Face recognition impairments despite normal holistic processing and face space coding: Evidence from a case of developmental prosopagnosia

Abstract

Holistic processing and face space coding are widely considered primary perceptual mechanisms behind good face recognition. Here, however, we present the case of S.P., a developmental prosopagnosic who demonstrated severe impairments in face memory and face perception, yet showed normal holistic processing and face space coding. Across three composite experiments, S.P. showed normal-strength holistic processing for upright faces and no composite effect for inverted faces. Across five aftereffect experiments, S.P. showed normal-sized face aftereffects, which derived normally from face space rather than shape-generic mechanisms. The case of S.P. implies: (a) normal holistic processing and face space coding can be insufficient for good face recognition even when present in combination; and (b) the focus of recent literature on holistic processing and face space should be expanded to include other potential face processing mechanisms (e.g., part-based processing). Our article also highlights the importance of internal task reliability in drawing inferences from single-case studies.

Keywords:

• Face perception
• Face recognition
• Developmental prosopagnosia
• Holistic processing
• Face space
• Composite effect
• Face adaptation
• Face aftereffect

Acknowledgments

We are extremely grateful to S.P. for her generosity, time, and interest, and we thank Devin Bowles for referring her to us. We also thank: Bradley Duchaine and Raka Tavashmi for providing us with their Cambridge Car Memory Test (CCMT); Anne Aimola Davies for directing us to literature on 95% confidence intervals (CIs) in neuropsychology; Davide Rivolta and Mary Broughton for testing some controls on the Macquarie Centre for Cognitive Science Famous Face Test (MACCS FFT). Daphne Maurer and Mayu Nishimura codesigned (with L.J. and G.R.) the original method for the identity aftereffect task used in Experiment 4. This research was supported by Australian Research Council Grants DP0450636 and DP0984558 to E.M., DP0770923 to L.J., and DP0877379 to G.R. T.S. is grateful for scholarship support from Australian National University (ANU) Centre for Visual Sciences and overseas student fee waiver from ANU Department of Psychology. T.S. and E.M. designed the study, wrote the paper, and take responsibility for task selection, data analysis, results interpretation, and theoretical claims. T.S. tested S.P. R.P. contributed to choice of neuropsychological tests. All other contributions involved provision of previously unpublished tasks and/or control data: H. Dennett for task creation and control data for Experiment 6 (face expansion–contraction aftereffect) and Experiment 7 (horse aftereffect); H. Darke for task creation and control data for Experiment 2 (composite front-view adult faces); A.H. and M.P. for control data and its analysis for CCMT; A.D. for control data and its analysis for Experiment 1 (three-quarter-view composite); E.W. and R.P. for task creation and some control data for MACCS FFT 2008; L.J., G.R., R.P., and E.W. for task creation and control data for Experiment 4 (identity aftereffect).

Notes

¹Note that little is known about the relationship between holistic processing and face space coding. It is not known whether one precedes the other in time course, or whether they occur in parallel, or even whether they might, in some currently not-understood manner, be two ways of tapping the same underlying mechanisms (for discussion, see Susilo, McKone, & Edwards, 2010a).

²Herzmann, Danthir, Schacht, Sommer, and Wilhelm (2008) computed a reliability of .23 for their composite task, but we excluded this report because their task did not produce a composite effect.

³For example, S.P. has not been tested as to whether she shows identity-contingent aftereffects (e.g., adaptation to contraction for one identity and simultaneously expansion for another; Robbins & Heck, 2009; Yamashita, Hardy, DeValois, & Webster, 2005).

⁴And also indicating that, in contrast to the suggestion of Konar et al. (2010), holistic processing can be functionally associated with face individuation ability.