Abstract

Recent evidence suggests that the perirhinal cortex is involved in perception of complex objects with ambiguous features. Anterior regions of the temporal lobes, including the perirhinal cortex as well as lateral cortex, are also thought to play a critical role in semantic memory. To understand how semantic factors might contribute to perceptual discrimination of complex objects, we studied visual object discrimination in patients with semantic dementia (SD)—a neurodegenerative condition characterized by progressive deterioration of semantic knowledge and atrophy to anterior temporal lobes (including perirhinal cortex). In 3 experiments, we assessed discrimination of meaningful (e.g., familiar real-world objects) and novel (e.g., blobs) objects with varying feature ambiguity levels. In a fourth experiment, we compared SD patients with amnesic patients with nonprogressive medial temporal lobe (MTL) lesions and less impaired semantic memory. Across studies, patients with perirhinal damage were impaired at discriminating objects with a high, but not low, degree of feature ambiguity, consistent with previous work indicating a perceptual role for this structure. Stimulus meaningfulness, however, differentially influenced performance in SD patients compared with MTL amnesics, suggesting that perceptual representations of complex objects (dependent upon perirhinal cortex) interact with higher-order abstract conceptual representations, even for tasks with no overt semantic component.

Introduction

There has been considerable debate as to whether structures in the medial temporal lobe (MTL), a set of heavily interconnected structures including the hippocampus and perirhinal cortex, support both memory and perception. A prominent model of memory proposes an exclusive role for the MTL in long-term declarative memory but not in perception or short-term working memory (Squire et al. 2004). There is increasing evidence, however, that the perirhinal cortex is critical for the performance of visual discrimination tasks that have high “feature ambiguity,” meaning that the task is not easily solved by using simple feature information alone (Bussey et al. 2002; Barense et al. 2005, 2007). For example, in a recent study, participants with perirhinal damage were impaired on a series of discriminations between objects with a large number of features in common but performed normally when the objects could be discriminated using simple visual features (Barense et al. 2007). These findings suggest that the perirhinal cortex may be best understood as an extension of the representational hierarchy within the ventral visual stream: rostral inferotemporal cortical regions, including perirhinal cortex, form representations of complex conjunctions of stimulus features, whereas more caudal regions (e.g., V4, TEO) represent the components from which these conjunctions are formed (Desimone and Ungerleider 1989; Riesenhuber and Poggio 1999; Bussey et al. 2002; Barense et al. 2005). Damage to the perirhinal cortex compromises the integrity of these complex conjunctive object representations and thus impairs not only memory but also object perception—should the task tax the ability to represent complex conjunctions of features.

It has also been proposed that the perirhinal cortex may be important for semantic memory (Murray and Bussey 1999). Given its widespread and polymodal afferents (Suzuki and Amaral 1994) and position at the boundary between putative mnemonic and perceptual regions, the perirhinal cortex is well placed to act as an interface between perceptual processing and conceptual knowledge. In support of this idea, some researchers have found evidence to suggest that perirhinal cortex integrates perceptual feature information into higher-level semantic memories of meaningful objects (Taylor et al. 2006). Related to this, studies with semantic dementia (SD), the clinical label given to the temporal variant of frontotemporal dementia (Snowden et al. 1989; Hodges et al. 1992), have highlighted a role for the anterior temporal lobe as an amodal semantic hub thought to support the interactive activation of surface representations across different modalities (e.g., shape, action, and color) (McClelland and Rogers 2003; Rogers et al. 2004; Patterson et al. 2007).

The goal of the current study was to understand whether and how semantic factors influence perceptual processing in visual discrimination. In 3 experiments, we assessed object processing for meaningful (e.g., familiar everyday objects) and unfamiliar (e.g., barcodes, blobs, greebles, and fribbles) stimuli in patients with SD who have anterior temporal lobe atrophy (including perirhinal cortex) and progressive loss of conceptual knowledge. In a fourth experiment, we compared the performance of SD patients with a small group of amnesic patients with nonprogressive MTL lesions who have less impairment to semantic memory.

Semantic Dementia

SD is a neurodegenerative condition that results in striking atrophy to anterior–inferior temporal regions, including the perirhinal cortex (Chan et al. 2001; Galton, Patterson, et al. 2001; Davies et al. 2002, 2004). The disease is characterized by a gradual, crossmodal deterioration of semantic knowledge, with relative sparing of other cognitive abilities such as working memory, syntax, phonology, visuospatial abilities, and nonverbal problem solving. For example, SD patients perform poorly when asked to name pictures of familiar objects and animals, match a spoken word to its appropriate pictorial representation, generate exemplars from a given category, sort words or pictures into categories (e.g., land animals vs. birds), or demonstrate the correct use of objects (Hodges et al. 1992, 1994, 1995, 2000; Bozeat et al. 2000; Rogers et al. 2004). In all of these tasks, the patients’ performance is strongly modulated by word or object familiarity/frequency.

Previous studies of SD provide some support for the view that the perirhinal cortex is necessary for storing and processing representations of complex object stimuli. For example, although SD patients typically perform well on recognition memory for pictures of everyday objects (Graham et al. 2000; Simons et al. 2001, 2002), deficits in recognition memory emerge when SD patients are tested on stimuli with a large number of overlapping features, such as faces (Clague et al. 2005). Furthermore, these impairments extend to face discrimination tasks with no requirement to remember stimuli across trials (Lee et al. 2006, 2007). To date, however, patients with SD have performed well on object perceptual processing tasks, consistent with their good object recognition memory. This may reflect the use of object stimuli that did not possess sufficient feature ambiguity to tax the perirhinal cortex.

Interactions between Semantic Memory and Perception

Theories of visual object recognition are divided regarding the relationship between perceptual and conceptual knowledge. One approach distinguishes between “semantic” representations that store meaning about the world and stored structural descriptions that contain information about the visuospatial structure of objects (Riddoch and Humphreys 1987a, 1987b; Humphreys et al. 1988; Tulving and Schacter 1990; Sheridan and Humphreys 1993). In this view, stored structural descriptions permit the recognition of objects from vision prior to the retrieval of conceptual information about their functional and associative properties. Perceptual representations are thought to be qualitatively distinct from semantic representations. An alternative approach does not consider conceptual knowledge to be wholly separate from the perceptual representations that code modality-specific surface properties. Instead, concepts and meanings are thought to emerge from the interactions of perceptual representations across different modalities (Allport 1985; Warrington and McCarthy 1987; Goldstone and Barsalou 1998; Humphreys and Forde 2001; Barsalou et al. 2003; McClelland and Rogers 2003; Rogers et al. 2004). In this view, modality-specific perceptual representations are also the modality-specific content-bearing parts of the semantic network. Amodal abstract semantic representations then code high-order relations among features and concepts and are structured in a way that allows generalization among similar concepts.

Anterior–temporal regions, which are severely affected early in SD, are considered possible loci for this crossmodal associative conceptual knowledge (McClelland and Rogers 2003; Patterson et al. 2006, 2007; Pobric et al. 2007). An important prediction is that interactions between semantic memory and perceptual representations are critical to normal functioning on tasks that initially appear to be independent of conceptual knowledge. This is especially true when the stimuli have atypical characteristics (e.g., the hump of a camel, the unusual spelling–sound correspondence of “leopard”; Hovius et al. 2003; Rogers et al. 2003, 2007). In a systematic investigation of 6 such tasks (reading aloud, writing to dictation, inflecting verbs, lexical decision, object decision, and delayed copy drawing), Patterson et al. (2006) report a frequency-by-typicality interaction in all tasks and across all patients. In other words, across disparate tasks that have usually been considered to be independent of conceptual knowledge, SD patients consistently performed poorly on low-frequency items with an atypical structure. The authors conclude that this striking consistency in performance is evidence for an interaction between distinct anatomical sites storing perceptual representations and a high-level amodal semantic system within the anterior–temporal lobe. Low-frequency and atypical items are particularly vulnerable to the loss of conceptual knowledge because, having atypical structural features, they require more support from semantic memory.

In summary, there are 2 main theoretical positions regarding the influence of semantic memory on perception: 1) A theory in which perception–recognition processes are independent of semantic processes. On this view, the impairments observed in SD on object decision and other purportedly nonsemantic tasks reflect disruptions to multiple presemantic structural systems and are unrelated to the central loss of semantic memory (Riddoch and Humphreys 1987b; Coltheart 2004; Tyler et al. 2004; Miozzo and Gordon 2005). 2) A theory in which perception–recognition processes interact with semantic processes, so that semantic knowledge impairments negatively influence perceptual processing, particularly for items of low frequency and atypical of their kind (McClelland and Rogers 2003; Rogers et al. 2004; Patterson et al. 2006).

The current work investigates whether visual discrimination in SD is influenced by the semantic meaningfulness of stimuli. It was predicted that SD patients would be impaired on discriminations with a high degree of feature ambiguity for both meaningless and meaningful stimuli, due to perirhinal damage, but that the patients’ degree of impaired performance relative to controls would be exaggerated for meaningful compared with meaningless stimuli, due to impaired conceptual processing.

Experiment 1: Concurrent-Discrimination Learning

Methods: Experiment 1

Participants

Eight SD patients (mean age = 63.2 years, standard deviation [SD] = 6.5 years; mean education = 13.5 years, SD = 3.7 years) participated in this experiment. Twenty healthy individuals (mean age = 62.9 years, SD = 5.5 years; mean education = 13.6 years, SD = 3.2 years) were also recruited. Independent sample t-tests revealed no significant differences between the control group and the patients in terms of age or years of education (all t < 0.12, P > 0.91). All participants described in this paper (Experiments 1–4) provided informed consent, and the study was approved by the Cambridgeshire Health Authority Local Research Ethics Committee, United Kingdom.

All SD patients reported in this paper presented through the Memory Clinic, Addenbrooke's Hospital, Cambridge, United Kingdom, and had been tested longitudinally on an extensive neuropsychological battery. In addition to neuropsychological assessment, each patient was given standard psychiatric rating scales to exclude major psychiatric disorders such as depression and schizophrenia. All patients fulfilled the Lund–Manchester consensus criteria for frontotemporal lobar degeneration (Neary et al. 1998): impaired receptive and expressive content–word vocabulary and impoverished semantic knowledge, with relative preservation of nonverbal reasoning, visuospatial abilities, phonology, syntax, and day-to-day memory (as measured by nonverbal tests of memory). Magnetic resonance imaging (MRI) scans at initial presentation indicated focal atrophy in anterior and inferior temporal lobe regions in all cases.

