J Cogn Neurosci. 2017 Mar;29(3):573-591. doi: 10.1162/jocn_a_01063.

Perceptual Learning of Faces: A Rehabilitative Study of Acquired Prosopagnosia.

Davies-Thompson J1,2, Fletcher K1,3, Hills C1, Pancaroglu R1, Corrow SL1, Barton JJ1.
1 University of British Columbia.
2 University of Nottingham.
3 Derby Hospitals NHS Foundation Trust.



Despite many studies of acquired prosopagnosia, there have been only a few attempts at its rehabilitation, all in single cases, with a variety of mnemonic or perceptual approaches, and of variable efficacy. In a cohort with acquired prosopagnosia, we evaluated a perceptual learning program that incorporated variations in view and expression, which was aimed at training perceptual stages of face processing with an emphasis on ecological validity. Ten patients undertook an 11-week face training program and an 11-week control task. Training required shape discrimination between morphed facial images, whose similarity was manipulated by a staircase procedure to keep training near a perceptual threshold. Training progressed from blocks of neutral faces in frontal view through increasing variations in view and expression. Whereas the control task did not change perception, training improved perceptual sensitivity for the trained faces and generalized to new untrained expressions and views of those faces. There was also a significant transfer to new faces. Benefits were maintained over a 3-month period. Training efficacy was greater for those with more perceptual deficits at baseline. We conclude that perceptual learning can lead to persistent improvements in face discrimination in acquired prosopagnosia. This reflects both acquisition of new skills that can be applied to new faces as well as a degree of overlearning of the stimulus set at the level of 3-D expression-invariant representations.

PMID: 28139958



Recognising faces is a seemingly easy process for most people. However, others can experience severe difficulties in face perception – a condition known as prosopagnosia. Such deficits cause social anxiety, feelings of isolation, chronic stress, and have a negative impact on relationships (Yardley, McDermott, Pisarski, Duchaine, & Nakayama, 2008). Evidence from neuroimaging and single-cell recording studies suggest that regions in inferior temporal cortex are involved in processing facial identity (Haxby, Hoffman, & Gobbini, 2000). This is supported by neuropsychological case studies that describe individuals with damage to these areas, resulting in difficulties in face recognition (Davies-Thompson, Pancaroglu, & Barton, 2014; de Renzi, 1986; Haxby et al., 2000; Sergent & Villemure, 1989). Although there have been many studies attempting to elucidate the mechanisms behind prosopagnosia, there have been few attempts at remedying this perceptual defect. These have generally involved single subjects and have employed a variety of approaches, some targeting perceptual deficits, others aiming at improving memory for faces. The results have been mixed, in terms of both immediate improvement and the persistence of any benefit.

The small number of rehabilitation studies in acquired prosopagnosia is mainly due to the rarity of this condition. The interpretation of these studies is further complicated by the fact that acquired prosopagnosia is a syndrome rather than a single condition, with different lesions and different functional deficits in different patients. Whether a single training approach will be effective for all patients is not known. Here, we assessed a perceptual learning face training programme in a cohort of 10 acquired prosopagnosics (Figure 1).

Perceptual learning is “the performance improvements in perceptual tasks as a result of practice or training” (Petrov, Dosher, & Lu, 2005). This technique has been previously used to improve face recognition in healthy subjects (Fine & Jacobs, 2002; Gold & Rabins, 1989; Hussain, Sekuler, & Bennett, 2009a, 2009b), as well as in individuals with a developmental form of prosopagnosia with the training showing promising results (Bate et al., 2015; DeGutis, Bentin, Robertson, & D’Esposito, 2007). We used perceptual learning to train patients to discriminate face pairs in an online training programme. Participants are presented with three faces and are asked to indicate which of the bottom two faces most resembles the top face. Different difficulty levels were created by morphing the two choice faces together (Figure 2A); the similarity between the two choice faces increases (becomes more difficult) after correct responses and decreases (becomes easier) after incorrect responses, so that participants are always training near their current limits. Sessions are repeated 3 times per week, with each session lasting approximately 30 minutes. Over the course of 11 weeks, the bottom two faces gradually increased in viewpoint and magnitude of expression, so that by the end of the eleventh week, participants were matching the faces across large changes in views (40°) and expression (100% neutral, happy, angry, sad) (Figure 2B). Finally, each participant completed a control task, which consisted of watching an assigned TV series, and lasting approximately the same time as the training (30mins, three times per week, 11 weeks). Half of the participants completed the training followed by the control task, whilst the other half completed the control task followed by the training.

To assess improvement, participants completed six online tests at three time-points: at baseline, after training, and after the control task. These tests assessed participants’ abilities to discriminate: 1) trained faces; 2) untrained views; 3) untrained expressions; and 4) new identities. Figure 3 shows the average improvement across all patients after training and after the control task. We found that participants significantly improved after training, showing a 39% improvement on trained views and expressions, 48% improvement on new views, and 57% improvement on new expressions. In contrast, no significant improvements were observed after the control task (same faces: -3%; new views: 0%, new expressions: -4%). Participants also significantly improved on their abilities to discriminate between new identities, improving 27% on trained views and expressions, 30% on new views, and 40% on new expressions, but no significant improvements after training (2%, -9%, and -4% respectively). These results show that our perceptual learning paradigm can improve face perception in acquired prosopagnosia, and that this training generalizes to new viewpoints, new expressions, and new identities. Next, we assessed whether the training benefits were sustained after three months of no training by assessing performance in participants (n=5) who undertook the training task followed by the control task. Average perceptual sensitivity across all conditions increased from 41% at baseline to 60% after training, with this increase being sustained after three months (57%). Finally, we found that the effectiveness of the training across patients did not differ as a function of the location or extent of neurological damage.

