At all scales, the movement patterns of organisms serve as dynamic read-outs of their behaviour and physiology. We devised a novel droplet microfluidics assay to encapsulate single algal microswimmers inside closed arenas, and comprehensively studied their roaming behaviour subject to a large number of environmental stimuli. We compared two model species, Chlamydomonas reinhardtii (freshwater alga, 2 cilia), and Pyramimonas octopus (marine alga, 8 cilia), and detailed their highly-stereotyped behaviours and the emergence of a trio of macroscopic swimming states (smooth-forward, quiescent, tumbling or excitable backward). Harnessing ultralong timeseries statistics, we reconstructed the species-dependent reaction network that underlies the choice of locomotor behaviour in these aneural organisms, and discovered the presence of macroscopic non-equilibrium probability fluxes in these active systems. We also revealed for the first time how microswimmer motility changes instantaneously when a chemical is added to their microhabitat, by inducing deterministic fusion between paired droplets - one containing a trapped cell, and the other, a pharmacological agent that perturbs cellular excitability. By coupling single-cell entrapment with unprecedented tracking resolution, speed and duration, our approach offers unique and potent opportunities for diagnostics, drug-screening, and for querying the genetic basis of micro-organismal behaviour.