Lesion Analysis

For 7 of the 8 SD patients, further MRI scans were acquired within 4 months of experimental testing, and in the remaining case, a scan was acquired 11 months prior to testing. To isolate common regions of brain atrophy, scans were warped into Montreal Neurological Institute (MNI) space using SPM5 (Wellcome Department of Functional Neuroscience, London, United Kingdom). Following normalization, the temporal lobes of each patient (both lateral and medial surfaces) from the tip of the temporal pole until the posterior limit of the hippocampus were outlined using MRIcro (Rorden and Brett 2000). These outlines were superimposed on an average T1 MNI brain template, and regions of atrophy for each patient were delineated as regions of interest (ROIs). The individual patient ROIs were then compared via MRIcro to identify the areas of intersection and produce a single ROI depicting the common lesion across all patients (Fig. 1a). In line with previous reports, we found common atrophy in anterior–temporal cortex bilaterally, including the perirhinal cortex, amygdala, anterior hippocampus, and temporal pole (with slightly more damage on the left) (Chan et al. 2001; Galton, Patterson, et al. 2001; Davies et al. 2004; Lee et al. 2006; Noppeney et al. 2007).

Figure 1.

Overlapping regions of atrophy within the temporal lobe are shown for the SD patients in (a) Experiment 1 and (b) Experiments 2–3, superimposed on an average T1 MNI average brain template. (c) Regions of atrophy specific to each patient group (SD cases [across both experiments] shown in blue; MTL cases shown in yellow), as well as damage shared between both groups (in purple). The y-coordinate of each brain slice is shown, and left on the image corresponds to left on the brain.

Figure 1.

Overlapping regions of atrophy within the temporal lobe are shown for the SD patients in (a) Experiment 1 and (b) Experiments 2–3, superimposed on an average T1 MNI average brain template. (c) Regions of atrophy specific to each patient group (SD cases [across both experiments] shown in blue; MTL cases shown in yellow), as well as damage shared between both groups (in purple). The y-coordinate of each brain slice is shown, and left on the image corresponds to left on the brain.

Neuropsychological Battery

The patients’ cognitive abilities were assessed using a series of standardized neuropsychological tests and compared with published normative control data (Table 1). The patients’ performance reflected typical SD deficits: an impairment in semantic knowledge, combined with relative preservation of syntax, phonology, visuospatial abilities, nonverbal problem solving, and aspects of episodic memory (Warrington 1975; Snowden et al. 1989; Hodges et al. 1992; Garrard et al. 1997). For example, as a group, the SD patients were profoundly impaired on measures of semantic memory, although with the usual substantial severity range. By contrast, performance on nonverbal episodic memory was relatively intact.

Table 1

Raw scores of SD patients from Experiment 1 and healthy controls on a range of standard neuropsychological tests

Tests Controlsa mean (SD) Patients mean (SD) NS RS TW WM FB AC AN JM 
MMSE 29.1 (1.0) 22.0 (4.8) 12 18 20 21 23 24 27 28 
Semantic memory 
    Picture naming (/48) 43.6 (2.3) 18.3 (16.7) 20 20 41 48 
    Word–picture matching (/64)b 63.8 (0.4) 44.6 (16.7) NT 22 38 21 60 47 62 62 
    PPT—pictures (/52) 51.2 (1.4) 38.5 (8.2) 27 38 30 38 47 31 48 49 
    Category fluency—3 categoriesc 48.9 (9.8) 7.9 (7.8) NT 24 14 
Episodic memory 
    Rey figure—immediate recall (/36)d 16.8 (5.6) 15.7 (4.4) NT NT NT 12 14.5 11 23 18 
    Rey figure—45-min delayed recall (/36) 15.3 (7.4) 11.3 (5.9) 1.5 11.5 14.5 10.5 10.5 23 13 
Executive and attentional tasks 
    Digit span—forwards 6.8 (0.9) 6.4 (1.2) 
    Digit span—backwards 4.7 (1.2) 4.6 (1.2) 
    WCST categories (/6)b 5.8 (0.5) 5 (1.5) NT NT 
    WCST perseverative errors  4.5 (4.6) 13 NT NT 
    WCST other errors  3.2 (3.7) NT 11 NT 
Visuospatial ability 
    Rey figure—copy (/36) 34.0 (2.9) 33.1 (2.7) 30 29 31 34 36 36 36 33 
    VOSP—dot count (/10) 9.9 (0.3) 9.9 (0.3) 10 10 10 10 10 10 10 
    VOSP—cube analysis (/10) 9.7 (2.5) 9.7 (0.4) NT 10 10 10 10 10 
Tests Controlsa mean (SD) Patients mean (SD) NS RS TW WM FB AC AN JM 
MMSE 29.1 (1.0) 22.0 (4.8) 12 18 20 21 23 24 27 28 
Semantic memory 
    Picture naming (/48) 43.6 (2.3) 18.3 (16.7) 20 20 41 48 
    Word–picture matching (/64)b 63.8 (0.4) 44.6 (16.7) NT 22 38 21 60 47 62 62 
    PPT—pictures (/52) 51.2 (1.4) 38.5 (8.2) 27 38 30 38 47 31 48 49 
    Category fluency—3 categoriesc 48.9 (9.8) 7.9 (7.8) NT 24 14 
Episodic memory 
    Rey figure—immediate recall (/36)d 16.8 (5.6) 15.7 (4.4) NT NT NT 12 14.5 11 23 18 
    Rey figure—45-min delayed recall (/36) 15.3 (7.4) 11.3 (5.9) 1.5 11.5 14.5 10.5 10.5 23 13 
Executive and attentional tasks 
    Digit span—forwards 6.8 (0.9) 6.4 (1.2) 
    Digit span—backwards 4.7 (1.2) 4.6 (1.2) 
    WCST categories (/6)b 5.8 (0.5) 5 (1.5) NT NT 
    WCST perseverative errors  4.5 (4.6) 13 NT NT 
    WCST other errors  3.2 (3.7) NT 11 NT 
Visuospatial ability 
    Rey figure—copy (/36) 34.0 (2.9) 33.1 (2.7) 30 29 31 34 36 36 36 33 
    VOSP—dot count (/10) 9.9 (0.3) 9.9 (0.3) 10 10 10 10 10 10 10 
    VOSP—cube analysis (/10) 9.7 (2.5) 9.7 (0.4) NT 10 10 10 10 10 

Note: MMSE (Folstein et al. 1975); PPT = Pyramids and palm trees test (Howard and Patterson 1992); VOSP = Visual Object and Space Perception Battery (Warrington and James 1991); and NT = not tested.

a

Controls from Hodges and Patterson (1995), n = 24.

b

Controls for WPM and WCST from Graham et al. (2004), n = 19.

c

Control n = 30 for category fluency.

d

Control n = 16 for Rey Immediate (30 s) recall test.

Behavioral Procedure

The current study employed an identical concurrent-discrimination task to that used in patients with static lesions to the MTL (Barense et al. 2005). Participants learned a series of discrimination problems, in which the number of objects to remember was held constant, but the degree of feature ambiguity was varied systematically (Fig. 2). Each discrimination problem consisted of 4 objects, presented in pairs approximately 15 cm apart on a touchscreen. Two of the 4 objects were arbitrarily designated correct (targets) and 2 incorrect (nontargets). In each pair, only 1 object was a target (left vs. right position of the target was randomized). Touching either stimulus resulted in the offset of the stimulus display, accompanied by a pleasant tone if the target was chosen and an unpleasant tone if the nontarget was chosen. The pairs of objects were continuously presented in a pseudorandom order until the participant selected the target objects for 8 consecutive trials. When this criterion of 8 consecutive correct trials was reached, the given task condition would terminate, and after a short delay, the participant would begin another condition.

Figure 2.

Concurrent object discrimination task (Experiment 1). Participants learned 11 discrimination problems, in which the number of objects was held constant, but the degree of feature ambiguity was varied systematically. Each discrimination problem consisted of 4 objects, presented in pairs. Two of the 4 objects were designated correct (targets) and 2 incorrect (nontargets). In each pair, only 1 object was a target (shown here on the left). The pairs of objects were continuously presented in a pseudorandom order until the participant selected the target objects for 8 consecutive trials. Each object consisted of the conjunction of 2 stimulus features: (a) barcode components (individual features shown as letters for illustrative purposes); (b) bug parts (body and legs); shape and fill for “blob” stimulus set; body plan, and coat pattern for the “beast” stimulus set. There were 3 feature ambiguity conditions: minimum ambiguity, in which no features were explicitly ambiguous (i.e., each feature was consistently either part of a target or a nontarget); intermediate ambiguity, in which 1 feature in each object (e.g., legs) was ambiguous; maximum ambiguity, in which all features were ambiguous (i.e., each feature was present in a target and nontarget object). Reproduced with permission from Barense et al. (2005).

Figure 2.

Concurrent object discrimination task (Experiment 1). Participants learned 11 discrimination problems, in which the number of objects was held constant, but the degree of feature ambiguity was varied systematically. Each discrimination problem consisted of 4 objects, presented in pairs. Two of the 4 objects were designated correct (targets) and 2 incorrect (nontargets). In each pair, only 1 object was a target (shown here on the left). The pairs of objects were continuously presented in a pseudorandom order until the participant selected the target objects for 8 consecutive trials. Each object consisted of the conjunction of 2 stimulus features: (a) barcode components (individual features shown as letters for illustrative purposes); (b) bug parts (body and legs); shape and fill for “blob” stimulus set; body plan, and coat pattern for the “beast” stimulus set. There were 3 feature ambiguity conditions: minimum ambiguity, in which no features were explicitly ambiguous (i.e., each feature was consistently either part of a target or a nontarget); intermediate ambiguity, in which 1 feature in each object (e.g., legs) was ambiguous; maximum ambiguity, in which all features were ambiguous (i.e., each feature was present in a target and nontarget object). Reproduced with permission from Barense et al. (2005).

Participants were presented with 4 types of object stimuli with varying levels of semantic familiarity/meaningfulness: “blobs,” “barcodes,” “bugs,” and “beasts.” Each object was the composite of 2 explicitly defined features and the different conditions represented a range of “meaningfulness.” Blobs were completely unfamiliar stimuli consisting of 2 novel shapes shaded with 2 novel textures. Barcodes are generically familiar stimuli from packaged items in shops, but participants have little knowledge to help them “interpret” the features. In this case, the 2 features consisted of 2 different barcode patterns for the left and right halves of the stimuli. Bugs were recognizable as beetle-like insects decomposable into identifiable parts (e.g., legs, head, and body), and varying in the shading of their body and the shape of their legs. Finally, beasts were individual recognizable animals varying in body shape (horse-like or cat-like) and body markings (stripes or spots). Here the items are, like the bugs, recognizable at a superordinate level (i.e., as animals); but each configuration of shape and body markings further yields a more specifically recognizable item—for instance, the cat shape with stripes is recognizable as a “tiger,” the cat shape with spots is a “leopard,” the horse shape with stripes is a “zebra,” and the horse shape with spots is a “pony.”

