A central question in neuroscience is how the brain selects relevant actions in response to physiological needs. Across species, the brain monitors physiological variables through peripheral nerve sensors and specialized brain nuclei that detect need-related hormone levels in the bloodstream. While the influence of certain hormones on behavior is well established, how hormones specifically bias neuronal dynamics remains unclear.In this talk, we will first present a simple theoretical framework that allows us to derive spiking neural networks implementing specific low-dimensional dynamics. We will then explore the consequences of implementing dynamics with both fast and slow components, which give rise to neural interactions over multiple timescales. These can be interpreted as fast synaptic interactions and slow hormonal modulatory signals, respectively. Finally, we will discuss how these models can be understood within a broader control-theoretical framework for regulating physiological homeostasis and how they connect to our recent experimental efforts to simultaneously measure hormonal dynamics and neural activity in live animals.We will conclude with an overview of future plans and an outlook on how this research will advance our understanding of physiology-driven decision-making and the embodiment of artificial agents.