Does frequent backchannel from voice agents help or hurt older adults?

Liu and colleagues (CHI 2025 full paper) find that backchannel signals — "yes," "uh-huh" — produced by their LLM-powered Barge-in voice agent did not uniformly improve older adults' experience of it. The authors state: "Some participants reported that the backchannel ... and active interruptions made by the Barge-in Agent interfered with their speech," and conclude that "excessive backchannel responses can lower users' evaluations of the robot, which is consistent with observations in the field of voice chat."

The authors' explanation is physiological rather than purely a matter of design tuning: aging changes "the central auditory system, including differences in processing times, alterations in neural response patterns, and a slowing down in the response speed to rapid sequences of sound events," which they argue makes it harder for older adults "to quickly process and distinguish between these rapid auditory cues" and "to seamlessly integrate these signals into ongoing conversations without feeling disrupted."

This qualifies the general finding in Do voice agents that interrupt feel more natural to older adults? — the same study that finds two-way interruption and backchannel improve naturalness overall also finds the backchannel component specifically needs to be dosed down for this population, not simply added. It parallels What pacing moves make conversational agents sound like active listeners? and Can pacing strategies make conversational agents feel more human? in treating conversational micro-signals — pacing, backchannel — as calibration problems rather than features that are simply better when present; here the calibration variable is sensory and physiological (auditory processing speed) rather than purely stylistic.

The excerpt reports this as a qualitative theme from participant feedback within the same small, companionship-focused, within-subject study described in the sibling note, not a quantified dose-response relationship between backchannel frequency and rated naturalness. The authors note that response-wait-time and interruption-frequency parameters were customizable but that "uncertainty in network latency and token calculation speed during the study prevented us from fully exploring their effects." The implication the excerpt supports is that backchannel frequency for older-adult voice agents likely needs individual or age-group tuning, not a universal more backchannel is more natural default.