Changes in Description Naming for Common and Proper Nouns After Left Anterior Temporal Lobectomy

Abstract

Numerous studies have shown that surgical resection of the left anterior temporal lobe (ATL) is associated with decline in object naming ability (1). In contrast, few studies have examined the effects of left ATL surgery on auditory description naming (ADN) or category-specific naming. Compared to object naming, which loads heavily on visual recognition processes, ADN provides a more specific measure of concept retrieval. The present study examined ADN declines in a large group of patients who were tested before and after left ATL surgery, using a 2 × 2 × 2 factorial manipulation of uniqueness (common vs. proper nouns), taxonomic category (living vs. nonliving things), and time (pre- vs. post-surgery). Significant declines occurred across all categories but were substantially larger for proper living concepts, i.e., famous individuals. The disproportionate decline in proper living noun naming relative to other conditions is consistent with the notion that the left ATL is specialized not only for retrieval of unique entity concepts, but also plays a role in processing social concepts and person-specific features.

1. Introduction

Examination of naming in patients prior to and following brain surgery provides a method for better understanding semantic memory, category-specific naming, and the neural substrate of language (2). Patients undergoing dominant hemisphere anterior temporal lobe (ATL) surgery for medically intractable epilepsy are at risk for language impairments, particularly object naming decline (1). Naming can be assessed using various paradigms differing in modality of presentation, for example using visual confrontation naming or auditory description naming (3, 4). In addition, naming can be evaluated by semantic category or level of uniqueness, for example, living versus nonliving or common versus proper noun retrieval.

Most studies examining naming outcome after left ATL surgery have used a visual object naming paradigm and common (i.e., non-unique) objects. These studies have found that visual object naming, most commonly assessed using the Boston Naming Test (5), often declines after left ATL surgery (1, 6) and that predictors of this decline include left language dominance on fMRI (7), older age at onset of epilepsy (8, 9), higher pre-operative naming scores (8, 10), and absence of medial temporal lobe sclerosis (10, 11).

Fewer studies have examined naming decline using a verbal cue procedure. With an auditory description naming (ADN) test paradigm, the patient provides a common or proper noun in response to a verbal cue such as “jewelry for the finger.” In a series of studies using intraoperative stimulation mapping, Hamberger and coauthors (3) found that left ATL stimulation disrupted ADN whereas posterior temporal stimulation disrupted both ADN and visual object naming. Given this, one might expect ADN to be more sensitive to decline with resection of the left ATL. However, in a follow-up study comparing visual object naming and ADN decline after left ATL surgery (12) visual naming and verbal memory declined while, counter-intuitively, ADN did not decline. It was hypothesized that resection of the hippocampus led to declines in both visual naming and verbal memory, but that the hippocampus was not involved in auditory description naming.

In addition to potential localization differences in naming depending on the stimulus modality used for testing (auditory verbal versus visual object), there is evidence that object concepts are spatially organized by semantic category (13–15). Some patients with ATL pathology, such as those with semantic dementia, herpes simplex encephalitis, or anterior temporal lobe epilepsy, have shown category-specific naming deficits for living things (16–19). In an influential study combining lesion-deficit correlation and functional imaging methods, Damasio et al. showed that patients with left ATL lesions had greater deficits for naming people and animals, whereas posterior left temporal-parietal-occipital lesions were associated with difficulties naming tools, a pattern mirrored in functional activation patterns obtained from healthy participants (15). Another principle proposed to underlie concept organization in the temporal lobe is the distinction between common and unique entities. A number of neuropsychological studies implicate the ATL in retrieval of proper nouns for unique entities, such as famous people and geographic locations (20). Many functional imaging studies have also suggested a specific role for the ATL in processing unique entity concepts (15, 21, 22). Proponents of this view of the ATL note evidence for a posterior-to-anterior organization of the ventral temporal lobe visual object recognition pathway that leads from low-level sensory features to object concept representations (23) and propose that this ascending hierarchy of response selectivity extends anteriorly to representation of unique entities, such as known individuals and unique geographical locations (15, 22).

