Naming and recognizing famous faces in temporal lobe epilepsy

Methods.

Results.

A one-way analysis of variance (ANOVA) of the number of correct spontaneous naming scores (table 2) yielded a group effect [F (4,68) = 8.5; p < 0.001]. Post hoc Newman-Keuls tests demonstrated that all epilepsy groups were impaired relative to healthy controls (p < 0.05), and that those patients who had undergone LATL were most impaired in spontaneously naming photographed faces of very famous people, differing from the three other patient groups (p < 0.05). It is important to note that LATL patients’ naming deficit for familiar faces was significantly greater than that seen in patients with LTLE who had not had surgery in the anterior left temporal lobe. These results, therefore, support a unique role for the left anterior temporal lobe in the retrieval of names of famous people. Using a cutoff score of two standard deviations below the mean of the healthy controls, 88% of LATL patients’ naming score fell below this value, identifying them as significantly impaired, as were 75% of the patients with LTLE, but only 54% of patients with RATL and 61% of patients with RTLE were similarly impaired in naming famous faces.

To separate name recall from other knowledge about the famous faces, a one-way ANOVA was conducted on the number of responses that demonstrated correct knowledge of the photographed face, either through spontaneous naming or correct description. A group effect was found [F (2,68) = 3.5; p < 0.05]. On this measure, in contrast to the results of spontaneous naming, however, post hoc Newman-Keuls tests revealed that only patients with RATL were impaired (p < 0.05) relative to controls. On this measure, 38% of patients with RATL scored two standard deviations or lower than the mean of the control group. Patients with LTLE and LATL did not differ significantly from controls, and patients with RTLE also did not differ from controls. There was a trend [t (37) = 1.9; p = 0.06] for patients with RATL to perform more poorly than patients with RTLE, suggesting that there may have been a loss of previously more intact abilities for recognizing famous faces that was specifically related to lesion in the right anterior, as compared to medial, temporal lobe.

Time since surgery did not correlate significantly with either naming or description of famous faces in the right and left post ATL groups.

When correct identification of the famous pictured faces was assessed by correct naming, or correct description, or correct response to a first name cue, the overall group difference was no longer significant [F (4,68) = 1.7; p > 0.10], suggesting that all groups had familiarity at least with the names of the famous people. First name cues are believed to be especially effective in eliciting the last name of familiar people as these two components of proper names are thought to be encoded in memory as single units,35,36⇓ and they function much like the phonologic components of single words (e.g., phonemic cues). Whereas retained knowledge of these phonologic lexical units cannot be taken to reflect complete semantic knowledge about the pictured person, retained familiarity with the name form does suggest that all the patients had some knowledge of the pictured person at some time in their past.

Two-way ANOVA examined differences related to side of epilepsy/surgery (left vs right) and surgical status (pre vs post surgery) on cognitive measures that did not involve famous face processing (table 3). For visual confrontation picture naming (Boston Naming Test) there was an effect of side of epilepsy/surgery [F (1,58) = 9.00; p < 0.01] and an interaction between side of epilepsy/surgery and surgical status that approached significance [F (1,58) = 2.84; p < 0.10]. Post hoc Newman-Keuls test demonstrated that the LATL group differed from both right sided groups, but there was no significant difference between the LATL and LTLE groups in naming common nouns. These data provide suggestive but weak evidence that retrieval of common nouns, like the findings for unique proper names, is specifically impaired by a lesion in the left anterior temporal lobe. Discrimination of unfamiliar faces (Benton Facial Recognition) revealed an effect of side of epilepsy/surgery [F (1,58) = 4.23; p < 0.05], but no effect of surgical status and no interaction. Whereas the patients with right-sided epilepsy/surgery performed below the levels of left-sided patients, there was no effect of anterior temporal lobe surgery on unfamiliar face discrimination scores. Similar results were found in an analysis of scores on the delayed unfamiliar face recognition memory test (Graduate Hospital Facial Memory), which yielded an effect of group [F (1,58) = 6.23; p < 0.05], but no other significant effects. Finally, analysis of delayed recognition memory for newly presented words (California Verbal Learning Test) showed an effect of side of epilepsy/surgery [F (1,58) = 9.00; p < 0.01] but no significant effect of surgical status and no interaction. Thus, unlike the results for famous faces and for visual confrontation naming of common nouns, which both indicated significant effects of anterior temporal lobe removal, other material-specific cognitive deficits associated with lateralization of epilepsy/surgery did not show a consistent association with whether the anterior temporal pole was lesioned.

