Table of Contents
Dr. Hiroko Dodge is a NewDays Clinical Advisory Board Member and Professor of Neurology, Massachusetts General Hospital / Harvard Medical School
Q: My mother has mild cognitive impairment. She's still very much herself, but conversations have gotten harder — she loses the thread, or we end up circling the same few subjects. Is that something that can actually get better, or do we just adapt to it?
This question matters more than it's usually given credit for. When researchers ask older adults living with MCI or dementia what they most want to hold on to, being able to talk with the people they love comes up again and again — often ahead of the abilities clinicians spend the most time measuring. So the honest answer to "does this improve?" is one families deserve to have.
We recently published a study that speaks to it directly.
Our team looked at conversations recorded during I-CONECT, a randomized controlled trial in which socially isolated adults aged 75 and older had 30-minute video chats with trained conversation partners four times a week for six months. Because every session was recorded, we could do something unusual: instead of measuring cognition with a test, we measured it in ordinary conversation.
We used a linguistic marker called semantic noise, which describes how a conversation moves across topics. At the start of the trial, participants with MCI had a measurably different pattern from participants with normal cognition. Over conversations of the same length, they ranged across fewer distinct subjects — a narrower, more repetitive path through a conversation. It is, more or less, a number for the thing you're describing.
After six months of frequent conversation, that difference was gone. The MCI group's pattern changed significantly, while the cognitively normal group's stayed steady, and by month six the two groups were no longer statistically distinguishable. Following the natural flow of a conversation appeared to have become easier.
I want to be careful about what this does and doesn't show. It's a proof-of-concept study of 52 participants, and we did not have conversation recordings from the trial's control group — so it demonstrates change in people who received the intervention rather than proving conversation alone caused it. It needs replication in larger and more diverse groups.
But it points at something I think families should hear. Conversation isn't only a pleasant way to pass an afternoon with your mother. For the skills you’re worried about her losing, conversation is the exercise. Keep talking with her - not just about the weather or the things she always talks about, but about things she hasn’t talked about lately. Those wandering, unexpected, off-topic conversations help her maintain the skills needed to communicate with others while providing some of the most valuable cognitive stimulation of all.
Source: Chen L, Wu C-Y, Dodge HH. Changes in conversational pattern as a clinical trial outcome: a proof-of-concept study using the I-CONECT data. Innovation in Aging. 2026;10(8):igag041. doi.org/10.1093/geroni/igag041
Have a question for our clinical team? Email us at hello@newdays.ai.