The logic of the design is as follows: If semantic knowledge does not facilitate perceptual discriminability of highly overlapping items, then patients with SD should perform similarly in all 4 conditions. Like the patients previously reported by Barense et al. (2005), they should show normal performance on minimum feature ambiguity conditions and should be impaired on maximum ambiguity conditions, regardless of the meaningfulness of the stimuli. In contrast, if semantic knowledge about objects facilitates the ability to discriminate perceptually overlapping items, then SD patients, because their semantic knowledge is impaired, may show serious impairments on high-ambiguity items when these are meaningful (bugs and beasts) and may show reduced or even no impairment on high-ambiguity items when these are nonmeaningful (blobs and barcodes).

To adjudicate these 2 hypotheses, we manipulated the level of difficulty of the perceptual discrimination. Thus for the blobs, barcodes, and bugs tasks, we included 1) a minimum feature ambiguity condition, in which no object features were ambiguous; 2) an intermediate feature ambiguity condition, in which half of the features were ambiguous; and 3) a maximum feature ambiguity condition, in which all features were ambiguous so that only the conjunction of the 2 features correctly distinguished targets from nontargets (e.g., black body and yellow legs). In the “beasts” condition, we tested only the minimum and maximum ambiguity conditions. Critically, the number of objects to be remembered was held constant across the different conditions, but as feature ambiguity increased, the necessity to learn conjunctions of features also increased. Thus, the 2 critical variables were 1) whether or not the discrimination required learning the associations of features within individual objects and 2) whether these features/objects were meaningful or novel.

The testing was divided into conditions based on stimulus type, with 11 conditions in total, fully crossing stimulus type (blobs, barcodes, bugs, and beasts) and ambiguity (minimum, intermediate, and maximum), but omitting the beasts–intermediate condition because it was not possible to generate 4 real animals with components that conformed to the requirements of this condition. The order of presentation of the conditions was counterbalanced such that the maximum condition of each stimulus type appeared before the minimum condition of the same stimulus type for half of the subjects and vice versa for the remaining participants. The blob stimuli were administered on a separate occasion to both patients and controls. An average of 6 months separated the 2 testing sessions. All patients were tested on all 4 experimental conditions (barcodes, blobs, bugs, and beasts), except for FB, who was unavailable for testing on the blobs condition.

Results: Experiment 1

To evaluate the performance of the SD group on the 4 concurrent-discrimination tasks, the performance data (shown as errors to criterion in Fig. 3) were subjected to 4 repeated-measures analyses of variance (ANOVAs) based on stimulus type (i.e., blobs, barcodes, bugs, and beasts). A single within-subject factor of ambiguity with 3 levels (or 2, in the case of the beast stimuli) corresponding to the degree of feature ambiguity (i.e., minimum, intermediate, and maximum), and a between-subject factor of subject group (i.e., SD patient vs. matched control), were incorporated. Significant interactions were examined further using independent sample t tests (Bonferroni corrected) to investigate differences between SD patients and controls on each individual condition. These analyses revealed that on discriminations involving novel stimuli with low levels of semantic meaning (i.e., blobs and barcodes), SD patients were unimpaired. The between-subject factor of subject group was not significant (blobs: F(1,25) = 0.73; P = 0.40 and barcodes: F(1,26) = 0.00; P = 0.99), illustrating that overall, the 2 participant groups did not perform differently from one another. The interaction between ambiguity and subject group was also not significant for either of these stimulus types (blobs: F(2,50) = 0.82; P = 0.45 and barcodes: F(2,52) = 1.05; P = 0.36), indicating that the patients did not perform differently from the controls as ambiguity increased.

Figure 3.

Mean errors to criterion (8 consecutive correct responses) for the SD patients and their controls in Experiment 1 for (a) blobs, (b) barcodes, (c) bugs, and (d) beasts are shown. Error bars represent standard error of the mean (SEM). *P < 0.05 and **P < 0.01 (SD patient vs. control).

Figure 3.

Mean errors to criterion (8 consecutive correct responses) for the SD patients and their controls in Experiment 1 for (a) blobs, (b) barcodes, (c) bugs, and (d) beasts are shown. Error bars represent standard error of the mean (SEM). *P < 0.05 and **P < 0.01 (SD patient vs. control).

A different pattern of performance emerged on discriminations of meaningful stimuli (i.e., bugs and beasts). The same analyses revealed a significant effect of subject group (bugs: F(1,26) = 11.86; P < 0.01 and beasts: F(1,26) = 12.86; P < 0.001), indicating that overall, the 2 subject groups performed differently. Importantly, there was a significant interaction between ambiguity and subject group (bugs: F(2,52) = 10.32; P < 0.001 and beasts: F(1,26) = 4.61; P < 0.05), indicating that on discriminations involving familiar stimuli, the increase in feature ambiguity had a differential impact on the 2 subject groups. Independent sample t-tests (Bonferroni corrected) to investigate this interaction further confirmed that the SD patients were impaired relative to their controls on discriminations between familiar stimuli with ambiguous features (intermediate bugs: t(26) = 2.64; P < 0.05; maximum bugs: t(26) = 3.56; P < 0.01; and maximum beasts: t(26) = 2.89; P < 0.05). By contrast, they were not impaired on the minimum bugs condition (t(26) = 0.44; P = 0.99). There was a trend toward an impairment on minimum beasts (t(26) = 2.10; P = 0.09), a result driven primarily by the poor performance of 1 patient.

Difficulty Analyses

To investigate the relationship between task difficulty and feature ambiguity, the data for the SD control group only were subjected to the same 4 repeated-measures ANOVAs described above. These analyses revealed no effect of ambiguity for the blob and bug stimuli (blobs: F(2,38) = 0.38; P = 0.69 and bugs: Greenhouse–Geisser corrected F(1.5,28.9) = 2.42; P = 0.12) but an effect of ambiguity for the barcode and beast stimuli (barcodes: F(2,38) = 4.05; P < 0.05 and beasts: F(1,19) = 17.77; P < 0.001). “Post hoc” analyses (Bonferroni-corrected t-tests) were then performed on the performance data of the control group across the different conditions within each stimulus type. Only 2 of the comparisons (minimum barcodes vs. maximum barcodes and minimum beasts vs. maximum beasts) were significant (both P < 0.05). For the large majority of the comparisons (8 of 10), however, there was no difference in difficulty across conditions (all adjusted P between 1 and 0.25). Furthermore, given that the SD patients performed normally on the most difficult class of discriminations (barcodes), it seems extremely unlikely that differences in task difficulty can account for the observed pattern of results in the SD patient group. Instead, the most likely explanation for the data presented here are that the SD patients are impaired on discriminations of meaningful stimuli with a high degree of feature ambiguity.

Discussion: Experiment 1

Using an object discrimination paradigm that is sensitive to perirhinal cortex damage in monkeys (Bussey et al. 2002) and humans (Barense et al. 2005), the present investigation confirmed the critical influence of feature ambiguity on learning object discriminations in patients with SD. As predicted, SD patients performed normally on all minimum feature ambiguity conditions. The patients were severely impaired, however, on learning conjunctions of features comprising meaningful stimuli (bugs and beasts). However, unlike the amnesic patients with focal perirhinal damage from Barense et al. (2005), SD cases were unimpaired on all discriminations involving novel stimulus items (blobs and barcodes). The bugs and beasts portion of this experiment, therefore, replicates previous findings from nonhuman primates (Bussey et al. 2002) and patients with nonprogressive lesions to the MTL (Barense et al. 2005) in a larger group of participants with perirhinal involvement. The observed pattern of performance seen in the SD patients on the novel stimuli (blobs and barcodes), however, poses a puzzle. Why should relatively focal damage to perirhinal cortex (with other MTL structures) seriously impair performance on this task across all stimulus types (Barense et al. 2005), whereas the pattern of atophy observed in SD spares performance on novel stimuli?

One hypothesis is that the neuroanatomical distribution of pathology differs in the 2 patient groups. Although the perirhinal damage in SD is more extensive in anterior regions of the temporal cortex (Davies et al. 2004), the MTL patients studied by Barense et al. (2005) had pronounced damage along the rostral-to-caudal extent of this region. It is possible, then, that patients with SD do have more subtle object discrimination impairments but perform well on the novel object discrimination tasks because these are not sufficiently challenging. On this view, patients with SD should show impairments with object discrimination tasks involving novel stimuli, so long as stimuli with a higher degree of feature ambiguity are used.

A second possibility is that the pattern of impairment on novel stimuli observed in MTL patients is at least partially attributable to the memory demands of the concurrent-discrimination task, rather than—or in addition to—perceptual impairments caused by perirhinal damage. In contrast to the SD cases studied here, the MTL patients described by Barense et al. (2005) were densely amnesic. We demonstrated that the pattern of impairment was not solely attributable to amnesia, because amnesic patients with damage limited to the hippocampus were not impaired on these tasks (Barense et al. 2005). It remains possible, however, that impairment on the high-ambiguity novel stimuli arises from a combination of perceptual impairment (following perirhinal damage) plus amnesia. On this view, patients with SD—because they are not densely amnesic—should not show impairments on object discrimination tasks that use novel stimuli with very high degrees of feature ambiguity.

A third hypothesis is that the impairments reported by Barense et al. (2005) are mainly attributable to damage in more posterior parts of the perirhinal cortex but that more anterior regions are less involved in the concurrent-discrimination task. For instance, it may be that conjunctions of features that have been very robustly stored in long-term memory—specifically, sets of features that denote familiar classes of objects, like bugs and beasts—are supported by more anterior regions of cortex, whereas newly learned conjunctions are initially supported by more posterior regions. On this view, patients with SD, whose damage is primarily to anterior regions, will not show the same novel object discrimination deficits observed in the MTL group but will be impaired when the same tasks involve familiar objects.

Experiments 2 and 3 were designed to adjudicate these hypotheses. We assessed object discrimination using a nonmnemonic task known to depend on perirhinal cortex (Buckley et al. 2001) and employing both real-world and novel stimuli with a much higher degree of feature ambiguity than in Experiment 1. In these experiments, minimal demands are placed on learning and memory, so amnesia should not influence performance. Very high demands, however, are placed on visual perception and discrimination. Moreover, we manipulated whether the stimuli are familiar or novel objects. If patients with SD have impairments similar to those with MTL damage, but subtler in form (Hypothesis 1), they should show deficits for both familiar and novel items in these tasks. If the previously reported deficits in MTL cases are partly attributable to the dense amnesia observed in these cases (Hypothesis 2), then patients with SD should perform well in these nonmnemonic tasks for both real and familiar items, even when feature ambiguity is very high. If anterior regions code feature conjunctions that correspond to familiar items, whereas more posterior regions code newly learned feature conjunctions (Hypothesis 3), then patients with SD should show a pattern of performance similar to that observed in Experiment 1: significant impairment for tests with familiar items but spared performance for tests with novel items.