Prior studies of category and uniqueness effects in naming have almost universally used object pictures as stimuli. To our knowledge, no studies have examined the effects of left ATL lesions on naming in response to auditory verbal cues for different taxonomic categories or levels of uniqueness. The ADN task used extensively by Hamberger and coauthors included only common nouns (24). From a purely practical perspective, given evidence that the ATL may be especially critical for naming unique entities, an ADN test designed to test retrieval of proper names may be more sensitive to the effects of ATL resection. The ADN procedure also offers an interesting alternative method for assessing conceptual knowledge, in that it does not require visual recognition processes and thus provides a more specific measure of concept retrieval compared to the visual object naming paradigm, which may be confounded by categorical effects on perceptual processing.

The present study examines ADN declines in a group of patients tested before and after left ATL resection in a multicenter study. The ADN test (25) includes verbal cues for four different categories: common living (CL), common nonliving (CN), proper living (PL) and proper nonliving (PN), enabling a 2 × 2 × 2 factorial manipulation of taxonomic category, uniqueness level, and time. We hypothesized that larger pre-to-postoperative declines in ADN performance would occur for proper nouns, in keeping with notions of the specificity of the left ATL for unique entities.

2. Materials and method

2.1. Participants

Participants were 52 patients who underwent left anterior temporal lobectomy for medically refractory epilepsy. Participants were enrolled prospectively as part of a multicenter study, Functional MRI in Anterior Temporal Epilepsy Surgery (FATES), aimed at identifying predictors of cognitive outcome in patients undergoing left ATL resection. Patients underwent inpatient video EEG monitoring, MRI, fMRI, and neuropsychological testing as part of a comprehensive work-up to determine their candidacy for epilepsy surgery. Only patients who had a standard left ATL resection were included in the present analysis. Those who underwent hippocampal laser ablation or focal neocortical resections were excluded. All participants were over the age of 17 and fluent speakers of English. Approval for human subjects research was obtained from the Human Research Protection Program at each enrolling institution. Written informed consent was obtained from all participants prior to initiation of the study protocol. The demographic characteristics of the sample are presented in Table 1.

Table 1.

Demographic and Seizure Characteristics of the Sample of 52 Left ATL Epilepsy Patients. Values in parentheses represent standard deviation. F = female, M = male, L = left, R = right, S = symmetric, n/a = no data available.

Sex	26F, 26M
Age	35.7 (12.2)
Education	12.9 (2.2)
Full Scale IQ	91.29 (12.16)
Handedness	44R, 7L, 1A
Age at Epilepsy Onset	20.5 (12.5)
Duration of Epilepsy	15.2 (12.3)
Seizure Frequency (events per month, excluding auras)	8.8 (15.2)
Language Lateralization	37L, 8R, 5S, 2 n/a

2.2. Procedures

Patients underwent a comprehensive neuropsychological battery prior to and approximately 6 months following left temporal lobectomy. Testing included the Boston Naming Test (BNT) (26) and an experimental ADN test based on previously normed stimuli for description naming (25). ADN requires naming in response to a verbal description (e.g., “an animal that builds dams”). The conditions for ADN included common living (CL) entities (e.g., beaver, snail, woodpecker), common nonliving (CN) entities (e.g., football, ring, suitcase), proper living (PL, i.e., animate) entities (e.g., Cher, Obama, Woody the Woodpecker), and proper nonliving (PN) entities (e.g., Chicago, Greece, Eiffel Tower). There were 24 items in each of the 4 conditions (96 trials in total), which were intermixed and presented in one of four random orders. The four ADN conditions were matched on mean accuracy and reaction time obtained from a previous sample of 20 healthy participants (25).

2.3. Statistical analyses

A repeated measures 2×2×2 ANOVA was conducted to examine the effects of category, uniqueness, and surgery as well as their interactions on performance, with follow-up analyses to elucidate the significant main and interaction effects. Raw scores were used in the ANOVAs, and percent change scores were also calculated to further examine differences in decline across conditions. Percent change scores were calculated by subtracting the presurgical raw score from the post-surgical raw score, then dividing this amount by the presurgical raw score. Each set of post-hoc analyses was corrected for multiple comparisons using the Benjamini-Hochberg False Discovery Rate (FDR) (27). Because the pre- and postsurgical distributions of the common noun and proper living conditions as well as the percent change distributions for the common living and proper noun conditions were found to be non-normal using the Shapiro-Wilks test, pairwise comparisons were also performed using Wilcoxon signed ranks tests.

