Silence, delay, and withdrawal are interpreted socially even when they arise from technical limits. They need to be designed, explained, and made reversible.
Relevant outcomes include the distribution of talk and initiative, human agency, handoff and repair, accountability, and what people retain after the interaction.
Making attention, uncertainty, and intent visible through gaze, orientation, or explanation helps people understand and repair an interaction.
What counts as well-timed involvement depends on culture, institution, social role, and power. Participation policies should remain contestable and adjustable.
Psychology offers behavioral, attentional, implicit, and model-based measures that can complement self-report and question whether engagement is the right objective.
Robots should affirm rather than judge, support rather than replace human experience, and contribute to the communities in which they participate.
Work on gaze-contingent grounding, embodied reference, group participation, and telepresence showed how visible attention can support correction and coordination.
Conditional participation includes when a robot enters, how strongly it participates, whom it addresses, when it withdraws, and how it returns and repairs.
Robots need architectures that can interrupt, acknowledge new events, balance planned and reactive action, and resume coherently after a pause.
Laboratory evidence and sidewalk-robot deployments both pointed toward limited response as a way to preserve trust, agency, and future cooperation.
Separate perception, decision, expression, handoff, and repair to identify which element produces an effect.
Move toward multi-party and sustained settings where withdrawal and re-entry can be observed.
Capture attention, agency, capability, accountability, and the human relationships that remain afterward.