Experiments 2–3: Object Oddity Discriminations

Methods: Experiments 2–3

Participants

Seven SD patients (mean age = 63.3 years, SD = 7.6 years; mean education = 12.3 years, SD = 2.1 years) participated in Experiments 2–3. Four of the patients are the same as those reported in Experiment 1 (ca. 18 months separated Experiment 1 from Experiments 2–3). Sixteen healthy individuals (mean age = 64.1 years, SD = 7.4 years; mean education = 13.8 years, SD = 3.0 years) were also recruited. Independent sample t-tests revealed no significant differences between the control group and patients in terms of age (all t < 1.21, P > 0.24).

Lesion Analysis

Structural MRI scans were available for 6 of the 7 SD patients. Five of these were all acquired within 3 months of experimental testing. The sixth scan was acquired 1 year before testing but was unusable for the lesion analysis due to extreme motion artifacts. Visual inspection of this scan, however, revealed focal atrophy in anterior and inferior temporal regions. An identical lesion analysis was performed to that described in Experiment 1, and again, we found common damage in anterior temporal cortex bilaterally, including the perirhinal cortex, amygdala, anterior hippocampus, and temporal pole (with slightly more left-sided damage) (Fig. 1b).

Neuropsychological Battery

The patients’ cognitive abilities were assessed using a series of standardized neuropsychological tests (Table 2) and reflected typical SD deficits. Due to his extremely poor comprehension, 1 patient (RS) was not administered any neuropsychological testing beyond the mini-mental state examination (MMSE). Importantly, the experimental tests reported here contained an extensive visual practice round, which ensured that he had understood the task instructions. Exclusion of this patient does not affect the results reported here, and interestingly, his performance on the experimental tests always placed him among the top 3 patients.

Table 2

Raw scores of SD patients from Experiments 2–3 and healthy controls on a range of standard neuropsychological tests

Tests Controlsa mean (SD) Patients mean (SD) RS NS WM EAD IB JM DG 
MMSE 29.1 (1.0) 16.1 (7.9) 19 20 20 21 23 
Semantic memory 
    Picture naming (/48) 43.6 (2.3) 11.0 (9.6) NT 10 25 20 
    Word–picture matching (/64)b 63.7 (0.5) 27.7 (21.5) NT 13 30 18 59 45 
    CCT—pictures (/64)c 58.4 (3.4) 38.3 (17.8) NT NT 25 NT 21 56 51 
Episodic memory 
    Rey figure—immediate recall (/36)d 16.8 (5.6) 9.2 (5.9) NT NT 15 11 14 
    Benton visual retention (multiple choice) (/15)d 13.1 (1.9) 13.3 (1.0) NT NT 14 13 14 12 NT 
Executive and attentional tasks 
    Digit span—forwards 6.8 (0.9) 6.0 (1.9) NT NT 
    Digit span—backwards 4.7 (1.2) 4.0 (1.2) NT NT 
Visuospatial ability 
    Rey figure—copy (/36) 34.0 (2.9) 34.3 (1.0) NT 34 35 33 34 36 34 
    VOSP—dot count (/10) 9.9 (0.3) 9.8 (0.4) NT 10 10 10 10 10 
    VOSP—position discrimination (/20)b 19.8 (0.6) 19.2 (0.8) NT 18 19 20 19 20 19 
Tests Controlsa mean (SD) Patients mean (SD) RS NS WM EAD IB JM DG 
MMSE 29.1 (1.0) 16.1 (7.9) 19 20 20 21 23 
Semantic memory 
    Picture naming (/48) 43.6 (2.3) 11.0 (9.6) NT 10 25 20 
    Word–picture matching (/64)b 63.7 (0.5) 27.7 (21.5) NT 13 30 18 59 45 
    CCT—pictures (/64)c 58.4 (3.4) 38.3 (17.8) NT NT 25 NT 21 56 51 
Episodic memory 
    Rey figure—immediate recall (/36)d 16.8 (5.6) 9.2 (5.9) NT NT 15 11 14 
    Benton visual retention (multiple choice) (/15)d 13.1 (1.9) 13.3 (1.0) NT NT 14 13 14 12 NT 
Executive and attentional tasks 
    Digit span—forwards 6.8 (0.9) 6.0 (1.9) NT NT 
    Digit span—backwards 4.7 (1.2) 4.0 (1.2) NT NT 
Visuospatial ability 
    Rey figure—copy (/36) 34.0 (2.9) 34.3 (1.0) NT 34 35 33 34 36 34 
    VOSP—dot count (/10) 9.9 (0.3) 9.8 (0.4) NT 10 10 10 10 10 
    VOSP—position discrimination (/20)b 19.8 (0.6) 19.2 (0.8) NT 18 19 20 19 20 19 

Note: MMSE (Folstein et al. 1975); CCT = Camel and cactus test (Bozeat et al. 2000); VOSP = Visual Object and Space Perception Battery (Warrington and James 1991); and NT = not tested.

a

Controls from Hodges and Patterson (1995), n = 24.

b

Controls for WPM and VOSP position discrimination from Bozeat et al. (2000), n = 14.

c

Controls for the CCT from Lambon Ralph et al. (2001), n = 15.

d

Control n = 16 for Rey Immediate (30 s) recall and Benton Visual Retention Test (Moses 1986).

Behavioral Procedure

The behavioral procedure was identical to that used in patients with static MTL lesions (Barense et al. 2007). All tasks were based on an oddity paradigm in which the subjects were instructed to select the “odd one out” from an array of simultaneously presented stimuli, as quickly but as accurately as possible (e.g., Buckley et al. 2001; Lee, Buckley et al. 2005). All stimuli were trial unique. During the test, touching any item on the touchscreen resulted in the offset of the stimulus display and the onset of the next trial. No feedback was given. Depending on the willingness of each individual, Experiments 2 and 3 were either administered in the same session or in different sessions separated by approximately 1 week. Half of the participants performed Experiment 2 prior to Experiment 3 and vice versa. One SD patient (EAD) could not be tested on the size and color conditions in Experiment 3.

Experiment 2: Feature Ambiguity Object Oddity

Participants were presented with an array of 7 stimuli and were asked to identify the object that did not have an identical pair (i.e., in each array, there were 3 sets of “identical twins” and 1 odd one out). This adaptation of the typical oddity paradigm reported in previous studies (e.g., Buckley et al. 2001; Lee, Buckley et al. 2005) allowed for the systematic manipulation of feature ambiguity across objects (see Fig. 4a). The stimuli were “fribbles” (Williams and Simons 2000), novel objects composed of a main body and 4 appendages (Fig. 4a). There were 12 categories (or “species”) of fribbles in total. Within a species, all fribbles consisted of the same main body, but each of the 4 appendages had 3 possible values.

Figure 4.

Object oddity tasks (Experiments 2–3). (a) A fribble comprises a central body and 4 attached features. The overlap of these features across fribbles was varied according to the letter schematic (the correct answer [i.e., the fribble without an identical pair] is shown in red). Representative trials from the (b) minimum, (c) intermediate, and (d) maximum conditions in Experiment 2. Representative trials from Experiment 3 (correct answer is located in the bottom left corner): (e) high- (greebles from same family and gender) and (f) low- (greebles from different families) ambiguity greebles, (g) high- and (h) low-ambiguity familiar objects, (i) size. and (j) color. To minimize mnemonic demands, all pictures within a given trial were presented simultaneously, and all stimuli are trial unique. Reproduced with permission from Barense et al. (2007).

Figure 4.

Object oddity tasks (Experiments 2–3). (a) A fribble comprises a central body and 4 attached features. The overlap of these features across fribbles was varied according to the letter schematic (the correct answer [i.e., the fribble without an identical pair] is shown in red). Representative trials from the (b) minimum, (c) intermediate, and (d) maximum conditions in Experiment 2. Representative trials from Experiment 3 (correct answer is located in the bottom left corner): (e) high- (greebles from same family and gender) and (f) low- (greebles from different families) ambiguity greebles, (g) high- and (h) low-ambiguity familiar objects, (i) size. and (j) color. To minimize mnemonic demands, all pictures within a given trial were presented simultaneously, and all stimuli are trial unique. Reproduced with permission from Barense et al. (2007).

As demonstrated in Figure 4b–d, each trial could be 1 of 3 different levels of perceptual discrimination: 1) minimum feature ambiguity (all features were unique to the odd-one-out and paired items), 2) intermediate feature ambiguity (2 features were held constant across all fribbles, whereas the remaining 2 appendages maximally overlapped across all fribbles), and 3) maximum feature ambiguity, in which all appendages were maximally overlapping (i.e., every feature was present on either 3 or 4 fribbles). Thus, in both the maximum and the intermediate conditions, it was only the conjunction of 2 features that correctly distinguished the odd one out from the remaining pairs. By contrast, the minimum ambiguity condition could be solved on the basis of a single feature (i.e., by comparing one appendage).

A short practice of 5 easy trials (2 trials that consisted of letters only to convey the concept of the task and 3 trials using fribbles from different species with different bodies) was given before commencing the experiment. The actual experiment consisted of 108 trials, split into 2 blocks of 54 trials each with a short break between blocks. The different trial types were randomized within a block, and an equal number of trials from the minimum, intermediate, and maximum trials were given (i.e., 36 trials per condition). The position of the odd one out was counterbalanced, and the positions of the 6 paired items were randomly determined. The trials were then checked and altered where necessary to ensure that no trial had 2 sets of identical fribble twins adjacent to one another.

Experiment 3: Familiar and Novel-Object Oddity

As the intermediate and maximum ambiguity conditions in Experiment 2 were significantly more difficult than the minimum ambiguity condition for controls (see Fig. 5a, both P < 0.001), we administered a second series of oddity discrimination tasks with 2 difficulty-matched control conditions. The task administration was identical to that reported in Experiment 2, except that only 4 items were presented per trial (similar to an oddity paradigm reported previously, Buckley et al. 2001; Lee, Buckley et al. 2005, Fig. 4e–j). Each trial included 3 pictures of the same stimulus and 1 picture of a different stimulus. Participants were instructed to choose the odd one out. There were 6 tasks in total: low-ambiguity familiar objects, high-ambiguity familiar objects, low-ambiguity greebles, high-ambiguity greebles, size, and color. The 4 object tasks were presented in a counterbalanced order across all subjects. The size and color tasks were always presented after these 4 conditions. Thirty-five trials were administered for the familiar object, greeble, and size tasks, and 65 trials for the color task. The order of the trials was fully randomized, and the position of the odd one out was counterbalanced across trials. It is important to emphasize that for each trial the stimuli were presented simultaneously, thus minimizing mnemonic demands of the task. Furthermore, to avoid any confounding effects of memory for previously viewed stimuli, all items were trial unique. Any deficits on these tasks are therefore unlikely to be explained by difficulties in learning and remembering stimuli across trials.