3. Results

3.1. Repeated measures ANOVA results

The results of the 2 × 2 × 2 factorial ANOVA revealed significant main effects of time (F(1,51) = 17.39, p < .001) and uniqueness (F(1,51) = 153.56, p < .0001) as well as significant category x time (F(1,51) = 17.41, p < .001), uniqueness x time (F(1,51) = 13.94, p < .001), category x uniqueness (F(1,51) = 6.01, p = .018), and category x uniqueness x time (F(1,51) = 17.56, p < .001) interaction effects. The main effect of category was not significant (F(1,51) = 1.17, p = .284). To elucidate these effects, pairwise comparisons were conducted for each time point and across time points.

3.2. Differences Across Conditions Pre-Operatively

Table 2 shows the pre- and postoperative scores for each ADN condition, and Table 3 shows the results of the comparisons across conditions. Stronger performance was seen on the CN condition pre-operatively relative to all other conditions. Performance on the CL condition was stronger than performance on either of the proper noun conditions, and performance on the PL condition was stronger than performance on the PN condition. Wilcoxon signed ranks tests yielded the same pattern of results.

Table 2.

Means and standard deviations for the ADN raw scores pre- and post-operatively and the percent change scores as well as the paired t-tests and FDR-corrected p values for the pre- to post-operative comparisons.

Measure	Pre-Operative	Post-Operative	% Change	t	p
Common Living	21.60 (2.57)	20.60 (3.94)	−4.33 (16.34)	2.13	.038
Common Nonliving	22.38 (1.81)	21.73 (2.53)	−2.84 (9.06)	2.38	.028
Proper Living	18.38 (5.16)	14.85 (6.36)	−18.88 (29.57)	5.51	<.0001
Proper Nonliving	16.94 (4.43)	15.89 (4.37)	−3.43 (25.53)	2.44	.028

Table 3.

Paired t values and FDR-corrected p values for the pairwise comparisons between ADN conditions.

Comparison	Pre-Operative		Post-Operative		Change
	t	p	t	p	t	p
Common Living – Common Nonliving	−2.6	0.012	−2.6	0.014	0.73	0.707
Common Living – Proper Living	6.32	<0.0001	9.54	<0.0001	4.49	<0.001
Common Living – Proper Nonliving	8.93	<0.0001	8.88	<0.0001	−0.29	0.867
Common Nonliving – Proper Living	6.94	<0.0001	9.34	<0.0001	4.22	<0.001
Common Nonliving – Proper Nonliving	10.42	<0.0001	12.49	<0.0001	0.17	0.867
Proper Nonliving – Proper Living	−2.88	0.007	1.72	0.091	4.14	<0.001

3.3. Differences across conditions post-operatively

There was also a significant difference in performance across conditions after surgery. As seen before surgery, performance was stronger on CN than all other conditions, and performance on CL was stronger than performance on the proper noun conditions. Post-operatively, the difference in performance across the ADN proper noun conditions was not significant, although there was a trend toward stronger performance on PN relative to PL, which is the reverse of the pattern seen pre-operatively. Similar results were obtained using the Wilcoxon signed ranks test, although the last finding was no longer a trend (p = .130).

3.4. Pre-operative to post-operative decline

Naming performance significantly declined for all conditions (see Table 2). Similar findings were obtained with the Wilcoxon signed ranks test, with the exception that the decline for the CL condition only approached significance (p = .076). As indicated by the significant category x uniqueness x time interaction effect, the magnitude of the decline differed across conditions. Pairwise condition x time ANOVAs revealed significant interaction effects (larger declines) for PL with all other conditions (corrected ps < .0001), while there were no significant interaction effects between the remaining conditions (corrected ps > .492). The same pattern was obtained using pairwise paired t-tests on the percent change scores (see Table 3 and Figure 1) as well as Wilcoxon signed rank tests, such that significantly greater decline was seen in PL performance relative to all other conditions.