Pearson product-moment correlation coefficients were computed to assess associations between performances on the test of famous faces and measures of other cognitive functions known to be affected in temporal lobe epilepsy37,38-40⇓⇓⇓ (table 4). There was a substantial relationship between naming of proper nouns indexing unique entities and famous faces and problems naming common nouns that refer to other classes of objects. Recognition memory for learning newly presented words also correlated with the naming of famous faces. These results seem to indicate that familiar face naming is affected, at least in part, by general problems in language encoding and language retrieval. There was a minor, but significant, relationship between perceptual discrimination of unfamiliar faces and famous face naming, suggesting that deficits in the perceptual processing of presented photographs might also contribute in some small fashion to the problems naming famous people. Famous face knowledge, which relies less on language ability, was not strongly related to common object naming or verbal memory measures. Rather, correct knowledge about famous faces correlated with perceptual matching of unfamiliar faces and with visual nonverbal episodic memory (recognition memory for recently presented unfamiliar faces). The pattern of these associations suggests that two independent factors may be related to identifying famous faces: one factor entails knowledge and retrieval of the names of famous people, and the other involves visuoperceptual memory for the identity of famous people.

Discussion.

We found symptoms of prosopagnosia in patients who had undergone RATL and who were impaired in assigning names to the faces of famous people, providing semantic information about these individuals, and learning and remembering new faces. Our relatively liberal scoring of the descriptions of the famous faces provided by participants likely underestimates the true degree of impaired familiar face recognition in patients with RATL. Evidence that a matched group of patients with RTLE, most of whom had pathology in medial structures of the temporal lobe, did not show a deficit in their recognition and description of famous faces when examined before ATL further suggests that the postsurgical deficit in famous face recognition was related to removal of anterior cortical structures in the right temporal lobe. Before ATL, knowledge of and access to this knowledge about famous people from facial stimuli appeared to be relatively intact in patients with RTLE. The fact that RATL patients’ performance increased to normal levels when face identification was tested using first name cues indicates that the names of presented faces were indeed within these patients’ vocabulary at some time in the past, before resective epilepsy surgery. We assume, in accord with the suggestion of others,23,41⇓ that semantic knowledge about famous people is represented bilaterally in the brain and that it is processed in more posterior regions of the temporal lobes. The deficit in patients with RATL, therefore, seems to lie in their ability to use visual facial information to access this knowledge about familiar famous people. Although we did not specifically test patients’ semantic knowledge about the famous people using nonface stimuli and tasks, based on RATL patients’ normal fund of general factual information and their normal knowledge of other famous events,21,42⇓ we expect that they would demonstrate adequate semantic knowledge concerning the famous people when visual face processing is not required. This needs to be assessed more directly, however, given reports of some individuals with right anterior temporal lobe lesions who have apparent loss of nonvisual person-specific knowledge.9,22⇓

Impaired recognition of the faces of familiar people was related to impaired perceptual matching and new learning of unfamiliar faces. However, deficits in perceptual discrimination and memory are not likely to completely explain the problems in processing famous faces. Whereas famous face discrimination and identification became selectively impaired after anterior temporal lobe resection, processing of unfamiliar faces for perceptual matching or memory did not decline significantly as a result of surgery.43 We cannot completely rule out the possibility that disruption in lower level capacities for perceiving visual stimuli holistically12 contributes to disturbed recognition of famous faces, but such low level deficits do not appear to explain the prosopagnosic deficit we found. Based on these data, it appears that the neural systems devoted to processing unfamiliar faces and those specialized for processing familiar faces do not overlap completely.43-45⇓⇓ Furthermore, our results support the claim that anterior temporal lobe structures play a special role in binding together information relevant to the perceptual identification of unique entities such as the faces of familiar or famous people. Additional studies may examine whether visual recognition of other unique familiar entities, such as landmarks and buildings, may also be affected by right anterior temporal lobe lesions.

