Brains and Muscles: Learning about Facial Expressions and Gestures using EMG and EEG Measurements

 

Special Session at the 2008 IEEE International Conference on Automatic Face and Gesture Recognition (FG 2008),

September 17-19, 2008, Amsterdam, the Netherlands

 

Organizers:

Anton Nijholt (University of Twente, Enschede, the Netherlands) and

Ioannis Patras (Queen Mary, University of London, UK)

 

Traditionally, the methodologies that have dominated the Face and Gesture Recognition conference rely primarily on visual information. In this session, we explore how neurological signals, and in particular EEG, MEG and EMG measurements of brain and muscle activity, can be utilized for the analysis of facial and body expressions and gestures. The presentations will cover the analysis of neurological signals that are produced both during the generation/production and during the perception of facial expressions and gestures. This will include popular (for the FG audience) topics such as the analysis of emotional facial expressions and body gestures, and the recognition of spontaneous vs posed facial expressions based on EEG and facial EMG. Among the topics that are discussed are MEG measures of facial expressions, emotions and empathy, EMG and EEG response patterns to exposure of facial expressions (facial mimicry). The audience will be introduced to the challenges of neurological signal analysis, but in most cases will be familiar with the signal processing and pattern recognition methodologies that are employed.

 

During a panel session, discussion will be encouraged on multidisciplinary approaches for facial and body gesture analysis and on possible applications that can benefit from a multimodal analysis (and in particular in vision-based and neurological signal - based analysis).

 

Invited Presentations

Emotional Contagion for Unseen Bodily Expressions: Evidence from Facial EMG

Marco Tamietto1,2 and Beatrice de Gelder1

1Cognitive and Affective Neuroscience Laboratory, Tilburg University, The Netherlands, b.degelder@uvt.nl

2Department of Psychology, University of Torino, Italy, M.Tamietto@uvt.nl

 

Inverse mapping the neuronal correlates of facial expression processing

Philippe Schyns, University of Glasgow, p.schyns@psy.gla.ac.uk

Lucy S. Petro, University of Glasgow, lucy@psy.gla.ac.uk

Marie L. Smith, University of Glasgow, m.smith@psy.gla.ac.uk

 

Investigating the production of facial expressions: an electroencephalographic (EEG) and electromyographic (EMG) approach

Sebastian Korb, Swiss Center for Affective Sciences and Department of Psychology, University of Geneva, Switzerland, sebastian.korb@pse.unige.ch

Didier Grandjean, Swiss Center for Affective Sciences and Department of Psychology, University of Geneva, Switzerland, Didier.Grandjean@pse.unige.ch

Klaus.Scherer, Swiss Center for Affective Sciences and Department of Psychology, University of Geneva, Switzerland, Klaus.Scherer@pse.unige.ch

 

Training Computer Vision Systems with Implicit Brain Processing

Ashish Kapoor, Microsoft Research, Redmond, USA, akapoor@microsoft.com

Desney Tan, Microsoft Research, Redmond, USA, desney@microsoft.com

Pradeep Shenoy, University of Washington, Seattle, USA,  psenoy@cs.washington.edu

Eric Horvitz, Microsoft Research, Redmond, USA, Horvitz@microsoft.com

 

Panel Discussion

 

Moderated by the organizers, the three speakers, the organizers and the audience will discuss the research presented in this special session and in particular, how EEG and EMG research approaches can add to existing knowledge about (audio-visual) detection of affective states from prosody, facial expressions and body postures.

 

 

Paper Abstracts and Authors Information

 

Emotional Contagion for Unseen Bodily Expressions: Evidence from Facial EMG

Marco Tamietto1,2 and Beatrice de Gelder1

1Cognitive and Affective Neuroscience Laboratory, Tilburg University, The Netherlands, b.degelder@uvt.nl

2Department of Psychology, University of Torino, Italy, M.Tamietto@uvt.nl

 

Emotional contagion refers to the tendency to automatically mimic and synchronize our facial expressions with those of another person. Recent EMG studies have shown that emotionally-congruent expressive reactions in the observer’s face may be also elicited by the perception of bodily expressions, thus challenging the view that emotional contagion is simply due to motor imitation based on conscious visual recognition. Here we investigated whether emotional contagion may be triggered by bodily expressions that cannot be consciously perceived. Facial EMG was recorded in response to the presentation of backwardly masked happy and fearful bodily expressions. The subjects reacted with emotionally congruent facial expressions (i.e., greate zygomaticus major activity for happy expressions, and greater corrugator supercilli activity for fearful expressions), despite the fact that they were unable to consciously detect the triggering body stimuli. The present findings suggest that expressive facial reactions may unfold as an automatic response driven by the activation of emotion specific affect programs that are independent from conscious visual recognition.