Figure 5.

Mean proportion correct for the SD patients and their controls for Experiments 2 and 3 (a,b, respectively). Error bars represent SEM. *P < 0.05 and **P < 0.01 (SD patient vs. control).

Figure 5.

Mean proportion correct for the SD patients and their controls for Experiments 2 and 3 (a,b, respectively). Error bars represent SEM. *P < 0.05 and **P < 0.01 (SD patient vs. control).

Greeble Tasks

Four pictures of “greebles” (e.g., Gauthier and Tarr 1997) were presented for each trial. Each greeble was rotated either 0, 90, 180, or 270 deg from the upright position. Within the greeble tasks, there were 2 different conditions: high and low ambiguity (Fig. 4e–f). The criteria for the high-ambiguity task were that the greebles be from the same family, gender, and symmetry (i.e., asymmetrical vs. symmetrical). Within those criteria, the greebles for each trial were selected to produce the maximum amount of possible feature overlap between the odd one out and the foils, while also matching the difficulty of this condition to that of the high-ambiguity familiar object oddity task. Difficulty was equated through a series of behavioral pilot experiments. The criterion for the low-ambiguity task was that the greebles be from different families. The greebles could be of either the same or different gender or of different symmetry.

Familiar Object Tasks

Four images of objects common to everyday life were presented in each trial, and each photograph was taken from 4 different nonspecific orientations. Objects were collected from the Hemera Photo-Objects Image Collection (Volumes 1–3). As with the greeble tasks, there were 2 conditions: high and low ambiguity (Fig. 4g–h). Unlike for the fribble and greeble tasks, the level of ambiguity was determined subjectively, with extreme care taken to ensure that within a high-ambiguity trial, the objects shared a high number of overlapping features. By contrast, within a low-ambiguity trial, the objects were from the same overall category (e.g., cars), but the objects were easily differentiated on the basis of a single, obvious feature. Furthermore, the stimulus types were matched across the low- and high-ambiguity conditions (e.g., there was a high and a low trial comprised of cars, a high and a low trial comprised of stereos).

Difficult Control Tasks: Size and Color

We designed 2 control tasks that were as difficult as the critical high-ambiguity discriminations described above but could be solved on the basis of a single feature alone (i.e., either size or color) and not require processing complex conjunctions of object features. For the size control task, 4 black squares were presented in each trial (Fig. 4i). The length of each side was randomly varied from 67 to 247 pixels, and the size of each square was trial unique. In each trial, 3 squares were of identical size and a fourth was either smaller or larger. The size difference varied between 9 and 15 pixels. The squares’ positions were jittered slightly so that the edges did not line up along vertical or horizontal planes. For the color control task, 4 colored squares of dimensions 425 × 275 pixels were shown for each trial (Fig. 4j). In each trial, 3 squares were of an identical color, and 1 was of a different color. The proportion of green and red that each color contained was varied, and the “blue” dimension was held constant at 0%. Each color was trial unique, and luminance was equated across all 4 squares.

Results: Experiments 2–3

Repeated-measures ANOVAs were conducted on the accuracy scores (as measured by percent correct). The performance accuracy data from Experiment 2 (shown as proportion correct in Fig. 5a) were subjected to a repeated-measures ANOVA with a within-subject factor of “ambiguity” (i.e., minimum, intermediate, and maximum) and a between-subject factor of “subject group” (i.e., SD patient vs. matched control). In Experiment 3, to assess the effects of 1) familiar versus novel objects and 2) high versus low ambiguity on performance on the 4 choice oddity conditions, the performance accuracy data (Fig. 5b) were subjected to a repeated-measures ANOVA. The same between-subject factor of subject group was incorporated, along with 2 within-subject factors of “familiarity,” with 2 levels corresponding to the semantic meaningfulness of the stimuli (i.e., greebles vs. objects) and ambiguity, with 2 levels corresponding to the ambiguity of the stimuli (i.e., low vs. high). Significant interactions were examined further using independent sample t tests (Bonferroni corrected) to investigate differences between SD patients and controls on each individual condition. As a variable of interest (feature ambiguity) was consistent across both Experiments 2 and 3, the results for these 2 experiments are described together below. Additional analyses on the accuracy data from Experiment 3 (i.e., the effect of familiarity and performance on the size and color conditions) are described separately.

Effect of Feature Ambiguity across Experiments 2 and 3

The repeated-measures ANOVAs described above revealed a significant between-subject factor of subject group (Experiment 2: F(1,21) = 21.33; P < 0.001, Experiment 3: F(1,21) = 32.93; P < 0.001), indicating that overall, the 2 subject groups performed differently from one another. There was also a significant interaction between ambiguity and subject group (Experiment 2: F(2,42) = 22.72; P < 0.001 and Experiment 3: F(1,21) = 36.53; P < 0.001), revealing a differential impact of ambiguity on the 2 subject groups. Independent sample t tests (Bonferroni-corrected) revealed that SD patients performed significantly worse than controls on the intermediate and maximum ambiguity conditions in Experiment 2 (t(21) = 6.52; P < 0.001 and t(21) = 2.62; P < 0.05, respectively) and the high-ambiguity conditions in Experiment 3 (greebles: t(21) = 5.71; P < 0.001 and objects: t(21) = 4.78; P < 0.001). By contrast, there was no difference between the groups on the minimum ambiguity condition in Experiment 2 (t(21) = 1.65; P = 0.33) or on the low-ambiguity conditions in Experiment 3 (greebles: t(21) = 0.52; P = 0.99 and objects: t(21) = −0.25; P = 0.99), confirming that the SD patients had a selective deficit on discriminations between objects with ambiguous features.

Effect of Stimulus Familiarity in Experiment 3 (Novel Greebles vs. Familiar Objects)

In Experiment 3, the repeated-measures ANOVA described above revealed no effect of familiarity (i.e., greebles vs. familiar objects, F(1,21) = 2.25; P = 0.15), no interaction between familiarity and subject group (F(1,21) = 0.03; P = 0.87) and no interaction between familiarity, ambiguity, and subject group (F(1,21) = 0.01; P = 0.93). Thus, these results indicate that the SD group was selectively impaired as ambiguity increased but not as familiarity increased. Furthermore, the SD group was not more impaired, relative to controls, on high-ambiguity familiar than on high-ambiguity novel discriminations, as might have been predicted from Experiment 1.

Control Tasks in Experiment 3: Color and Size

Independent sample t tests revealed no significant differences between the SD patients and controls on either control task (color: t(20) = 0.75; P = 0.46 and size: t(20) = −0.86; P = 0.40).

Difficulty Analyses

The size and color control tasks were designed to provide discrimination tasks that were as difficult as the critical high-ambiguity discriminations. To investigate whether this difficulty manipulation was successful, the data from all control participants in Experiment 3 were subjected to a repeated-measures ANOVA. A within-subject factor of “task” (corresponding to all 6 oddity tasks) was included. This showed a significant effect of task (Greenhouse–Geisser corrected F(2.9,43.9) = 29.25; P < 0.001). Post hoc t tests (Bonferroni-corrected) revealed that the control tasks were either as difficult (all P > 0.66) or approached being more difficult (color vs. high-ambiguity object, P = 0.08) than the high-ambiguity greeble and object conditions. Performance on the high object and greeble conditions was matched (P > 0.99). Given that controls found the control conditions as difficult as the high object and greeble conditions, a pattern not true of the SD group, the observed results cannot easily be explained by task difficulty.

Discussion: Experiments 2–3

Consistent with the findings reported in other patient populations with perirhinal damage, patients with SD were impaired on object oddity judgments with a high degree of feature ambiguity but performed normally if the discriminations could be solved on the basis of a single feature. This pattern of impairment is similar to that reported for learning discriminations of familiar stimuli (Experiment 1) but extends the deficit to a task with little or no learning component and to discriminations involving novel stimuli with low levels of semantic meaning.

Experiments 2 and 3 were designed to address the discrepancy observed between the SD cases in Experiment 1 and previously reported findings in amnesic patients with nonprogressive lesions to the MTL (Barense et al. 2005): In Experiment 1, patients with SD were impaired on familiar, but not novel, object discriminations, whereas the focal MTL cases reported by Barense et al. (2005) were impaired on both novel and familiar discriminations. Here in Experiments 2–3, when objects with a higher degree of feature ambiguity were used, SD patients were impaired on the novel fribble discriminations (Experiment 2) and demonstrated equivalent impairments on the high-ambiguity greeble and familiar object oddity (Experiment 3). These findings suggest that the perceptual deficits observed in SD may be subtler than those in the MTL cases and that more complex objects with a high number of overlapping features (e.g., greater than 2) are required to elicit robust perceptual impairments for novel stimuli in SD: The objects from Experiment 1 contained 2 features that systematically overlapped, whereas the fribbles had 4 manipulated features. Although the features on the greebles and familiar objects were not systematically quantified, these objects were far more complex than either the fribbles or the objects used in Experiment 1.

An alternative (and not mutually exclusive) interpretation is that the SD patients in Experiments 2–3 had more extensive perirhinal damage than those in Experiment 1 and that this additional damage was necessary to elicit deficits on novel discriminations. The patients reported here were at a more advanced stage of the disease (average picture naming score of 11.0 and average word–picture matching score of 27.7 in Experiments 2–3 compared with 18.3 and 44.6, respectively, in Experiment 1). Additionally, 4 of the patients were identical across Experiment 1 and Experiments 2–3, and in the intervening 2 years between the different studies, the general cognitive functioning of these individuals had substantially deteriorated (see Tables 1and 2). Given the progressive nature of the disease, the amount of perirhinal cortex damage in these individuals had also increased, especially in more posterior regions (Chan et al. 2001; Galton, Patterson, et al. 2001; Davies et al. 2002, 2004). This additional perirhinal damage may have been sufficient to cause the robust deficits in processing novel objects reported here.

In summary, across the different experiments, we also observed differential influences of stimulus familiarity. In Experiment 3, SD patients were impaired on discriminations of both meaningful and novel items, whereas in Experiment 1, SD patients were impaired relative to controls at learning conjunctions of meaningful, but not novel, object features. Interestingly, in previous experiments, we have reported that patients with static lesions to the MTL showed the opposite pattern, with better performance on discriminations involving meaningful, compared with novel stimuli on both mnemonic and perceptual tasks (Barense et al. 2005, 2007). For a more systematic comparison of the performance across these different patient groups, we performed a reanalysis of the SD data described in Experiments 1 and 3 with previously published data from patients with static MTL lesions (Barense et al. 2005, 2007).

