FR

EN

Homepage Inria website
  • Inria login
  • The Inria's Research Teams produce an annual Activity Report presenting their activities and their results of the year. These reports include the team members, the scientific program, the software developed by the team and the new results of the year. The report also describes the grants, contracts and the activities of dissemination and teaching. Finally, the report gives the list of publications of the year.

  • Legal notice
  • Cookie management
  • Personal data
  • Cookies


Section: New Results

Observing and Modeling Awareness and Expertise During Problem Solving

Participants: Thomas Guntz, Dominique Vaufreydaz, James Crowley, Philippe Dessus, Raffaella Balzarini

Observing and Modelling Competence and Awareness from Eye-gaze and Emotion

We have constructed an instrument for capturing and interpreting multimodal signals of humans engaged in solving challenging problems. Our instrument captures eye gaze, fixations, body postures, and facial expressions signals from humans engaged in interactive tasks on a touch screen. We use a 23 inch Touch-Screen computer, a Kinect 2.0 mounted 35 cm above the screen to observe the subject, a 1080p Webcam for a frontal view, a Tobii Eye-Tracking bar (Pro X2-60 screen-based) and two adjustable USB-LED for lighting condition control. A wooden structure is used to rigidly mount the measuring equipment in order to assure identical sensor placement and orientation for all recordings.

As a pilot study, we observed expert chess players engaged in solving problems of increasing difficulty [Guntz et al 18a]. Our initial hypothesis was that we could directly detect awareness of significant configurations of chess pieces (chunks) from eye-scan and physiological measurements of emotion in reaction to game situation. The pilot experiment demonstrated that this initial hypothesis was overly simplistic.

In order to better understand the phenomena observed in our pilot experiment, we have constructed a model of the cognitive processes involved, using theories from cognitive science and classic (symbolic) artificial intelligence. This model is a very partial description that allows us to ask questions and make predictions to guide future experiments. Our model posits that experts reason with a situation model that is strongly constrained by limits to the number of entities and relations that may be considered at a time. This limitation forces subjects to construct abstract concepts (chunks) to describe game play, in order to explore alternative moves. Expert players retain associations of situations with emotions in long-term memory. The rapid changes in emotion correspond to recognition of previously encountered situations during exploration of the game tree. Recalled emotions guide selection of situation models for reasoning. This hypothesis is in accordance with Damasio's Somatic Marker hypothesis, which posits that emotions guide behavior, particularly when cognitive processes are overloaded [Damasio 91].

Our hypothesis is that the subject uses the evoked emotions to select from the many possible situations for reasoning about moves during orientation and exploration. With this interpretation, the player rapidly considers partial descriptions as situations composed of a limited number of perceived chunks. Recognition of situations from experience evokes emotions that are displayed as face expressions and body posture.

With this hypothesis, valence, arousal and dominance are learned from experience and associated with chess situations in long-term memory to guide reasoning in chess. Dominance corresponds to the degree of experience with the recognized situation. As players gain experience with alternate outcomes for a situation, they become more assured in their ability to spot opportunities and avoid dangers. Valence corresponds to whether the situation is recognized as favorable (providing opportunities) or unfavorable (creating threats). Arousal corresponds to the imminence of a threat or opportunity. A defensive player will give priority to reasoning about unfavorable situations and associated dangers. An aggressive player will seek out high valence situations. All players will give priority to situations that evoke strong arousal. The amount of effort that player will expend exploring a situation can determined by dominance.

In 2019 we will conduct an additional experiment designed confirm and explore this hypothesis. Results will be reported in a journal paper (under preparation) as well as in the doctoral thesis of Thomas Guntz, to be defended in late 2019.

Bibliography

[Damasio 91] Damasio, A., Somatic Markers and the Guidance of Behavior. New York: Oxford University Press. pp. 217–299, 1991.

[Guntz et al. 18a] T. Guntz, R. Balzarini, D. Vaufreydaz, and J.L. Crowley, "Multimodal Observation and Classification of People Engaged in Problem Solving: Application to Chess Players". Multimodal Technologies and Interaction, Vol 2 No. 2, p11, 2018.

[Guntz et al. 18b] T. Guntz, J.L. Crowley, D. Vaufreydaz, R. Balzarini, P. Dessus, The Role of Emotion in Problem Solving: first results from observing chess, Workshop on Modeling Cognitive Processes from Multimodal Data, at the 2018 ACM International Conference on Multimodal Interaction, ICMI 2018, Oct 2018.