 

Marco Tamietto is a neuropsychologist and holds a PhD in neuroscience. His main research interest is on the neuro-functional bases of non-conscious emotional processing in neurological patients with blindsight and spatial neglect. He has been recently awarded with a Veni grant from NWO for a research project focusing on the neurobiological correlates of conscious and non-conscious emotional communication that will be pursued in collaboration with Bea de Gelder at Tilburg University.

 

Bea de Gelder’s research group (http://www.beatricedegelder.com/) focuses on cognitive and affective neuroscience of intersensory perception, between different sensory systems, primarily on the interaction between seeing and hearing and on how emotion and cognition interact in humans. Behavioural and neurofunctional approaches (ERPs, fMRI, MEG and TMS) are used in an integrated fashion.

 

 

Inverse mapping the neuronal correlates of facial expression processing

Philippe Schyns, Lucy S. Petro and Marie L. Smith, University of Glasgow, m.smith@psy.gla.ac.uk & p.schyns@psy.gla.ac.uk

 

The brain computations that underlie our ability to recognize facial expressions involve the extraction of relevant information from the faces of our peers, and allow us to readily respond in an appropriate manner to the displayed emotion. Here we present recent advances in understanding the brain processes underlying the categorization of facial expressions of emotion using classification image techniques to link both the brain dynamics (EEG) and the behavioural strategies of three observers with specific facial features.

 

Philippe Schyns' research deals with applications of reverse correlation methods to understand the information processing functions of brain signals (EEG, MEG and fMRI).  The current emphasis is on face processing.

 

Marie L. Smith’s research interests include the extension of her work into other neuroimaging modalities (e.g. MEG, fMRI), the application of new analytical techniques (e.g. ICA) to the interpretation of brain signal measurements and the study of the brain's response to facial expressions of emotion.

 

Investigating the production of facial expressions: an electroencephalographic (EEG) and electromyographic (EMG) approach

Sebastian Korb, Swiss Center for Affective Sciences and Department of Psychology, University of Geneva, Switzerland, sebastian.korb@pse.unige.ch

Didier Grandjean, Swiss Center for Affective Sciences and Department of Psychology, University of Geneva, Switzerland, Didier.Grandjean@pse.unige.ch

Klaus.Scherer, Swiss Center for Affective Sciences and Department of Psychology, University of Geneva, Switzerland, Klaus.Scherer@pse.unige.ch

 

Facial expressions are part of emotional reactions. However, humans can voluntarily pose a specific emotional expression without having the corresponding underlying feeling, or voluntarily modify (e.g. reduce or enhance) their spontaneous expression in reaction to an emotional event. Few studies have attempted to distinguish these different processes at the level of the central nervous system (CNS), even though spontaneous and voluntary facial expressions are long thought to rely upon distinct neural circuitries. Here, we review the neural bases of spontaneous and voluntary facial expressions, report the results of a first study assessing the Bereitschaftspotential (BP) before voluntary smiles, and outline a combined EEG/EMG approach for investigating facial expressions at the level of the CNS.

 

Since September 2006 Sebastian Korb is a Ph.D. student at the Swiss Center for Affective Sciences and the faculty of psychology of the University of Geneva, under the supervision of Didier Grandjean and Klaus Scherer. His research domain is the “physiology of emotion regulation”. Using EEG and facial EMG he is investigating the differences between voluntary (controlled or posed) and natural (spontaneous, automatic and involuntary) emotional facial expressions, both at the production and perception level.

 

 

Training Computer Vision Systems with Implicit Brain Processing

Ashish Kapoor, Microsoft Research, Redmond, USA, akapoor@microsoft.com

Desney Tan, Microsoft Research, Redmond, USA, desney@microsoft.com

Pradeep Shenoy, University of Washington, Seattle, USA,  psenoy@cs.washington.edu

Eric Horvitz, Microsoft Research, Redmond, USA, Horvitz@microsoft.com

 

Computer vision has been a popular tool for researchers in the Face and Gesture community. However, there is a tremendous disparity between the workings of computer vision algorithms and how a human brain processes visual information. In our work, we investigate how to exploit this disparity for vision tasks. Specifically, we focus on human aided computing, in which we use an electroencephalograph (EEG) device to measure the implicit cognitive processing that occurs in the brain as users see and process images. We discuss how the challenging task of categorizing objects, including faces, from images can benefit by analyzing human brain responses. We describe our efforts exploring how we can ideally combine the efforts and competencies of human and machine computation to achieve improved recognition performance and highlight methodologies that could allow us to use the brain signal to train better computer vision systems