the choices of other flying conspecifics to reach reward locations. We can simulate this challenging feat in the lab environment while using high-speed positional tracking and neural recording methods to investigate how bats solve this high-dimensional multi-agent foraging problem. However, in the lab the multi-agent foraging problem has a latent extra player: the human experimenter. While most studies involving a spatial reward task ignore the presence of the human in reporting and controls, we sought to explicitly include the experimenter in the task paradigm to test whether there is an appreciable effect of human presence on ongoing neural signal.

The first section of this thesis explores the neural representation of the human experimenterduring a multi-agent foraging task. We find that a substantial amount of variance in the code for allocentric spatial representation can be accounted for simply by controlling for the identity of the human the bat lands on. This study shows, for the first time, that non-conspecifics are encoded in the canonical place cell code for spatial position. On a practical level, these findings serve as a cautionary tale to encourage future studies to limit the uncontrolled effects of human presence and intervention in behavioral neuroscience tasks.

The second section of this thesis strips down the complex foraging problem to its simplest formto study the stability of the neural code that supports flight in a familiar foraging environment. We find that, across days, the code for flight is largely stable and apparent variance may be accounted for simply by behavioral drift, not changes in the underlying neural code.

The thesis concludes with caveats of the current research, possibilities for future investigation,and avenues for combining modeling work with cross-species representation in the environment.