MIT Media Lab: Affective Computing Group

Home People Projects Publications Policy Press FAQ Vintage

Policy for Research in Affective Computing at the MIT Media Lab

Statement of Policy for Research Experiments

For all participants that are involved in our research experiments and studies, we adopt the following policies:

We require that all experiments involving humans receive approval by the MIT Committee on the Use of Humans as Experimental Subjects, prior to the running of the experiments.
The subject of the experiment maintains absolute authority over the release of any information that in any way breaches his or her anonymity.
Studies may involve participants who, in some cases, are asked to deliberately create and/or express an emotion; in other cases, the administrators of an experiment may attempt to evoke the emotions of participants via passive stimuli, such as films, or more actively, for example by engaging participants in activities that are predicted to evoke certain emotional responses. In all cases, participants are willing, informed and have the right to withdraw their consent at any time.
Each participant in our experiments is given the option of being alerted to information an experiment may reveal about the participant. They can also contact us to have their data withdrawn and deleted at any time.

This policy is in accordance with the relevant federal guidelines concerning human subjects research:

Statement of Policy for System Development

Affective computing systems detect, process, recognize, and communicate information that a user may feel is private. We pledge to use any reasonable means at our disposal to ensure the privacy and empowerment of users of affective computing systems. This includes encryption, hardware/software firewalls, real-time algorithm development without storage of sensitive information, separation of identifying information from affective information, communication and sharing protocols that require informed consent, and privacy evaluations of affective systems.

Research on Ethical Issues related to Affective Computing

When humans observe and communicate emotions to one another, they are often bound by ethical and social constraints. The extension of affective capabilities to computers raises important moral and ethical questions. Consequently, we are actively working to anticipate harmful or malicious ramifications of our work, (as well as the potential benefits) and develop means of maximizing the potential for beneficial outcomes, while minimizing that for potential harm. Near-term concerns for systems and research involving affective computing include issues of privacy, autonomy, calmness, and informed consent. In the longer term, there are deep questions about the virtue of providing emotional capabilities for artificial intelligence. We take a perspective that seeks a mutualistic relationship between human and artificial intelligence (AI), but one where AI remains subordinate to and in service of human values.

R. W. Picard, Affective Computing ( MIT Press, 1997), Chapter 4, "Potential Concerns".
R. W. Picard, Jonathan Klein (2002), "Computers that Recognise and Respond to User Emotion: Theoretical and Practical Implications", Interacting with Computers, 14, 2 (2002), 141-169. TR 538
C. Reynolds and R. W. Picard (2004), "Affective Sensors, Privacy, and Ethical Contracts," Proceedings of CHI, Vienna, April 2004. PDF
Friedman, B. and Kahn, P. H., Jr. (2002) "Human values, ethics, and design". In J. Jacko and A. Sears (eds.), Handbook of Human-Computer Interaction. Lawrence Erlbaum Associates, Mahwah NJ.