This study presents a cybernetic model of emotions based on Michael Apter’s reversal theory, which dynamically links emotions and motivations. Unlike classical descriptive approaches, this theory introduces a systemic and interactive structure built upon pairs of opposing mental states, whose frequent transitions account for emotional instability. The core concept of emotivation captures the interdependence between emotional states and motivational dynamics. The theoretical framework distinguishes two experiential domains—somatic and transactional—each producing positive or negative emotions depending on whether the underlying motivations are satisfied or frustrated. Three support strategies are modeled accordingly: adjustment, reversal, and refocusing. The proposed cybernetic approach relies on a formal representation of these mechanisms: emotions are modeled as information flows regulated by feedback loops that enable real-time adjustment. These loops dynamically tune motivational parameters to restore emotional balance in response to contextual disruptions. This model is implemented in SABER, a conversational agent embedded in a companion robot. Designed to support elderly individuals experiencing psychological distress, SABER detects emotions and their causes, converts them into emotivations, and guides users towards beneficial actions through adaptive dialogue, supported by APIs for emotion analysis and language generation. The results obtained validate the relevance of the model for effective emotional regulation.