Recommendation systems are prevalent on the Internet but are prone to feedback loops that cause ‘echo chamber’ effects. We present an allostatic regulator for recommendation systems based on opponent process theory and behavioral posology principles to combat these effects. When applied as a code wrapper to a supervised K-Nearest Neighbors algorithm for movie recommendations, our prototype algorithm can dynamically restrict the proportion of potentially harmful content recommended to users. This technique is adaptable to other domains and is scalable to more complex machine learning algorithms with minimal changes to their internal parameters. Combining allostatic regulation with insights from human-derived models on the healthy limits of digital content consumption may provide app developers with a flexible tool to help users regulate their online experiences. In turn, allostatic regulation moderates positive feedback loops, potentially reducing echo chamber effects while increasing the transparency and interpretability of machine learning models.