Figure 1.

Percent declines by condition for Auditory Description Naming

4. Discussion

The utility of auditory description naming in the neuropsychological evaluation for drugresistant temporal lobe epilepsy has remained unclear. Although initial evidence pointed to a particular role for the left ATL in ADN (3), subsequent data from a few small ATL surgery outcome studies suggested little or no effect on ADN performance from left ATL resection (12, 28). Notably, these ADN tests were restricted to common noun targets. Drawing on evidence that left ATL damage can impair proper noun picture naming more than common noun picture naming, and evidence from functional imaging studies showing proper noun > common noun activation differences in some parts of the ATL, we hypothesized that proper noun ADN would be more affected than common noun ADN by left ATL resection. The results indicated that left ATL resection results in much larger declines in ADN for proper living concepts (i.e., specific animate individuals) compared to other categories (including proper nonliving concepts). Small but significant declines were also observed in the other ADN conditions, which did not differ from each other.

The pattern of results suggests that the left ATL plays a particular role in storing knowledge about specific animate individuals, consistent with claims for a role of the ATL in processing social concepts, people’s names, and specific individual’s faces (29). The ATLs are proposed to contain a person-selective region that is functionally connected with the larger social cognition brain network (30). Studies of nonhuman primates also show that the ATL is a zone for multisensory interactions supporting conspecific recognition via auditory and visual streams involved in identity perception (31). These converging results suggest that the role of the ATL in conceptual processing is better characterized as supporting conspecific knowledge rather than unique entity concepts more generally. A brain specialization for conspecific knowledge also seems likely from an evolutionary perspective, given the central importance of conspecific interactions in everyday life and the development of human culture (32).

One notable advantage of the ADN task is that it eliminates visual object recognition processes as a factor in determining category effects, and therefore provides a more specific assessment of conceptual retrieval. Given the well-documented decline in picture naming that occurs with left ATL resection, it may seem surprising that declines in common noun ADN are relatively small. We hypothesize that this difference may reflect the greater demands on high-level visual object recognition processes in the case of picture naming. That is, ATL resection may produce naming impairment in two distinct ways: by impairing visual object recognition (or the mapping from this stage to a conceptual representation) in the case of picture naming, or by a direct effect on conceptual knowledge stores. The present data suggest that in the latter case, the effect is restricted to a particular conceptual domain (conspecific knowledge) for which the ATL is especially critical.

Although the ADN conditions were matched for difficulty using accuracy and response time data obtained from a normative sample, the left temporal lobe epilepsy patients in the current sample still performed worse preoperatively on the two proper noun conditions (and slightly worse on common nonliving compared to common living items). However, the pattern of change scores across the four conditions is unlikely to be due to baseline differences in task difficulty. At baseline, patients performed significantly worse on the PN condition than on all other conditions (including PL), yet decline on PN was not different from decline on the two common noun conditions and was much less than decline on the PL condition.

A limitation of the present study is that variables beyond category and uniqueness were not included in the analyses. Other factors that may influence cognitive outcome after left temporal lobe surgery such as the presence of hippocampal pathology, age at onset of seizures, and length of resection were not examined in the current study. Directions for future research include examining the inter-relationship between surgical and seizure variables, stimulus modality (auditory vs visual), and various aspects of conceptual knowledge.

Ultimately the clinical goal of understanding language networks in people with epilepsy is to obtain information that improves outcome following epilepsy surgery. In the case of left ATL surgery, understanding the specific processes supported by the ATL, and their differential localizations within the ATL, is a critical step toward this goal. Ideally, precise localization of seizure foci along with better understanding of functional network organization can improve cognitive outcome from epilepsy surgery without compromising seizure freedom.

Highlights.

• Auditory description naming provides a more specific measure of concept retrieval.

• Description naming declines are noted for common and proper, living and nonliving nouns after left anterior temporal resection.

• Naming of proper living concepts (famous individuals) declines disproportionately after left anterior temporal resection.

• The left anterior temporal lobe is specialized not only for retrieval of unique entity concepts, but also plays a role in processing social concepts and person-specific features.