Processing faces in Alzheimer’s disease patients: how a familiar face becomes unfamiliar

Face recognition in Alzheimer’s disease

In the healthy brain, the processing of familiar faces has been examined extensively (Johnston and Edmonds, 2009). Face preferential regions, such as the fusiform face area, are important in facial processing, independent of whether the face is familiar or unfamiliar (Rossion, Schiltz, and Crommelinck, 2003). However, when processing familiar faces, there are two additional regions of the brain that are involved. Medial temporal lobe structures, including the hippocampus, amygdala, and the perirhinal cortex, along with inferior regions of the temporal cortex, respond abruptly when a person recognizes a face as familiar (Ramon, Vizioli, Liu-Shang, and Rossion, 2015). Interestingly, the medial temporal lobe is an area highly implicated in the pathology of Alzheimer’s disease. Atrophy of the medial temporal lobe, in particular the hippocampus and the amygdala, has been well-documented as a biomarker for Alzheimer’s disease (Visser, Verhey, Hofman, Scheltens, and Jolles, 2002). Atrophy of these regions in Alzheimer’s disease patients may play a role in the breakdown of their face recognition processes as the disease progresses.

A remaining question is, what aspect of cognition are the face recognition deficits in Alzheimer’s disease specifically attributed to? There has been speculation that the perceived deficit in facial recognition is actually due to a breakdown in semantic stages of recognition, as Alzheimer’s patients commonly have trouble remembering names of familiar people (Werheid and Clare, 2007). Hodges, Salmon, and Butters (1993) looked at the processing of familiar faces in Alzheimer’s disorder by spontaneous and cued naming of famous faces. In a group of mildly impaired Alzheimer’s patients, performance in identifying famous faces declined with increasing symptom severity. While patients were able to still spontaneously name famous people, they took longer to recognize and identify faces, and could not identify and name as many faces as the control group. Conditions of semantic cueing did not improve their naming abilities. The researchers concluded that Alzheimer’s patients may have a breakdown in the pre-semantic stages of facial processing. As such, there must be a breakdown of facial processing in a stage prior to name retrieval.

An alternative hypothesis posits that Alzheimer’s patients are impaired at the perceptual level of face processing, i.e., their ability to holistically process a face is impaired. Lavellée et al. (2016) investigated the processing of upright and inverted faces and cars in mildly impaired Alzheimer’s patients. The inversion effect, in which it takes longer to process inverted faces versus upright faces, occurs because people normally process faces as an integrated whole, and are impaired in this ability when a face is inverted (Taubert, Apthorp, Aagten-Murphy, and Alais, 2011). Compared to controls, patients in the study showed a reduced inversion effect for faces, but not for cars. The researchers suggest that Alzheimer’s disease patients are impaired in higher-level visual processing that requires holistic representation. Holistic processing of faces has been heavily implicated in the ability to recognize a face, such that the loss of holistic processing in Alzheimer’s patients may be a direct link to the loss in recognition of familiar faces (Richler et al., 2011).

What then, is the neural basis for the loss of familiar face recognition? Certain prefrontal brain regions have been associated with familiarity recognition. This includes the bilateral superior frontal cortex, as well as the right middle orbital gyrus (Montaldi, Spencer, Roberts, and Mayes, 2006). Failure to activate these regions when exposed to familiar stimuli has been documented in patients with early stage Alzheimer’s disease (Donix et al., 2013). There are also notable differences in cerebellar activity in patients with Alzheimer’s disease, with a higher activation of the right cerebellum for unfamiliar stimuli in Alzheimer’s patients as compared to controls. Alzheimer’s patients have less activation in the medial frontal cortex, which has been associated with explicit access to contextual information surrounding a stimulus. In addition, Alzheimer’s patients present with a reduced ability to recruit frontal brain regions during the perception of personally familiar stimuli (Li et al., 2015). Together, a reduction in access to contextual knowledge and episodes surrounding a familiar stimulus translates to reduced ability to recognize familiar stimuli in everyday life (Donix et al., 2013).

However, the loss of the ability to holistically process faces seems to be independent from the loss of processing other familiar stimuli. Another common feature in Alzheimer’s patients is loss of memory for familiar locations (Pai and Jacobs, 2004). The mechanism for this ability has been found to be independent from face recognition, as impairment for scene recognition occurs earlier in the progression of the disease than does the impairment for face recognition (Cheng and Pai, 2010). Thus, there is not a generalized deficit in recognition per se, but rather separate mechanisms may be co-occurring to contribute to the loss of familiarity for different types of stimuli. Research indicates that Alzheimer’s patients are better at recognizing objects such as houses than they are at recognizing faces (Kawagoe, Matsushita, Hashimoto, Ikeda, and Sekiyama, 2017). Further evidence for the specificity of facial processing impairment is indicated by deficits in emotion detection. As compared to mild cognitive impairment, patients with Alzheimer’s disease are worse at detecting emotions in faces. In predementia stages of the disorder, there has been no documented impairment in emotion processing (Bediou et al., 2009). The deficit in emotional expression processing in Alzheimer’s patients indicates that faces are more difficult to process in this population, and thus there is a specific deficit in holistic facial processing, rather than a general deficit in recognition of familiar stimuli.