Experiment 4: Comparison with MTL Amnesics

Methods: Experiment 4

Participants and Behavioral Procedure

In addition to the SD patients described in the above experiments, we previously tested 3 patients with focal lesions to the MTL (including the perirhinal cortex) on an identical series of object discrimination tests to those described in Experiments 1 and 3 (data originally reported in Barense et al. 2005, 2007). Of the 3 patients in the MTL group, 2 were viral encephalitis cases, and the third had experienced traumatic intracerebral bleeding. The average age was 68.0 years (SD = 8.3 years) for the concurrent-discrimination learning task (Experiment 1) and 69.8 years (SD = 8.2 years) for the oddity judgment tasks (Experiment 3). The average education level was 10.3 years (SD = 1.5 years). The control participants reported in the present study were used. All controls were age and education matched to both the SD and MTL patient groups (all t < 1.9; P > 0.1).

The neuropsychological profile of the MTL group has been described in detail elsewhere (Barense et al. 2005, 2007). In summary, although both diseases affect anterior temporal lobe structures, each causes a very different profile of cognitive impairment. As demonstrated in Tables 1 and 2, SD results in a progressive loss of semantic knowledge, with other cognitive domains largely intact. By contrast, cases with static MTL lesions generally demonstrate profoundly impaired episodic memory but only mildly impaired semantic memory (see Table 1 in Barense et al. 2005, 2007). For example, the MTL patients were very impaired on measures of episodic recall, such as delayed recall of the Rey complex figure (mean raw score = 3.8, SD = 3.5) and logical memory (WMS-R, Stories 1 and 2: immediate recall mean raw score = 18.0, SD = 9.5; delayed recall mean raw score = 2.3, SD = 2.1). By contrast, they were only mildly impaired on tests of semantic memory, including Word–Picture matching (mean raw score = 56.7, SD = 2.5) and the pyramids and palm trees test (pictures, mean raw score = 46.7, SD = 2.1). Visuoperceptual performance as measured by the Rey complex figure copy (mean raw score = 33.2, SD = 2.8) and visual object space perception battery (passed all object and space tests) was within the normal control range.

Lesion Analysis

Structural MRI scans were acquired for all 3 MTL focal lesion patients, but an electronic version of one patient's scan was not available, and thus, only 2 scans were available for the lesion analysis here. Previous structural analysis of the missing scan using an established rating scale validated against volumetric measures (Galton, Gomez-Anson, et al. 2001; Barense et al. 2005; Lee, Buckley et al. 2005; Graham et al. 2006) indicated that the profile of damage in this individual was indistinguishable from the other MTL patients (see Barense et al. 2007). To isolate the common regions of brain atrophy in the remaining patients, an identical lesion analysis to that used in Experiments 1–3 was performed. This produced an ROI of the lesion common to the MTL patients. To compare brain damage across MTL and SD patients, the common SD lesion (i.e., the intersection of the individual patient lesion ROIs collapsed across Experiments 1–3; Fig. 1a,b), and the common MTL lesion was overlaid on an average T1 MNI brain template. Using MRIcro, the SD group ROI (Fig. 1a,b) was then subtracted from the MTL group ROI to produce a third ROI image that highlighted the regions of damage specific to each patient group and the areas of atrophy that were common to both groups (Fig. 1c). Consistent with previous findings (Noppeney et al. 2007), both patient groups had overlapping damage in anterior temporal regions, including the temporal pole and the perirhinal cortex bilaterally. The MTL group, however, had more damage to the right temporal pole, perirhinal cortex, and posterior hippocampus bilaterally. In general, the MTL patients’ damage extended more posteriorly than did the damage in the SD cases.

Results and Discussion: Experiment 4

Concurrent Object Discrimination (Experiment 1)

The performance of the 3 groups across all maximum ambiguity discriminations of novel stimuli (i.e., total number of errors committed on both maximum blobs and barcodes) was compared with their performance across all familiar stimuli (i.e., total number of errors committed on both maximum bugs and beasts) using a repeated-measures ANOVA with a within-subject factor of stimulus familiarity (i.e., novel vs. familiar) and a between-subject factor of subject group (i.e., SD patients, MTL patients, and controls) (Fig. 6a). This confirmed a significant effect of group (F(2,27) = 16.00; P < 0.001), a significant effect of familiarity (F(1,27) = 21.27; P < 0.001), and a significant interaction between the 2 (F(2,27) = 12.07; P < 0.001). Two one-way ANOVAs to separately compare performance on the familiar and novel discriminations revealed that the 3 groups performed differently on both novel and familiar discriminations (both F > 10.97; P < 0.001). Post hoc analyses (Bonferroni-corrected t tests) revealed that, although the MTL participants were impaired relative to controls on both novel and familiar discriminations (both P < 0.001), the SD patients were impaired on familiar (P < 0.05) but not novel (P > 0.99) discriminations. When the 2 patient groups were compared with each other, the MTL patients performed significantly worse than the SD group on novel discriminations (P < 0.001), and there was a trend toward worse performance in the MTL group on familiar discriminations (P = 0.09). Given the small sample size of the MTL group, we also compared each individual MTL patient with the control and SD groups. Each of the 3 cases fell more than 2 standard deviation outside the control mean for both the familiar and novel discriminations. Relative to the SD group, all 3 cases fell more than 2 standard deviation outside the SD mean for novel discriminations, whereas for familiar discriminations, only 1 MTL case was more than 2 standard deviation outside the SD mean. Thus, the above results do not reflect an artifact of averaging in this small sample size.

Figure 6.

Comparison across patient groups (Experiment 4). (a) Mean errors to criterion on the maximum ambiguity conditions of novel (blobs and barcodes) and familiar (bugs and beasts) concurrent discriminations for controls, SD patients, and patients with focal MTL lesions. (b) Proportion of errors committed on the high-ambiguity conditions of novel (greebles) and familiar (objects) oddity discriminations for controls, SD patients, and patients with focal MTL lesions. *P < 0.05 and **P < 0.01 (patient vs. control).

Figure 6.

Comparison across patient groups (Experiment 4). (a) Mean errors to criterion on the maximum ambiguity conditions of novel (blobs and barcodes) and familiar (bugs and beasts) concurrent discriminations for controls, SD patients, and patients with focal MTL lesions. (b) Proportion of errors committed on the high-ambiguity conditions of novel (greebles) and familiar (objects) oddity discriminations for controls, SD patients, and patients with focal MTL lesions. *P < 0.05 and **P < 0.01 (patient vs. control).

Object Oddity (Experiment 3)

A repeated-measures ANOVA with a within-subject factor of stimulus familiarity (i.e., high-ambiguity greebles vs. high-ambiguity familiar objects) and a between-subject factor of subject group was performed (Fig. 6b; for consistency with Experiment 1, results are illustrated in terms of errors). This confirmed a significant effect of subject group (F(2,23) = 31.92; P < 0.001) and familiarity (F(1,23) = 16.55; P < 0.001), and a significant interaction between the 2 (F(2,23) = 7.33; P < 0.001). Two 1-way ANOVAs on the familiar and novel discriminations revealed that the groups performed differently to each other on both novel and familiar discriminations (both F > 15.19; P < 0.001). Post hoc t tests (Bonferroni-corrected) revealed that both patient groups were impaired relative to controls on all high-ambiguity discriminations regardless of their familiarity (all P < 0.01). When the patient groups were compared relative to each other, the MTL patients performed significantly worse than the SD group on novel discriminations (P < 0.01), but performance in the 2 patient groups was not different on familiar discriminations (P > 0.99). When compared individually, each of the 3 MTL cases fell more than 2 standard deviations outside the control and SD population means for novel discriminations. For familiar discriminations, all 3 MTL cases fell more than 2 standard deviations outside the control mean, whereas all 3 cases were within 2 standard deviations of the SD group mean.

Visual inspection of the graphs in Figure 6 indicates that, in general, the pattern of performance of the MTL group was similar across the 2 experiments: MTL patients were severely impaired on both novel and familiar discriminations, but their deficit was attenuated by the use of familiar stimuli. As illustrated by the upward shift (i.e., more errors) of the line for SD performance in Figure 6b compared with that in Figure 6a, the SD group performed worse on the oddity discriminations (Experiment 3) than the concurrent discriminations (Experiment 1) for both familiar and novel stimuli. Furthermore, as demonstrated by the relatively flat line for SD performance in both Figure 6a and b, SD patients showed equivalent performance on novel and familiar stimuli within each experiment. These patterns of patient performance were true regardless of control performance: Controls showed facilitation for meaningful stimuli on the concurrent object-discrimination tasks in Experiment 1 (t(19) = 3.7; P < 0.01), but similar performance for novel and meaningful stimuli in Experiment 3 (t(15) = −1.0; P = 0.32).

To further investigate this, we compared the facilitation by the use of familiar stimuli across the different subject groups in each task using 1-way ANOVAs (i.e., a “facilitation score” for each participant was obtained by subtracting the number of errors committed on novel discriminations from the number of errors on familiar discriminations). These revealed that in both tasks, the groups showed different amounts of facilitation (both F > 7.33, P < 0.001). Post hoc t tests (Bonferroni corrected) revealed that the MTL patients showed significantly more facilitation from the use of meaningful stimuli on both experiments than either controls or SD patients (all P < 0.01). By contrast, SD patients and controls did not demonstrate significantly different amounts of facilitation from each other on either experiment (both P > 0.57).

In summary, across 2 different object discrimination tasks, stimulus familiarity differentially influenced performance in patients with SD compared with those with focal lesions to the MTL. On both tasks, the MTL group was severely impaired on both novel and familiar discriminations, but their deficit was attenuated by the use of semantically meaningful stimuli (see also Moses et al. 2008). By contrast, the SD patients showed no benefit from the use of familiar stimuli. These results suggest that performance in the MTL group was facilitated by input from a semantic system that was severely damaged in SD. Object processing, therefore, appears to involve a dynamic interaction between semantic and perceptual processes, even for tasks with no overt semantic component.

General Discussion

A Role for Perirhinal Cortex in Complex Object Perception

Across 3 experiments, patients with SD showed a deficit on visual discriminations with a high degree of feature ambiguity for 1) 2-feature items corresponding to familiar but not novel items (Experiment 1) and 2) both familiar and novel multifeature items with a very high degree of feature ambiguity (Experiments 2–3). These findings are consistent with those reported in other populations (rat, monkey, and human) with perirhinal damage (e.g., Buckley et al. 2001; Bussey et al. 2002; Lee, Buckley et al. 2005; Bartko et al. 2007) and in functional magnetic resonance imaging studies (Devlin and Price 2007; Lee et al. 2008; Barense et al. 2009). When taken together, they strongly suggest a role for the perirhinal cortex in processing complex conjunctions of object features, on both mnemonic and perceptual tasks.

