In this paper, I examine the way humans interact with domestic companion animals, with a focus on ‘positive reward-based training’ methods, particularly for dogs. From a biosemiotic perspective, I discuss the role of animal training in today’s society and examine what binary reward- based reinforcement schedules communicate, semiotically. I also examine the extent to which reward-based training methods promote better welfare, when compared to the more traditional methods which rely on aversive stimuli and punishment, if and when they are relied upon excessively. I conclude that when used as the primary means of communication, they have the potential to be detrimental to animal welfare, because the underlying social signal is control and resource dominance. As an alternative view to behaviourist-based learning theory and conditioning, I outline how enactivist theories of cognition support a semiotic approach to interspecific human-animal communication. I therefore propose a move toward a dynamic semiosis and mutual understanding based upon Peirce’s phenomenology, resulting in a more balanced merging of Umwelten. The aim is to create rich and more complex semiospheres around humans and domestic animals, which allow for individual agency and autonomy.