Importantly, improvements in face recognition abilities should benefit patients in their day-to-day lives. Colloquially, our patients reported a range of subjective improvements, from none, to being able to recognize family resemblances. However, subjective improvements did not correlate with performance on the objective assessments, suggesting that there may be a ‘missing link’ to make objective improvements more applicable to day-to-day face recognition performance. Interestingly, after training the majority of our patients reported an increase in relying on the eye region when attempting to recognizing a face; this is of interest because acquired prosopagnosics show reduced fixations and viewing times of the eye region, as well as impaired performance on matching and identification of the eye region in relation to other face parts (Caldara et al., 2005; Ramon & Rossion, 2010; Rossion, Kaiser, Bub, & Tanaka, 2009).

In sum, our study suggests that rehabilitation of acquired prosopagnosia is possible, and that perceptual learning shows promise as a tool for such rehabilitation, enabling generalisation to untrained viewpoints, expressions, and identities, as well as maintenance after a period of no training. However, future training studies are required to improve the generalizability of face training to day-to-day life.


Figure 1. MRI scans of the ten prosopagnosic patients. 



Figure 2. Example training trials. Participants are asked to indicate which of the bottom two faces most resembles the top. A) Within a session, the bottom two faces become more similar if participants answer correctly, and more dissimilar if participants answer incorrectly. Similarity was manipulated by creating a morph continuum of the two choice faces. B) Throughout the weeks, patients progress through different training blocks. In week 1, the faces are front-on and contain neutral expressions. During weeks 2-4, the view difference of the bottom two choice faces gradually increase in view (10°, 20°, 40°). During weeks 5-8, the view remains front-on but the faces gradually increase in emotional expression (10%, 33%, 66%, 100%). Finally, during weeks 9-11, the view difference increases (10°, 20°, 40°) for the 100% emotional expression.



Figure 3. Average %-change in perceptual sensitivity after training (top) and after the control task (bottom) for trained faces, new views, and new expressions. After training, performance on new views and new expressions was as good as performance on the trained faces, indicating that the training generalized. There was also significant improvement for untrained identities, indicating transfer. In contrast, there was no effect on performance after the control task.



Bate, S., Bennetts, R., Mole, J. A., Ainge, J. A., Gregory, N. J., Bobak, A. K., & Bussunt, A. (2015). Rehabilitation of face-processing skills in an adolescent with prosopagnosia: Evaluation of an online perceptual training programme. Neuropsychological Rehabilitation, 25(5), 733-762.

Caldara, R., Schyns, P., Mayer, E., Smith, M., Gosselin, F., & Rossion, B. (2005). Does Prosopagnosia Take the Eyes Out of Face Representations? Evidence for a Defect in Representing Diagnostic Facial Information following Brain Damage. J Cogn Neurosci, 17(10), 1652-1666.

Davies-Thompson, J., Pancaroglu, R., & Barton, J. (2014). Acquired prosopagnosia: structural basis and processing impairments. Front Biosci (Elite Ed), 6, 159-174.

de Renzi, E. (1986). Prosopagnosia in two patients with CT scan evidence of damage confined to the right hemisphere. Neuropsychologia, 24(3), 385-389. Retrieved from http://www.sciencedirect.com/science/article/pii/0028393286900230

DeGutis, J. M., Bentin, S., Robertson, L. C., & D’Esposito, M. (2007). Functional plasticity in ventral temporal cortex following cognitive rehabilitation of a congenital prosopagnosic. J Cogn Neurosci, 19(11), 1790-1802. doi:10.1162/jocn.2007.19.11.1790

Fine, I., & Jacobs, R. A. (2002). Comparing perceptual learning across tasks: A review. Journal of vision, 2(2), 5-5.

Gold, K., & Rabins, P. V. (1989). Isolated visual hallucinations and the Charles Bonnet syndrome: a review of the literature and presentation of six cases. Compr Psychiatry, 30(1), 90-98.

Haxby, J. V., Hoffman, E. A., & Gobbini, M. I. (2000). The distributed human neural system for face perception. Trends Cogn Sci, 4(6), 223-233. Retrieved from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=10827445

Hussain, Z., Sekuler, A. B., & Bennett, P. J. (2009a). How much practice is needed to produce perceptual learning? Vision research, 49(21), 2624-2634.

Hussain, Z., Sekuler, A. B., & Bennett, P. J. (2009b). Perceptual learning modifies inversion effects for faces and textures. Vision research, 49(18), 2273-2284.

Petrov, A. A., Dosher, B. A., & Lu, Z. L. (2005). The dynamics of perceptual learning: an incremental reweighting model. Psychol Rev, 112(4), 715-743. doi:2005-13637-001 [pii]


Ramon, M., & Rossion, B. (2010). Impaired processing of relative distances between features and of the eye region in acquired prosopagnosia-Two sides of the same holistic coin? Cortex, 46(3), 374-389.

Rossion, B., Kaiser, M. D., Bub, D., & Tanaka, J. W. (2009). Is the loss of diagnosticity of the eye region of the face a common aspect of acquired prosopagnosia? Journal of Neuropsychology, 3(1), 69-78.

Sergent, J., & Villemure, J.-G. (1989). Prosopagnosia in a Right Hemispherectomized Patient. Brain, 112(4), 975-995. Retrieved from http://brain.oxfordjournals.org/content/112/4/975.abstract

Yardley, L., McDermott, L., Pisarski, S., Duchaine, B., & Nakayama, K. (2008). Psychosocial consequences of developmental prosopagnosia: A problem of recognition. Journal of Psychosomatic Research, 65(5), 445-451. Retrieved from http://www.sciencedirect.com/science/article/pii/S0022399908001578