A limitation of this study is that the impairments reported in the SD cases may reflect dysfunction in regions beyond the MTL, specifically to the lateral temporal lobe (e.g., areas TE/TEO). Although lesion analysis revealed minimal structural damage to these areas (see Fig. 1), SD patients have demonstrated hypometabolism in more caudal lateral temporal regions distant from their structural damage in a positron emission tomography study (Mummery et al. 1999). We cannot be certain that more lateral and/or posterior abnormality did not contribute to the current results; however, this possibility seems unlikely. First, if these impairments were indeed due to lateral posterior temporal lobe dysfunction, one would expect the SD patients to demonstrate more widespread perceptual problems, such as color discrimination, which is thought to be dependent upon Area TE (Buckley et al. 1997). In contrast, the SD patients were unimpaired on color and size discriminations, all minimum ambiguity problems, and all conditions of the most difficult class of discriminations (barcodes). Second, the findings agree with a number of other neuropsychological studies across different subject groups (Barense et al. 2005, 2007; Lee, Buckley et al. 2005; Lee, Bussey et al. 2005; Lee et al. 2006), suggesting they are not unusual. Third, the patterns reported here replicate animal lesion studies, in which it is possible to undertake more selective lesions. These studies have precisely localized the effects of feature ambiguity to perirhinal cortex and thereby help to discredit the involvement of other temporal regions to the cognitive process in question (Buckley et al. 2001; Bussey et al. 2002; Saksida et al. 2006; Bartko et al. 2007). Thus, although one must be cautious in interpreting findings from a neurodegenerative condition that does not selectively affect perirhinal cortex, it seems unlikely that pathology beyond the MTL is responsible for the observed discrimination deficits.

Finally, an important contribution of the current work is the suggestion that semantic knowledge about objects contributes to performance in visual discrimination tasks. Across 2 different discrimination tasks, stimulus meaningfulness differentially influenced performance in patients with SD compared with those with focal lesions to the MTL. Although amnesic patients with focal MTL lesions were impaired on discriminations of both novel and meaningful stimuli, in both tasks, their deficit was attenuated by the use of meaningful stimuli. This pattern was true regardless of control performance: Controls showed facilitation for meaningful stimuli on the concurrent object-discrimination tasks in Experiment 1, but in Experiment 3, performance for novel and meaningful stimuli was matched. In contrast to the MTL amnesics, the SD patients showed no facilitation from the use of familiar stimuli in either experiment. Thus, although they were impaired on both familiar and novel discriminations, the MTL cases were able to engage semantic support that was presumably not available to the SD patients (see also Moses et al. 2008). These behavioral differences between patient groups may reflect differences in the nature of their pathology (i.e., neurodegenerative disease vs. viral infection or acute head injury, Lambon Ralph et al. 2007). In summary, these data support an account that holds that representations of complex objects (dependent upon perirhinal cortex) interact with higher-order conceptual representations (damaged in SD) (Hovius et al. 2003; Rogers et al. 2003; 2004; Patterson et al. 2006).

Funding

Medical Research Council, a Peterhouse Research Fellowship (to M.D.B.), a Discovery Grant from the Natural Sciences and Engineering Research Council of Canada (to M.D.B.), and the Wales Institute of Cognitive Neuroscience (to K.S.G.).

We thank John Hodges for access to patients, Karalyn Patterson for helpful comments on the manuscript, and Michael J. Tarr (Carnegie Mellon University) for providing the greeble and fribble stimuli. Conflict of Interest: None declared.

References

Allport
DA
Newman
SK
Epstein
R
Distributed memory, modular systems and dysphasia
Current perspectives in dysphasia
 , 
1985
Edinburgh (UK)
Churchill Livingstone
Barense
MD
Bussey
TJ
Lee
ACH
Rogers
TT
Davies
RR
Saksida
LM
Murray
EA
Graham
KS
Functional specialization in the human medial temporal lobe
J Neurosci
 , 
2005
, vol. 
25
 
44
(pg. 
10239
-
10246
)
Barense
MD
Gaffan
D
Graham
KS
The human medial temporal lobe processes online representations of complex objects
Neuropsychologia
 , 
2007
, vol. 
45
 
13
(pg. 
2963
-
2974
)
Barense
MD
Henson
RN
Lee
AC
Graham
KS
Medial temporal lobe activity during complex discrimination of faces, objects, and scenes: effects of viewpoint
Hippocampus
 , 
2009
 
Published online June 4, doi: 10.1002/hipo.20641
Barsalou
LW
Simmons
WK
Barbey
AK
Wilson
CD
Grounding conceptual knowledge in modality-specific systems
Trends Cogn Sci
 , 
2003
, vol. 
7
 
2
(pg. 
84
-
91
)
Bartko
SJ
Winters
BD
Cowell
RA
Saksida
LM
Bussey
TJ
Perirhinal cortex resolves feature ambiguity in configural object recognition and perceptual oddity tasks
Learn Mem
 , 
2007
, vol. 
14
 
12
(pg. 
821
-
832
)
Bozeat
S
Lambon Ralph
MA
Patterson
K
Garrard
P
Hodges
JR
Non-verbal semantic impairment in semantic dementia
Neuropsychologia
 , 
2000
, vol. 
38
 
9
(pg. 
1207
-
1215
)
Buckley
MJ
Booth
MC
Rolls
ET
Gaffan
D
Selective perceptual impairments after perirhinal cortex ablation
J Neurosci
 , 
2001
, vol. 
21
 
24
(pg. 
9824
-
9836
)
Buckley
MJ
Gaffan
D
Murray
EA
Functional double dissociation between two inferior temporal cortical areas: perirhinal cortex versus middle temporal gyrus
J Neurophysiol
 , 
1997
, vol. 
77
 
2
(pg. 
587
-
598
)
Bussey
TJ
Saksida
LM
Murray
EA
Perirhinal cortex resolves feature ambiguity in complex visual discriminations
Eur J Neurosci
 , 
2002
, vol. 
15
 
2
(pg. 
365
-
374
)
Chan
D
Fox
NC
Scahill
RI
Crum
WR
Whitwell
JL
Leschziner
G
Rossor
AM
Stevens
JM
Cipolotti
L
Rossor
MN
Patterns of temporal lobe atrophy in semantic dementia and Alzheimer's disease
Ann Neurol
 , 
2001
, vol. 
49
 
4
(pg. 
433
-
442
)
Clague
F
Dudas
RB
Thompson
SA
Graham
KS
Hodges
JR
Multidimensional measures of person knowledge and spatial associative learning: can these be applied to the differentiation of Alzheimer's disease from frontotemporal and vascular dementia?
Neuropsychologia
 , 
2005
, vol. 
43
 
9
(pg. 
1338
-
1350
)
Coltheart
M
Are there lexicons?
Q J Exp Psychol A
 , 
2004
, vol. 
57
 
7
(pg. 
1153
-
1171
)
Davies
RR
Graham
KS
Xuereb
JH
Williams
GB
Hodges
JR
The human perirhinal cortex and semantic memory
Eur J Neurosci
 , 
2004
, vol. 
20
 
9
(pg. 
2441
-
2446
)
Davies
RR
Xuereb
JH
Hodges
JR
The human perirhinal cortex in semantic memory: an in vivo and postmortem volumetric magnetic resonance imaging study in semantic dementia, Alzheimer's disease and matched controls
Neuropathol Appl Neurobiol
 , 
2002
, vol. 
28
 (pg. 
167
-
168
)
Desimone
R
Ungerleider
LG
Boller
F
Grafman
J
Neural mechanisms of visual processing in monkeys
Handbook of neuropsychology
 , 
1989
New York
Elsevier Science
(pg. 
267
-
299
)
Devlin
JT
Price
CJ
Perirhinal contributions to human visual perception
Curr Biol
 , 
2007
, vol. 
17
 
17
(pg. 
1484
-
1488
)
Folstein
MF
Folstein
SE
McHugh
PR
“Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician
J Psychiatr Res
 , 
1975
, vol. 
12
 
3
(pg. 
189
-
198
)
Galton
CJ
Gomez-Anson
B
Antoun
N
Scheltens
P
Patterson
K
Graves
M
Sahakian
BJ
Hodges
JR
Temporal lobe rating scale: application to Alzheimer's disease and frontotemporal dementia
J Neurol Neurosurg Psychiatry
 , 
2001
, vol. 
70
 
2
(pg. 
165
-
173
)
Galton
CJ
Patterson
K
Graham
K
Lambon-Ralph
MA
Williams
G
Antoun
N
Sahakian
BJ
Hodges
JR
Differing patterns of temporal atrophy in Alzheimer's disease and semantic dementia
Neurology
 , 
2001
, vol. 
57
 
2
(pg. 
216
-
225
)
Garrard
P
Perry
R
Hodges
JR
Disorders of semantic memory
J Neurol Neurosurg Psychiatry
 , 
1997
, vol. 
62
 
5
(pg. 
431
-
435
)
Gauthier
I
Tarr
MJ
Becoming a “Greeble” expert: exploring mechanisms for face recognition
Vision Res
 , 
1997
, vol. 
37
 
12
(pg. 
1673
-
1682
)
Goldstone
RL
Barsalou
LW
Reuniting perception and conception
Cognition
 , 
1998
, vol. 
65
 
2–3
(pg. 
231
-
262
)
Graham
KS
Scahill
VL
Hornberger
M
Barense
MD
Lee
AC
Bussey
TJ
Saksida
LM
Abnormal categorization and perceptual learning in patients with hippocampal damage
J Neurosci
 , 
2006
, vol. 
26
 
29
(pg. 
7547
-
7554
)
Graham
KS
Simons
JS
Pratt
KH
Patterson
K
Hodges
JR
Insights from semantic dementia on the relationship between episodic and semantic memory
Neuropsychologia
 , 
2000
, vol. 
38
 
3
(pg. 
313
-
324
)
Graham
NL
Emery
T
Hodges
JR
Distinctive cognitive profiles in Alzheimer's disease and subcortical vascular dementia
J Neurol Neurosurg Psychiatry
 , 
2004
, vol. 
75
 
1
(pg. 
61
-
71
)
Hodges
JR
Bozeat
S
Lambon Ralph
MA
Patterson
K
Spatt
J
The role of conceptual knowledge in object use evidence from semantic dementia
Brain
 , 
2000
, vol. 
123
 
Pt 9
(pg. 
1913
-
1925
)
Hodges
JR
Graham
N
Patterson
K
Charting the progression in semantic dementia: implications for the organisation of semantic memory
Memory
 , 
1995
, vol. 
3
 
3–4
(pg. 
463
-
495
)
Hodges
JR
Patterson
K
Is semantic memory consistently impaired early in the course of Alzheimer's disease? Neuroanatomical and diagnostic implications
Neuropsychologia
 , 
1995
, vol. 
33
 
4
(pg. 
441
-
459
)
Hodges
JR
Patterson
K
Oxbury
S
Funnell
E
Progressive fluent aphasia with temporal lobe atrophy
Brain
 , 
1992
, vol. 
115
 