Alzheimer’s patients are not just impaired in recognizing the faces of familiar people. As the disease progresses, patients increasingly have difficulty with identifying a person, independent of face recognition processes. A final facet to explain the impairment of face recognition in Alzheimer’s disease involves the ability to invoke theory of mind. Theory of mind involves the ability to understand that others have differing beliefs, and allows engagement in social and introspective behaviors (Dohnel et al., 2012). Areas of the brain implicated in theory of mind processing include prefrontal cortical areas; the medial prefrontal cortex has been implicated in conceptualizing mental states of others, while the superior frontal gyrus is implicated in self-related processes (Dohnel et al., 2012; Goldberg, Harel, and Malach, 2006). Theory of mind has been posited to be associated with face recognition processes. Impairment of this ability is well-documented in disorders such as schizophrenia and autism spectrum disorders, disorders which are also accompanied by impaired face recognition abilities (Irani et al., 2006). According to a review by Gobbini and Haxby (2007), theory of mind is necessary for familiarity recognition, as theory of mind regions of the brain are implicated in the non-visual cues of person recognition. These regions are hypothesized to act in conjunction with visual cues to invoke a mental representation of a familiar person.

In Alzheimer’s disease patients, a genuine deficit in theory of mind has been documented at early stages of the disorder, and progresses as patient condition worsens (Moreau, Rausy, Viallet, and Champagne-Lavau, 2016). Furthermore, when Alzheimer’s disease patients process familiar and self-face images, there is a negative correlation between connectivity in theory of mind implicated brain regions and ability to recognize faces. The current hypothesis for this phenomenon is that a decrease in the segregation of implicated neural networks leads to an increase in interference (Kurth et al., 2015). This implies that the medial prefrontal cortex and the superior frontal gyrus, regions involved in theory of mind, are implicated in the processing of familiar faces, and could explain the connection between worsened performance and unusual connectivity of these brain regions seen in Alzheimer’s disease patients. This, taken with evidence from non-Alzheimer’s research, implied that theory of mind plays a role in the loss of familiarity recognition in Alzheimer’s disease. Thus, a reduced ability to invoke theory of mind in Alzheimer’s patients, causing a deficit in understanding the minds of others, likely plays an indirect role in their impairment in familiar face recognition.

The time course of familiar face recognition in Alzheimer’s disease

As Alzheimer’s disease is neurodegenerative, it is important to examine how deficits in face recognition manifest throughout the course of the pathology. Genetic causes have been implicated in the pathology of Alzheimer’s, with the discovery of autosomal-dominant Alzheimer’s disease mutations. There is emerging evidence that for carriers of these mutations, dysfunction of neural circuits involved in memory occur prior to behavioral changes associated with diagnosis. In an fMRI study, carriers were shown to have abnormal activation in recognition tasks of face processing. Specifically, less activation was found in bilateral parietal regions of the brains of the subjects (Norton, Baena, Pulsifier, Lopera, and Quiroz, 2016). It is evident therefore that neural changes are taking place during the preclinical stages of the disease. In these preclinical stages, patients are still able to compensate for these neural changes and maintain normal cognitive abilities.

Eventually, neural changes overtake the brain’s ability to compensate, and cognitive decline begins. Once behavioral changes are evident in the progression of Alzheimer’s disease, face recognition will deteriorate in a more or less linear manner. In the mild cognitive impairment stage of Alzheimer’s disease, there is evidence that patients are still able to identify familiar faces at the same rate as healthy individuals (Hodges et al., 1993). The major difference is length of time for recognition to occur; patients need more time to recognize a familiar face, and have more trouble identifying familiar faces by name (Hodges et al., 1993). So, while at the earlier stages of Alzheimer’s disease patients will still be able to recognize familiar people, it is more difficult and time-consuming.

Patients with mild cognitive impairment (GDS stages 2 and 3) due to Alzheimer’s disease will also not have difficulty recognizing recent photographs of themselves, which is a deficit that appears as the disease progresses. With moderate cognitive impairment (GDS stages 4 and 5), patients become better at recognizing more dated photographs of both themselves and their spouses in comparison to more recent images. Moderately impaired patients will also perform worse both on time taken to recognize familiar faces and on rate of recognition (Kurth et al., 2015). This indicates that with disease progression, patients are worse at recognizing familiar faces in general, and their mental representations of themselves and familiar people are likely to be temporally inaccurate.

Despite a decline in the ability to recognize familiar faces, there is evidence that Alzheimer’s patients are still able to acquire representations of faces in the earlier to intermediate stages of the disorder. In a study by Hawley and Cherry (2004), patients with moderate Alzheimer’s (GDS stage 4 and 5) were able to learn face-name recognition. Patients were trained to learn faces and names based on pictures, and were subsequently tested both in picture conditions and on live person transfer abilities, i.e., recognizing the people from the images in real life. Improvements were displayed over a two-week period, and maintenance of learned information was displayed. Thus, the ability to acquire familiarity of a face is still preserved at some level at these stages, and is displayed at a functional, real-world level. However, it is unclear from this study at what stage the ability to learn or recognize faces is lost, and whether it is consistent across Alzheimer’s patients.

In the final and most severe stages of Alzheimer’s disease (GDS Stage 6 and 7), patients have often lost all ability to recognize familiar faces and, most notably, can no longer recognize their own faces. Biringer and Anderson (1992) tested self-recognition in Alzheimer’s patients in stages 5, 6, and 7 on the GDS scale, by presenting the patients with a mirror after marking their foreheads with paint. At stages 6 and 7, there was no response to the mark on the forehead, while the ability to self-recognize was still intact in GDS stage 5 patients. This provides evidence that patients at the final stages of the disease have lost all concept of the self and theory of mind. Without a concept of the self, the ability to recognize one’s own face is lost, as well as the recognition of loved ones.