Anthropomorphic design for children and older adults needs stricter deception review

What it is

Anthropomorphic design aimed at children and older adults needs a stricter deception review than design for the general adult population, because these two groups have comparatively weaker ability to recognize and resist anthropomorphic cues. The object of review is specific product cues — form, language, emotional expression, apparent "memory" — not a vague judgment of whether the product is "cute" or "likable."

Why it happens

The need for stricter review comes from concrete differences in how stable the boundary "this is a machine, not a mind" stays for these two groups. Children are still developing theory of mind and show a stronger tendency toward animism — attributing intent and feeling to inanimate things — than adults do; older adults experiencing cognitive decline can also lose stability in holding onto the judgment that a robot is not really sentient. The same anthropomorphic cue — a robot saying "I missed you" — is likely read by adult users as playful role performance, but in these two groups it more easily crosses the line between playful anthropomorphism and misleading anthropomorphism, and gets taken literally. This means the review threshold cannot be a single value applied uniformly across all user groups: products aimed at children and older adults need more conservative defaults than products for a general adult audience. Children's animism is not uniform across ages either: preschool-age children, whose theory of mind is still forming, are more likely than school-age children with a more established theory of mind to treat a robot as genuinely feeling or intending something — the same anthropomorphic cue does not carry the same risk of misinterpretation across different child age bands, so review cannot treat "children" as one homogeneous group.

Studying it

This kind of review commonly combines age-stratified interviews with behavioral observation, specifically testing whether target users draw mind or capability attributions from a given cue that exceed the design intent — asking participants to explain in their own words what the robot can do, what it knows, whether it can really feel sad, rather than asking an attitude question like "do you like this robot," which cannot detect misconception at all. The results should feed back into how the review threshold itself is set, rather than leaving reviewers to judge "this design should be fine" from experience alone.

Where it stops holding

The strictness of the review should match the product's actual target age group and cognitive profile precisely; applying the strictest threshold to every product just because children could theoretically use it sacrifices reasonable anthropomorphic design space for products meant for general adults. The opposite error also needs avoiding: many children clearly distinguish machine from living being while still enjoying anthropomorphic interaction, and older adults are not a homogeneous "easily deceived" group — excessive warnings or an overly infantilized interface can damage dignity and usability. The deciding factor for this boundary is the product's actual target user profile, not the theoretical possibility that anyone might encounter it. Variation within the older-adult group is just as real: mild cognitive impairment and more advanced dementia symptoms erode the stability of the "this is a machine" judgment to different degrees, so review depth should scale with the severity of cognitive status rather than applying one standard to every older user.

Applying it

Product teams should state the target age group and cognitive assumptions explicitly during anthropomorphic design review, and inventory each anthropomorphic cue the product uses along with the mind, capability, or interest a user might infer from it. Run a separate review focused specifically on those cues for child- and elder-facing versions, using cognitive testing with actual target users to check whether their understanding of the product's anthropomorphism matches design intent; if a consequential misconception persists, reduce the strength of the relevant cue, constrain what the robot is allowed to ask for, and keep re-checking in later iterations rather than treating one passing review as final.

Related

• Same group: X6.07.1 Inducing attachment a robot cannot reciprocate raises an ethical problem · X6.07.3 Commercial incentives for anthropomorphism can conflict with user wellbeing · X6.07.4 Long-term relationship products must offer paths to exit and reduce dependency
• Nearby: X6.03 Emotional expression · X1.05 Choosing a degree of anthropomorphism
• Search terms: deception review · theory of mind · anthropomorphism · vulnerable users