Affective Computing

Phase
Technology Trigger
UMN Assessment
No Investment Needed
Time Frame
5 to 10 Years
Last Updated on Jul 20, 2014

Affective computing technologies sense the emotional state of a user by measuring such indicators as changes in heart rate and galvanic skin response (GSR), body temperature, posture and gestures, verbal content, the force or rhythm of one's keystrokes, facial expressions—ideally, some combination of these—and using computer software to analyze these indicators to estimate the user's emotional state and then provide an appropriate response.

Affective computing may prove particularly valuable in supporting distance education. In the face-to-face classroom, instructors can read and respond to the reactions of learners. In an online learning interaction between a human and a computer, the goal is to be able to determine what the learner needs to remain engaged and to enhance the learning experience: more in-depth explanation, review of a previous topic, more challenging material, etc. Achieving this goal in the online environment requires a multidisciplinary approach, involving such fields as computer science, psychology, cognitive science, design, and others.

These ideas aren't new. In 1995, Rosalind Picard authored a paper titled "Affective Computing" which suggested both models for computer recognition of emotion and practical applications for it. Picard now directs Affective Computing research at MIT, which has numerous projects exploring affective computing and its applications. One challenge has been the development of sensors that are accurate and unobtrusive; another is the development of algorithms that can accurately assign emotional states and respond accordingly. In addition, more effort seems to have been directed to commercial applications, such as affdex for marketing research, as opposed to educational purposes.

It is likely that continued research in affective computing eventually will lead to effective online computer-based learning environments that will respond seamlessly to learners' needs. However, at present, even if excellent algorithms were available, the equipment required to achieve good results is not yet cost-effective for widespread use.

Resources

  • 07/20/14

    Related Research

    Ben Cowley, Martino Fantato, Charlene Jennett, Martin Ruskov, and Niklas Ravaja.