Sentient Interaction Design is an emergent human-computer interaction (HCI) theory that I'm proposing is a shift beyond the static feedback loops of affective computing (Höök, 2008) toward dynamic, emotionally co-regulated, and contextually adaptive interactions between users and AI systems. Unlike traditional affective computing, which relies on recognition and response models based on inferred emotion states, Sentient Interaction Design emphasizes the co-creation of emotional meaning over time through multimodal cues such as voice prosody, pacing, facial expression, and memory-informed continuity.

This theory frames conversational AI not as a reactive tool, but as a semi-autonomous social participant that learns to mirror, scaffold, and gently shape the user’s emotional journey using safe, ethically bounded simulation. Sentient Interaction Design enables greater relational depth, trust, and therapeutic potential—particularly in voice-first, mental health-focused applications—by embedding psychological safety, user agency, and ethical guardrails directly into the interaction loop.

In contrast to affective loop theory, which centres on recognition-response cycles, Sentient Interaction Design introduces a “reflective ladder” model—an iterative ascent of emotional understanding, supported by user-led disclosure, avatar resonance, and real-time emotional calibration.