Social capabilities across human development | Part 21
An action we expected to help may later turn out to be unwelcome. Social capability appears not only in getting things right initially, but also in receiving a person's response and changing the next action.
A joke expected to be welcome
Someone shows a colleague a light joke, but the colleague's reaction changes. "I don't like that topic," they say. The first person stops rather than explaining why the joke should be acceptable. In the next conversation, they turn to the work topic the colleague wanted.
This is fictional. Wanting to be liked and the actual impact are distinct. "I meant no harm" should not invalidate the information that the person disliked it.
Updating information and changing behavior
Executive-function research organizes capacities including working memory and cognitive flexibility, providing background for holding new information and changing an earlier approach. See Diamond: Executive functions.
That research alone cannot explain every unsuccessful joke or relationship repair. This article uses it to distinguish remembering a correction from acting on it. Adults' self-regulation can also be supported, but that does not mean everyone's interpersonal capability improves identically. See Harvard: Adult Capabilities.
Reflection should lead to observable changes
"Next time I'll be careful" remains vague unless something changes. Stop a particular topic, ask beforehand, or let the other person interrupt. Translate reflection into actions that can be observed.
Avoid fixing someone's preferences forever from one response. "I disliked this here today" differs from "I do not want this topic in future." Where necessary, check the scope of the requested change.
Reflection also differs from endless self-blame. Acknowledge responsibility, then organize what happened, its effects, and the next choices. Preserve the person's response rather than rewriting it for convenience.
A design hypothesis for physical AI
Displaying "Corrected" is insufficient if response generation continues using old information. A possible design retains the correction target, applicable scope, and current version, with both conversation and movement using that version.
Remembered preferences need a purpose, expiry, and methods for correction or deletion. Every reaction need not be stored indefinitely. Check that changes do not extend beyond the person's wishes or authority.
Evaluation could repeat the same scene, test against overapplying a condition elsewhere, and include updated preferences. The proposal examines whether necessary behavior changes correctly rather than whether records accumulate.
A decision in four panels
- A joke is shown and the other person appears uncomfortable.
- They ask for it to stop.
- The speaker puts it aside and listens to their preference.
- A later conversation uses another topic.
This fictional comic depicts updating after a mistake rather than treating one failure as proof of no social capability.
References
About this series
This series uses published literature to examine everyday situations. Its 7 domains and 22 capabilities are an editorial framework, not a standard academic classification. Differences in age or culture are not rankings of people's social worth or capability. The proposed AI designs do not establish humanlike emotional experience or demonstrated product performance. The comics illustrate fictional examples.
LOGIHEART's research and development vision
LOGIHEART is researching how to implement social capabilities in physical AI that interacts with people.
Explore the research and development vision at the LOGIHEART website and in the English introduction video.
Top comments (0)