The face databases described below are freely available to other researchers who intend to conduct non-profit, academic research. Researchers who download the faces should only use these faces for non-profit research and should acknowledge the proper sources of the faces, as well as relevant references. The relevant references for each database are described in the downloadable agreement for the specific set. The first two datasets consist of unmanipulated face images that have been rated on a number of trait dimensions (e.g. attractiveness, threat, trustworthiness). The other four datasets consist of faces that have been manipulated in computer models developed by Oosterhof & Todorov (2008). The image on this webpage shows variations of one facial identity on trustworthiness.
- 300 randomly generated computer faces and their mean ratings on 9 trait dimensions
- 66 faces from the Karolinska dataset and their mean ratings on 14 trait dimensions
- 175 computer generated faces that have been manipulated on trustworthiness (25 facial identities x 7 versions of trustworthiness: -3, -2, -1, 0, 1, 2, 3 SD)
- 175 computer generated faces that have been manipulated on dominance (25 facial identities x 7 versions of dominance: -3, -2, -1, 0, 1, 2, 3 SD)
- 175 computer generated faces that have been manipulated on threat (25 facial identities x 7 versions of threat: -3, -2, -1, 0, 1, 2, 3 SD)
- 300 computer generated faces that have been manipulated on trustworthiness (100 facial identities x 3 versions of trustworthiness: -3, 0, 3 SD)