Theory of Mind for a Humanoid Robot
Autonomous Robots
Artificial Intelligence: A Modern Approach
Artificial Intelligence: A Modern Approach
Proceedings of the 9th international conference on Multimodal interfaces
Joint attention and language evolution
Connection Science - Social Learning in Embodied Agents
Three way relationship of human-robot interaction
HCI'07 Proceedings of the 12th international conference on Human-computer interaction: intelligent multimodal interaction environments
| Hi-index | 0.00 |
This paper presents a novel probabilistic approach to joint attention. Joint attention is a communicative activity that allows communicators to share perceptual experience by attending on the same visual object. This communicative activity is conceptualized as a conditional probability over jointly given test and cue stimuli. To formalize the joint attention with mathematical terms, our approach starts from a simple decision task in which a response of subjects is determined by a test stimulus only. Our approach extends to an attentional cueing task in which subjects make a decision on a test that are jointly given with a cue stimuli. The joint relationship between test and cue stimuli yields attentional cueing effects -- faster and more accurate response if the two stimuli are consistent and slower and less accurate response if not. With our model, a series of simulations were carried out to show interesting properties of the model that can not be captured using a test stimulus alone. The model successfully locates a visual object guided by a cue stimulus such as color and pointing gesture. These results indicate that joint attention can be considered as a cooperative decision process on a visual object among many objects with a referential cue driven from a communicator.