A method for obtaining 3-dimensional facial expressions and its standardization for use in neurocognitive studies

https://doi.org/10.1016/S0165-0270(02)00006-7Get rights and content

Abstract

Facial expressions of emotion are increasingly being used in neuroscience as probes for functional imaging and as stimuli for studying hemispheric specialization for face and emotion processing. Available facial stimuli are 2-dimensional and therefore, their orientation is fixed and poorly suited for examining asymmetries, they are often obtained under poorly specified conditions, usually posed, lack ethnic diversity, and are of restricted age range. We describe a method for accurately acquiring and reconstructing the geometry of the human face and for display of this reconstruction in a 3-dimensional format. We applied the method in a sample of 70 actors and 69 actresses expressing happiness, sadness, anger, fear and disgust, as well as neutral expressions. Each emotion was expressed under three levels of intensity and under both posed and evoked conditions. Resulting images are of high technical quality and are accurately identified by raters. The stimuli can be downloaded in digital form as ‘movies’ where angle and orientation can be manipulated for inclusion in functional imaging probes or in tests that can be administered as measures of individual differences in facial emotion processing. The database of emotional expressions can also be used as a standard for comparison with clinical populations.

Introduction

Neuroscience investigations increasingly use facial expressions of emotions as probes for functional imaging (Adolphs et al., 1996, Breiter et al., 1996, Hyman, 1998, Morris et al., 1996, Morris et al., 1998) and in tasks that can be correlated with neuroanatomic measures and effects of brain disorders (Adolphs et al., 1994, Blonder et al., 1991, Borod, 1993, Kohler et al., 2000). Studies have applied stimuli of varying quality obtained under differing conditions, most including posed emotions. Facial stimuli are typically of a restricted ethnicity and age range. Furthermore, they are 2-dimensional photographs, where facial orientation is either poorly controlled or artificially made at straight angle. Such 2-dimensional stimuli are not amenable for manipulations of angle and orientation, and raise methodological concerns when applied to examination of facial asymmetries which could be related to hemispheric specialization (Sackeim et al., 1978; see review in Borod et al., 1997).

We describe a method for obtaining digitized high-quality 3-dimensional photographs of facial expressions and its implementation under standardized conditions in a sample of 139 actors (70 male, 69 female) of diverse ethnicity and age. The method has generated a digital database of facial expressions of happiness, sadness, anger, fear, and disgust, under both posed and evoked conditions, each at three levels of intensity, and neutral expressions. The validity of the emotions expressed was established in a sample of raters. The stimuli are available for downloading from the internet as pictures or 3-dimensional movies.

Section snippets

Image acquisition methodology

We developed a protocol to accurately acquire and reconstruct the geometry of the human face and display this reconstruction in a 3-dimensional format. Such methods of 3D reconstruction from multiple stereo camera setups are called polynocular stereo. Image capture is achieved with four Pulnix TM9701 digital video cameras, and a Nikon N90 35 mm SLR camera (with Kodak Ektachrome 320T film, exposed at 1/60 s and f/5.6) for color rendering is centered among them. The digital cameras are mounted on

Discussion

The method described for acquiring 3-dimensional photographs of emotional displays is feasible, produces uniformly good results across the range of age and ethnicity, and shows evidence of validity. The reconstructions are of high quality and can be easily classified by raters with regard to the identity of the emotion displayed. Furthermore, the accuracy of detecting the different emotions is similar to that reported in earlier studies (Erwin et al., 1992, Wallbott, 1998). Finally, a

Acknowledgments

We thank David Ingraham and Mary Jackman, Directors and the dedicated actors who produced the emotional expressions. We also thank Brian Klimas for assistance. Supported by the Bosworth Fund, NIMH Grants MH60722, MH43880, Grant Agency of the Czech Republic (Project GACR 102/01/1371) and Czech Ministry of Education (Project MSM 212300013).

References (23)

  • S.E. Hyman

    A new image of fear and emotion

    Nature

    (1998)
  • Cited by (0)

    View full text