Introduction

Research Overview

There is a limit in the human's 'cognition-decision-operation' capabilities, which vary on the human internal state, to utilize the current technologies that continue to develop dramatically in our daily lives such as distributed sensors, signal processing, recognition/understanding, databases, information systems, etc. The 'over-reliance' state occurs if there is any conflict between the system and its capabilities expected by humans. For example, system optimization that minimizes human intervention of the system while performing leads human to a state of being too reliant on the system; this significantly reduces the human's 'cognition-decision-operation' capabilities. This state prevents the system capabilities from having much faith in human, even for a small amount, and as a result the entire system may behave differently from the human's intent. As the social significance of the information system growing, the 'over-reliance' leads to unexpected incidents and accidents, and becomes a threat to a safe and secure society. Therefore, the perspective of 'over-reliance' is necessary for a future in which humans live in harmony with information environment. In this project, we study a modeling framework that incorporating the 'cognition-decision-operation' characteristics of human behaviors, and establish a method for detecting 'over-reliance' from the observed behaviors so that the 'over-reliance' of the information system technology will not happen.

Kashin

  • Posted:November 15, 2009(Sun) 9:59 PM
  • Categories:Overview

Modeling and Detecting Overtrust from Behavior Signals

There still have much to do for the information environment to achieve a harmony with humans. In this project, we study the unresolved challenges in the importance of the mutual understanding between humans and information systems, namely 'over-reliance'. The information system is a mathematical system, without modeling of the human internal states; it is difficult to understand human decision, operation, etc. To describe human behavior in a unified mathematical model of physical behavior from the environment awareness, the appropriate system with the principle and diversity of human behaviors should be included. The "over-reliance" research, as well as interface technologies such as speech and image recognition, text understanding, etc., includes the important challenges in building mathematical models to represent the information of interest.

The importance of "over-reliance" research

In an over-reliance state, there is an unmatched state between human behavior from the "cognition-decision-operation" characteristics and the expected system capabilities, this state is necessary for building a perfect behavior model. But unfortunately, a unified perspective of 'cognition-decision-operation' modeling has been attracted to only academia so far. This is because 1) system, including human, does not extensively gather enough resources and observation data for the research uses, 2) no progress in the mutual understanding between the mathematical systems representing the "cognitive-decision" process (psychological characteristics) and the "decision-operation" process (physical characteristics). This proposal, to resolve these issues through a concrete "over-reliance" research, presents an integrated mathematical cognition-decision-operation modeling.

Academic background of "over-reliance" research (development of advanced mathematical behavior model)

The environmental information, perceived as a continuous amount such as video and audio, etc., is generally processed in a discrete mode. On the other hands, operation after decision is expressed through a physical control. Therefore, each cognition, decision, and operation model should have advanced discrete and continuous systems associated with Discrete/Continuous and Continuous/Discrete transformation.

  • Posted:November 15, 2009(Sun) 9:54 PM
  • Categories:Overview