The impact of perceived human and computer interaction on collaborative learning: insights from the Map Task

Learning new skills and information becomes more challenging as people get older, but learning collaboratively can provide benefits to learning in older age. Prior work has predominantly focused on improved learning in terms of accuracy and efficiency when interacting with another human compared to learning alone, while the potential benefits of a human learning with a computer are understudied. It has been reported that older adults learn more efficiently and accurately when they believe they are interacting with a human compared to a computer to perform the Barrier Task. In the current study, we examined the effect of agency on performance on the Map Task, a test that measures spatial cognition, where participants must orally collaborate to reproduce a route printed on one participant's map onto another's, and which is considered a more ecologically valid measure of real-life communication compared to other materials used to study collaborative learning. We used a Wizard-of-Oz paradigm, in which participants interacted with a computer system, but were told that they were interacting with either a human assistant or computer. Twenty-four older adult participants (aged 66-79 years) completed an adapted version of the Map Task with a (perceived) human partner or computer to plan and execute a route, doing so through conversation. Overall, participants engaged similarly with their human and computer partner, but recall was more accurate when the information had been provided by the human, compared to the computer. Participants' negative perceptions about the computer-provided information may have driven this effect. These findings may influence how voice-based learning systems are introduced to older end-users, and strategies for negotiating agency between the human user and the voice-based system, to influence collaborative learning in older age. Future work should attempt to replicate these findings in a larger sample.