Consistent with prior studies, LTLE and LATL was associated with a deficit in naming faces of very famous people,21,46⇓ but not identifying characteristics related to the famous faces. Both left medial and anterior temporal lobe lesions affected naming, but not recognition, of famous faces. The deficit in retrieving proper names of familiar people worsened significantly after surgical resection of the anterior left temporal lobe, consistent with the idea that this cortical region plays a special role in the retrieval of names for unique objects and faces.6 These conclusions, however, must be tempered by the fact that LATL resulted not only in substantial loss of the ability to retrieve names for unique people, but also a worsening of the naming impairment for common objects. Furthermore, there was a relatively strong correlation between impaired retrieval of proper names of famous people and difficulties naming other familiar objects, as well as impaired memory for newly presented words. Our results, therefore, are not completely supportive of the claims that proper nouns processing is subserved by regions different from those involved in processing common nouns, at least within the anterior regions of the left temporal lobe. There are several reports of patients with focal lesions who have disproportionate or exclusive anomia for proper nouns, compared to common nouns.47 This48,49⇓ has led to the proposal that there are neural systems dedicated to processing the names associated with familiar people, which are different from those involved in naming other objects. The exact localization of the neural systems devoted to retrieving names of familiar people is being debated. Our results, in accord with other studies,18 raise questions as to whether the polar region of the left temporal lobe is the sole or primary locus for deficits in proper noun or familiar face naming.

The current results demonstrate a face recognition deficit that is specific to surgical lesions in the right anterior temporal lobe, because patients with right TLE with lesions in the medial temporal lobe demonstrated normal recognition of famous faces when examined before surgery. Our results differ from those of two recent studies that suggested that impaired recognition of famous faces may be present in patients with right TLE before surgery.46,50⇓ One study reported that patients with RTLE were impaired in recognizing whether a pictured face was of a famous person and identifying the person’s occupation from his or her face. Based on these findings, the authors suggested that the right hippocampal system plays a critical role in the acquisition of semantic information about faces. In patients with chronic dysfunction in the right temporal lobe, impaired perceptual processing of faces might combine with a deficit in encoding new faces into long-term memory representations to disrupt acquisition and memory for familiar faces. Although we did not find a significant deficit in famous face recognition in our RTLE sample, our findings do not rule out the possibility that our patients might have had a subtle problem in recognizing familiar faces before ATL, because our test may not have been sensitive enough to detect a mild problem in patients with RTLE before surgery. As indicated, we were relatively liberal in our scoring description of the pictured faces as correct, which may have resulted in missing a mild deficit in the knowledge about these faces in patients with RTLE. Even if our test was insensitive to mild deficits in the recognition of familiar faces before epilepsy surgery, however, this would not detract from our finding of a significant decline from a previously more intact ability for recognizing famous faces after surgery in the right anterior temporal lobe. Testing the same group of patients pre- and post-ATL would be necessary to distinguish more between the role of medial temporal lobe dysfunction in chronic epilepsy vs surgical lesion in the anterior right temporal lobe and deficits recognizing familiar faces.

We found that patients with lesions in the right anterior temporal lobe were impaired in recognizing previously known famous people from visual facial cues, as well as in discriminating and learning new unfamiliar faces. Although we did not test directly for recognition of faces uniquely familiar to individual patients with RATL, based on existing data we might expect that patients with right anterior temporal lobe lesions would have difficulties discriminating personally relevant familiar faces as they do publicly famous people. Clinical experience has shown that after RATL certain patients experience difficulties discriminating among previously familiar people (such as physicians and nurses who had cared for them before surgery) when they must rely solely on visual cues. The ability to distinguish and recognize different people from their faces is a basic skill for social interaction. Chronic problems in the ability to learn and recognize familiar faces might be expected to have detrimental consequences for the everyday functioning and social adjustment of patients with right TLE. Future studies might explore whether some of the reported disturbances in the social and emotional adjustment of patients with RTLE51 are related to their inability to utilize information from faces to select responses appropriate to different people.