Feasibility Study: Interactive Entertainment of Elder Persons with Intelligent and Emotional Personality Agents

In most listings of the benefits of virtual butlers or other technical companions for elder people one aspect is missing: games. Games can fulfill at least two tasks: First, they can entertain these people, they can amuse them, then people can become happier. Secondly, games can train emotional and cognitive capabilities. In this study we investigated which games could be used, which role an intelligent and emotional agent could play, and how the equipment should function and look in order to be accepted by elderly people.

In a first step we held several conversations with seniors and with gerontologists and we carefully checked the sparse literature on the application of computerized games for elderly people. Since card games play a big role not only in these few games but also in the day-to-day interactions in this group, we decided to choose a memory game. In order to make the game more attractive to people with no prior computer experience, two different versions were developed: A classical memory game and a personalised memory for which the people created pairs of cards, one of which bore the name of a family member or of a friend and the other card the picture of an object which the player associated with this other person.

This second game also allowed the players to recall their own history and narrate it in a interactive way during the game. As a connection was created between long term memory content (the relation between person and object) and short term memory content (the location of the cards) we assumed that this would result in better cognitive training.

A touch table was used as the technical equipment; the emotional software agent was displayed on a vertical screen at the opposite of the touch table so that the player had the impression of playing against the agent. In order not to discourage the players, the speed of the agent was adapted to that of the player. During the game interaction, the agent's emotional facial expressions and utterances simulated human reactions. The emotional personality agent chose cards according to its adaptive AI level memory; emotions shown during the game interactions were dependent on its emotional memory. E.g., the agent became bothered when it could not find a pair that it could have found.

The whole setup was incrementally improved in two pretests steps with small samples of players. In the final version, 18 elderly people (mean age 84) played the classical memory game and the personalised version in a cross design. Of these 18 people, all but one had no experience in computer use. Comments made by the participants during the game interactions were recorded for later analysis. Additionally, the participants filled out a questionnaire about a broad range of aspects, from the agent's perceived playing speed and preferred game types to their characterization of the software agent.

All participants reported that they enjoyed both versions of the memory game. Interviews and observations showed that the majority of participants considered the software agent to be important and amusing and they interacted with it in various ways during the game; only a minority focused more on the cognitive aspects of the memory task. This confirmed the hypothesis that games for a varied target group should better include multiple aspects of social interaction and cognitive training to reach all people. As regards the technical equipment, the touch table was a suitable device for these elderly people. Even the two participants with tremors had no problems playing since large cards were used in the memory games.

As regards evaluation of the emotional personality agent, the characteristics that were ascribed most often to it were intelligent, self-confident, honest, attentive, committed, believable, friendly, active, likeable, and interesting. Observations of the game interactions showed that nearly twice as many comments were made towards the software agent in the game with personal content than in the classical memory game. Furthermore, the game with personal content was perceived as being easier than the classical game, despite its higher complexity. This result, combined with the rating of the memory game with personal content as being more enjoyable for the participants, demonstrated the benefits of including personal aspects in cognitive training.

The study showed that in the interaction with an emotional personality agent in a well-developed game situation nearly all elderly people can spend entertaining, happy hours. For evaluating the improvements in the cognitive capabilities of these people the available time-span was too short; it is recommended that the study be continued with this successful setting in a larger time frame and with a larger group of people.

Research staff