Pt 6
(pg. 
1783
-
1806
)
Hodges
JR
Patterson
K
Tyler
LK
Loss of semantic memory: implications for the modularity of mind
Cogn Neuropsychol
 , 
1994
, vol. 
11
 
5
(pg. 
505
-
542
)
Hovius
M
Kellenbach
ML
Graham
KS
Hodges
JR
Patterson
K
What does the object decision task measure? Reflections on the basis of evidence from semantic dementia
Neuropsychology
 , 
2003
, vol. 
17
 
1
(pg. 
100
-
107
)
Howard
D
Patterson
K
Pyramids and palm trees: a test of semantic access from pictures and words
 , 
1992
Bury St Edmunds (UK)
Thames Valley Test
Humphreys
GW
Forde
EM
Hierarchies, similarity, and interactivity in object recognition: “category-specific” neuropsychological deficits
Behav Brain Sci
 , 
2001
, vol. 
24
 
3
(pg. 
453
-
476
discussion 476–509
Humphreys
GW
Riddoch
MJ
Quinlan
PT
Cascade processes in picture identification
Cogn Neuropsychol
 , 
1988
, vol. 
5
 (pg. 
67
-
103
)
Lambon Ralph
MA
Lowe
C
Rogers
TT
Neural basis of category-specific semantic deficits for living things: evidence from semantic dementia, HSVE and a neural network model
Brain
 , 
2007
, vol. 
130
 
Pt 4
(pg. 
1127
-
1137
)
Lambon Ralph
MA
Powell
J
Howard
D
Whitworth
AB
Garrard
P
Hodges
JR
Semantic memory is impaired in both dementia with Lewy bodies and dementia of Alzheimer's type: a comparative neuropsychological study and literature review
J Neurol Neurosurg Psychiatry
 , 
2001
, vol. 
70
 
2
(pg. 
149
-
156
)
Lee
AC
Buckley
MJ
Gaffan
D
Emery
T
Hodges
JR
Graham
KS
Differentiating the roles of the hippocampus and perirhinal cortex in processes beyond long-term declarative memory: a double dissociation in dementia
J Neurosci
 , 
2006
, vol. 
26
 
19
(pg. 
5198
-
5203
)
Lee
AC
Buckley
MJ
Pegman
SJ
Spiers
H
Scahill
VL
Gaffan
D
Bussey
TJ
Davies
RR
Kapur
N
Hodges
JR
, et al.  . 
Specialization in the medial temporal lobe for processing of objects and scenes
Hippocampus
 , 
2005
, vol. 
15
 
6
(pg. 
782
-
797
)
Lee
AC
Bussey
TJ
Murray
EA
Saksida
LM
Epstein
RA
Kapur
N
Hodges
JR
Graham
KS
Perceptual deficits in amnesia: challenging the medial temporal lobe ‘mnemonic’ view
Neuropsychologia
 , 
2005
, vol. 
43
 
1
(pg. 
1
-
11
)
Lee
AC
Levi
N
Davies
RR
Hodges
JR
Graham
KS
Differing profiles of face and scene discrimination deficits in semantic dementia and Alzheimer's disease
Neuropsychologia
 , 
2007
, vol. 
45
 
9
(pg. 
2135
-
2146
)
Lee
AC
Scahill
VL
Graham
KS
Activating the medial temporal lobe during oddity judgment for faces and scenes
Cereb Cortex
 , 
2008
, vol. 
18
 
3
(pg. 
683
-
696
)
McClelland
JL
Rogers
TT
The parallel distributed processing approach to semantic cognition
Nat Rev Neurosci
 , 
2003
, vol. 
4
 
4
(pg. 
310
-
322
)
Miozzo
M
Gordon
P
Facts, events, and inflection: when language and memory dissociate
J Cogn Neurosci
 , 
2005
, vol. 
17
 
7
(pg. 
1074
-
1086
)
Moses
JA
Jr
Factor structure of Benton's tests of visual retention, visual construction, and visual form discrimination
Arch Clin Neuropsychol
 , 
1986
, vol. 
1
 
2
(pg. 
147
-
156
)
Moses
SN
Ostreicher
ML
Rosenbaum
RS
Ryan
JD
Successful transverse patterning in amnesia using semantic knowledge
Hippocampus
 , 
2008
, vol. 
18
 
2
(pg. 
121
-
124
)
Mummery
CJ
Patterson
K
Wise
RJ
Vandenbergh
R
Price
CJ
Hodges
JR
Disrupted temporal lobe connections in semantic dementia
Brain
 , 
1999
, vol. 
122
 
Pt 1
(pg. 
61
-
73
)
Murray
EA
Bussey
TJ
Perceptual-mnemonic functions of the perirhinal cortex
Trends Cogn Sci
 , 
1999
, vol. 
3
 
4
(pg. 
142
-
151
)
Neary
D
Snowden
JS
Gustafson
L
Passant
U
Stuss
D
Black
S
Freedman
M
Kertesz
A
Robert
PH
Albert
M
, et al.  . 
Frontotemporal lobar degeneration: a consensus on clinical diagnostic criteria
Neurology
 , 
1998
, vol. 
51
 
6
(pg. 
1546
-
1554
)
Noppeney
U
Patterson
K
Tyler
LK
Moss
H
Stamatakis
EA
Bright
P
Mummery
C
Price
CJ
Temporal lobe lesions and semantic impairment: a comparison of herpes simplex virus encephalitis and semantic dementia
Brain
 , 
2007
, vol. 
130
 
Pt 4
(pg. 
1138
-
1147
)
Patterson
K
Lambon Ralph
MA
Jefferies
E
Woollams
A
Jones
R
Hodges
JR
Rogers
TT
“Presemantic” cognition in semantic dementia: six deficits in search of an explanation
J Cogn Neurosci
 , 
2006
, vol. 
18
 
2
(pg. 
169
-
183
)
Patterson
K
Nestor
PJ
Rogers
TT
Where do you know what you know? The representation of semantic knowledge in the human brain
Nat Rev Neurosci
 , 
2007
, vol. 
8
 
12
(pg. 
976
-
987
)
Pobric
G
Jefferies
E
Lambon Ralph
MA
Anterior temporal lobes mediate semantic representation: mimicking semantic dementia by using rTMS in normal participants
Proc Natl Acad Sci U S A
 , 
2007
, vol. 
104
 
50
(pg. 
20137
-
20141
)
Riddoch
MJ
Humphreys
GW
A case of integrative visual agnosia
Brain
 , 
1987
, vol. 
110
 
Pt 6
(pg. 
1431
-
1462
)
Riddoch
MJ
Humphreys
GW
Visual object processing in optic aphasia: a case of semantic access agnosia
Cogn Neuropsychol
 , 
1987
, vol. 
4
 (pg. 
131
-
185
)
Riesenhuber
M
Poggio
T
Hierarchical models of object recognition in cortex
Nat Neurosci
 , 
1999
, vol. 
2
 
11
(pg. 
1019
-
1025
)
Rogers
TT
Hodges
JR
Patterson
K
Lambon Ralph
MA
Object recognition under semantic impairment: the effects of conceptual regularities on perceptual decisions
Lang Cogn Proc
 , 
2003
, vol. 
18
 
5–6
(pg. 
625
-
662
)
Rogers
TT
Lambon Ralph
MA
Garrard
P
Bozeat
S
McClelland
JL
Hodges
JR
Patterson
K
Structure and deterioration of semantic memory: a neuropsychological and computational investigation
Psychol Rev
 , 
2004
, vol. 
111
 
1
(pg. 
205
-
235
)
Rogers
TT
Patterson
K
Graham
K
Colour knowledge in semantic dementia: it is not all black and white
Neuropsychologia
 , 
2007
, vol. 
45
 
14
(pg. 
3285
-
3298
)
Rorden
C
Brett
M
Stereotaxic display of brain lesions
Behav Neurol
 , 
2000
, vol. 
12
 
4
(pg. 
191
-
200
)
Saksida
LM
Bussey
TJ
Buckmaster
CA
Murray
EA
No effect of hippocampal lesions on perirhinal cortex-dependent feature-ambiguous visual discriminations
Hippocampus
 , 
2006
, vol. 
16
 
4
(pg. 
421
-
430
)
Sheridan
J
Humphreys
GW
A verbal-semantic category-specific recognition impairment
Cogn Neuropsychol
 , 
1993
, vol. 
10
 
2
(pg. 
143
-
184
)
Simons
JS
Graham
KS
Galton
CJ
Patterson
K
Hodges
JR
Semantic knowledge and episodic memory for faces in semantic dementia
Neuropsychology
 , 
2001
, vol. 
15
 
1
(pg. 
101
-
114
)
Simons
JS
Graham
KS
Hodges
JR
Perceptual and semantic contributions to episodic memory: evidence from semantic dementia and Alzheimer's disease
J Mem Lang
 , 
2002
, vol. 
47
 (pg. 
197
-
213
)
Snowden
J
Goulding
P
Neary
D
Semantic dementia: a form of circumscribed cerebral atrophy
Behav Neurol
 , 
1989
, vol. 
2
 (pg. 
167
-
182
)
Squire
LR
Stark
CE
Clark
RE
The medial temporal lobe
Annu Rev Neurosci
 , 
2004
, vol. 
27
 (pg. 
279
-
306
)
Suzuki
WA
Amaral
DG
Perirhinal and parahippocampal cortices of the macaque monkey: cortical afferents
J Comp Neurol
 , 
1994
, vol. 
350
 
4
(pg. 
497
-
533
)
Taylor
KI
Moss
HE
Stamatakis
EA
Tyler
LK
Binding crossmodal object features in perirhinal cortex
Proc Natl Acad Sci U S A
 , 
2006
, vol. 
103
 
21
(pg. 
8239
-
8244
)
Tulving
E
Schacter
DL
Priming and human memory systems
Science
 , 
1990
, vol. 
247
 
4940
(pg. 
301
-
306
)
Tyler
LK
Stamatakis
EA
Jones
RW
Bright
P
Acres
K
Marslen-Wilson
WD
Deficits for semantics and the irregular past tense: a causal relationship?
J Cogn Neurosci
 , 
2004
, vol. 
16
 
7
(pg. 
1159
-
1172
)
Warrington
EK
The selective impairment of semantic memory
Q J Exp Psychol
 , 
1975
, vol. 
27
 
4
(pg. 
635
-
657
)
Warrington
EK
James
M
Visual object and space perception battery (VOSP)
 , 
1991
Oxford
Harcourt Assessment
Warrington
EK
McCarthy
RA
Categories of knowledge. Further fractionations and an attempted integration
Brain
 , 
1987
, vol. 
110
 
Pt 5
(pg. 
1273
-
1296
)
Williams
P
Simons
DJ
Detecting changes in novel, complex three-dimensional objects
Vis Cogn
 , 
2000
, vol. 
7
 
1/2/3
(pg. 
297
-
322
)