Semantic memory: Which side are you on?

Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Q1 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

Contents lists available at ScienceDirect

Neuropsychologia journal homepage: www.elsevier.com/locate/neuropsychologia

Semantic memory: Which side are you on? Karalyn Patterson a,b, Michael D. Kopelman c, Anna M. Woollams d, Sonia L.E. Brownsett e, Fatemeh Geranmayeh e, Richard J.S. Wise e,n a

Department of Clinical Neurosciences, University of Cambridge, UK MRC Cognition & Brain Sciences, Cambridge, UK c Institute of Psychiatry, Kings College London, UK d School of Psychological Sciences, University of Manchester, UK e Computational, Cognitive, and Clinical Neuroimaging Laboratory, Imperial College London, Hammersmith Hospital, London W12 0NN, UK b

art ic l e i nf o

a b s t r a c t

Article history: Received 3 June 2014 Received in revised form 10 November 2014 Accepted 20 November 2014

We present two patients in whom the combination of lesion site and cognitive performance was un- Q7 iquely informative about the organisation and functional anatomy of semantic memory. One had had a single lobar stroke with an unusual distribution, largely destroying the whole of the left temporal lobe ventral to the superior temporal sulcus. The other patient had had herpes simplex encephalitis with destruction that was confined to the left cerebral hemisphere. The lesion again mainly encompassed the left temporal lobe, but also extended to the left inferior frontal gyrus. Cognitive outcomes in the two patients were compared with each other and with published results from patients with semantic dementia. This is because, whereas the majority of semantic dementia patients present with more prominent atrophy of the left rostroventral temporal lobe, they invariably have a degree of atrophy in the mirror region on the right that progresses. Semantic dementia therefore provides no clear evidence about the specific role of the left rostroventral temporal lobe. The two patients showed a highly consistent cognitive profile. Their deficits were also similar in many respects to that observed in patients with mild-moderate semantic dementia, including severe anomia that was not resolved by phonological cues and impairment on non-verbal as well as verbal semantic tasks. Certain key features of the semantic dementia profile, however—including sensitivity to the familiarity and typicality of the stimulus materials—appeared only in tasks requiring verbal output in these two patients with unilateral left temporal lesions. Results in these cases provide some of the first definitive evidence regarding the specific functions of the left anterior temporal lobe. & 2014 Published by Elsevier Ltd.

Keywords: Left temporal lobe Stroke Herpes simplex virus encephalitis Semantic dementia Semantic aphasia

1. Introduction Semantic memory became the subject of serious experimental study about 40 years ago, following the proposal that episodic and semantic knowledge are distinct forms of long-term memory (Tulving, 1972). Clinically, it was common for patients to present with a predominant impairment in episodic memory, i.e. difficulty in encoding specific life events. In contrast, patients with a predominant deficit of semantic memory were not recognised until the study by Warrington (1975) of three cases with what is now Abbreviations: ACE-R, Addenbrooke's Cognitive Examination; HSVE, herpes simplex virus encephalitis; rvTL, rostroventral temporal lobes; SA, semantic aphasia; SD, semantic dementia; SD-L, semantic dementia with most atrophy in the left temporal lobe; SD-R, semantic dementia with most atrophy in the right temporal lobe; TLE, temporal lobe epilepsy; TROG, test for reception of grammar n Corresponding author. E-mail address: [email protected] (R.J.S. Wise).

known as the semantic variant of fronto-temporal dementia, or semantic dementia (SD). Such patients demonstrate a relatively selective progressive loss of conceptual knowledge. Common manifestations of SD include failure to understand the meanings of words, to name or draw objects, to understand the properties of objects and other concepts, and to recall widely known facts. Once the clinical descriptions of SD began to appear in the literature (Hodges et al., 1992; Snowden et al., 1989), cases were increasingly identified by clinicians and referred for specialist study. Imaging data and, to a lesser extent, post mortem results in these patients have been extensively analysed, in part because focal regions of damage can be presumed to constitute an important component of the network for conceptual knowledge in the healthy brain. Neuroradiological results in SD can of course vary somewhat from one case to another; but relative to most forms of brain disease, there is striking consistency in the location of atrophy in SD, focused on the rostroventral temporal lobes (rvTL) (Acosta-Cabronero et al.,

http://dx.doi.org/10.1016/j.neuropsychologia.2014.11.024 0028-3932/& 2014 Published by Elsevier Ltd.

