Stéphane Raffard1, Catherine Bortolon1, Mahdi Khoramshahi2, Robin N Salesse3, Marianna Burca4, Ludovic Marin3, Benoit G Bardy5, Aude Billard2, Valérie Macioce6, Delphine Capdevielle7. 1. Epsylon Laboratory Dynamic of Human Abilities & Health Behaviors, University of Montpellier 3, Montpellier, France; University Department of Adult Psychiatry, Hôpital de la Colombière, CHRU Montpellier, Montpellier University, Montpellier, France. 2. Learning Algorithms and Systems Laboratory, School of Engineering, EPFL, Lausanne, Switzerland. 3. EuroMov, Montpellier University, 700 Avenue du Pic Saint-Loup, 34090 Montpellier, France. 4. Epsylon Laboratory Dynamic of Human Abilities & Health Behaviors, University of Montpellier 3, Montpellier, France. 5. EuroMov, Montpellier University, 700 Avenue du Pic Saint-Loup, 34090 Montpellier, France; Institut Universitaire de France, France. 6. Clinical & Epidemiological Research Unit, CHU, Montpellier, France. 7. University Department of Adult Psychiatry, Hôpital de la Colombière, CHRU Montpellier, Montpellier University, Montpellier, France; INSERM U-1061, Montpellier, France.
Abstract
BACKGROUND: The use of humanoid robots to play a therapeutic role in helping individuals with social disorders such as autism is a newly emerging field, but remains unexplored in schizophrenia. As the ability for robots to convey emotion appear of fundamental importance for human-robot interactions, we aimed to evaluate how schizophrenia patients recognize positive and negative facial emotions displayed by a humanoid robot. METHODS: We included 21 schizophrenia outpatients and 17 healthy participants. In a reaction time task, they were shown photographs of human faces and of a humanoid robot (iCub) expressing either positive or negative emotions, as well as a non-social stimulus. Patients' symptomatology, mind perception, reaction time and number of correct answers were evaluated. RESULTS: Results indicated that patients and controls recognized better and faster the emotional valence of facial expressions expressed by humans than by the robot. Participants were faster when responding to positive compared to negative human faces and inversely were faster for negative compared to positive robot faces. Importantly, participants performed worse when they perceived iCub as being capable of experiencing things (experience subscale of the mind perception questionnaire). In schizophrenia patients, negative correlations emerged between negative symptoms and both robot's and human's negative face accuracy. CONCLUSIONS: Individuals do not respond similarly to human facial emotion and to non-anthropomorphic emotional signals. Humanoid robots have the potential to convey emotions to patients with schizophrenia, but their appearance seems of major importance for human-robot interactions.
BACKGROUND: The use of humanoid robots to play a therapeutic role in helping individuals with social disorders such as autism is a newly emerging field, but remains unexplored in schizophrenia. As the ability for robots to convey emotion appear of fundamental importance for human-robot interactions, we aimed to evaluate how schizophreniapatients recognize positive and negative facial emotions displayed by a humanoid robot. METHODS: We included 21 schizophrenia outpatients and 17 healthy participants. In a reaction time task, they were shown photographs of human faces and of a humanoid robot (iCub) expressing either positive or negative emotions, as well as a non-social stimulus. Patients' symptomatology, mind perception, reaction time and number of correct answers were evaluated. RESULTS: Results indicated that patients and controls recognized better and faster the emotional valence of facial expressions expressed by humans than by the robot. Participants were faster when responding to positive compared to negative human faces and inversely were faster for negative compared to positive robot faces. Importantly, participants performed worse when they perceived iCub as being capable of experiencing things (experience subscale of the mind perception questionnaire). In schizophreniapatients, negative correlations emerged between negative symptoms and both robot's and human's negative face accuracy. CONCLUSIONS: Individuals do not respond similarly to humanfacial emotion and to non-anthropomorphic emotional signals. Humanoid robots have the potential to convey emotions to patients with schizophrenia, but their appearance seems of major importance for human-robot interactions.
Authors: Laura Cohen; Mahdi Khoramshahi; Robin N Salesse; Catherine Bortolon; Piotr Słowiński; Chao Zhai; Krasimira Tsaneva-Atanasova; Mario Di Bernardo; Delphine Capdevielle; Ludovic Marin; Richard C Schmidt; Benoit G Bardy; Aude Billard; Stéphane Raffard Journal: Sci Rep Date: 2017-11-03 Impact factor: 4.379