Talk about robots in long-term care always makes me uneasy. It’s partly suspicion that this is just another way to avoid dealing with the growing direct-care workforce crisis: look, Ma, no workforce! Crisis solved.

But it’s also a reaction to how robot engineers tend to look at direct-care work. Like most people who’ve never done it, they seem to see it as just a series of tasks — lifting, transferring, etc. – that a machine can do at least as well as a person. In fact, most long-term care recipients need more interaction with other living beings, not less. Can a robot provide companionship and comfort? Can a robot form a meaningful relationship?

According to an article in the current issue of the Journal of the American Medical Directors Association (free to subscribers only), the answer to those questions could be “Yes.” For the study, Saint Louis University researchers set up pet therapy visits to nursing homes by two dogs: a real one named Sparky and a robot named AIBO. (Pictured above, in a photo by Grant Crawley, is resident Gladys Moore playing with AIBO.) It took the residents longer to warm up to AIBO than to Sparky, but after eight weeks of visits he did almost as good a job of relieving loneliness and fostering attachments in the people he visited.

The robot dog wags his tail, makes noises, and blinks lights when patted or talked to. In an Associated Press story about the article, study coauthor Dr. William Banks calls AIBO “an engaging sort of guy” and says the study could mean that robots could substitute for living dogs in helping people.

The AP article also quotes a professor of computer science and human-computer interaction at Carnegie Mellon University who has her doubts, noting that it may have been the human who bought the robotic dog into the room who reduced the loneliness, not the dog itself.