Please cite this article as: Patterson, K., et al., Semantic memory: Which side are you on? Neuropsychologia (2014), http://dx.doi.org/ 10.1016/j.neuropsychologia.2014.11.024i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89

2

1 2 Q3 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

K. Patterson et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

2011; Agosta et al., 2010; Binney et al., 2010; Desgranges et al., 2007; Guo et al., 2013; Libon et al., 2013; Mion et al., 2010; Nestor et al., 2006; Snowden et al., 1996; Yamamoto et al., 2009). The atrophy in this region is dramatic, with 50
80% grey-matter loss even in mild-moderate stages of the condition (Hodges and Patterson, 2007; Mesulam et al., 2013). Having been virtually absent from discussions of the neuroanatomical basis of semantic memory a few decades ago, the rvTL is now a prominent player. This is primarily due to research on SD, complemented by recent fMRI studies with improved ability to visualise signal from this region in normal participants performing semantic tasks (e.g., Binney et al., 2010; Peelen and Caramazza, 2012) and by the use of transcranial magnetic stimulation in healthy participants (Pobric et al., 2010). Like most neurodegenerative conditions, SD is a bilateral disease, though typically asymmetrical at least until the later stages. In the majority of cases, the left temporal lobe shows the most atrophy (SD-L), while in a smaller number the atrophy is either greater on the right (SD-R) or approximately symmetrical (Snowden et al., 2004). Some differences in cognitive profile between the SD-L and SD-R subgroups have been demonstrated. For example, when asked to supply information about famous people, SD-L patients performed more poorly when probed with the people's names whereas SD-R cases had more trouble when the stimuli were faces (Snowden et al., 2004); and performance on tasks requiring verbal output, such as naming objects or generating the names of members of a category, is more impaired in SD-L than SD-R cases matched on other measures of conceptual knowledge (Lambon Ralph et al., 2001; Thompson et al., 2003). In these and other studies, however, both groups were substantially impaired relative to controls on all tests requiring semantic memory, whatever the modality of input or output. Therefore, although some researchers have placed considerable weight on the cognitive differences (e.g. Gainotti, 2012), others consider that SD is associated with a rather consistent pattern of semantic deficit, in which SD-L and SD-R cases are more similar than different. As atrophy in SD is bilateral, if often asymmetrical, such cases do not provide conclusive evidence on the contribution, jointly or separately, of the left and right temporal lobes to the various aspects of semantic memory. Two studies by Lambon Ralph et al. (2010, 2012) each identified 20 patients with unilateral temporallobe lesions: the former with a variety of aetiologies but mostly surgical resections for tumours or temporal lobe epilepsy (TLE), the latter all with resections for TLE. When assessed with the tests commonly administered in research on SD, the patients in these studies did exhibit some semantic deficits relative to matched controls, especially on more demanding tests or (as in Lambon Ralph et al. (2012)) on reaction-time measures. In neither study, however, did the performance of the patients with either unilateral left or right lesions resemble SD, suggesting that the deterioration of semantic memory characteristic of SD is the result of bilateral temporal lobe pathology. Nevertheless, it can be argued that the growth of tumours and the chronicity of repeated focal epileptic discharges allow time for reorganisation of function. Furthermore, the areas resected in TLE are more discrete than the regions of atrophy and hypometabolism in SD. The most robust evidence regarding the contribution of the rvTL in one hemisphere would come from patients with an acute or sub-acute onset of focal damage in this region; but such patients are very rare. Owing to the anatomy of the arterial supply, stroke lesions are almost never confined to this region; and herpes simplex virus encephalitis (HSVE) affecting the ventromedial temporal lobes is rarely a unilateral disease (for example, see Fig. 9 in Kopelman et al. (1999); and Fig. 2 in Reed et al. (2005)). We have identified two patients, one with a stroke and the other with HSVE, with atypical lesion distributions. Both lesions are entirely

unilateral, encompassing all of the left rvTL. With these two cases, we addressed the question of whether a left-lateralised lesion in the territory affected bilaterally in SD would produce a cognitive profile similar to, or distinctly different from, SD. The level of performance on any semantic task in patients with SD can be largely summarised or predicted in terms of five main factors (Jefferies and Lambon Ralph, 2006; Patterson et al., 2007). The first, which does not apply to patients without a progressive condition, is the stage of disease. The remaining four formed the focus of this investigation, and are as follows: 1.1. The familiarity of the stimulus materials used to probe conceptual knowledge The performance of patients with SD is always significantly worse with less familiar stimuli, regardless of either the materials used (spoken words, written words, pictures, real objects, environmental sounds etc.) or the task modality (expressive or receptive). 1.2. The typicality of the stimulus materials Objects or concepts vary in how much they resemble others in the same semantic category, and category members lacking one or more salient features shared by most other instances (for example birds, like the ostrich or penguin, that do not fly) are considered atypical. SD patients have consistently and significantly more failures in semantic tasks such as naming (Woollams et al., 2008) on atypical category instances. Note that, whilst to some extent correlated, the dimensions of familiarity and typicality are separable: robins and kingfishers are both fairly typical birds, but the robin is much more frequently encountered and thus more familiar (Rogers and Patterson, 2007). The powerful typicality effect in SD extends well beyond the domain of explicit semantic processing. For example, SD patients are likely to pronounce the written word sew as “sue”, a pronunciation typical for sew's orthographic neighbours new, few, stew etc. These regularisation errors are more likely for less frequent words, yielding a frequency-by-typicality interaction that occurs in every domain thus far assessed in SD (e.g. Lambon Ralph et al., 2012; Patterson et al., 2006). 1.3. The specificity of the semantic knowledge required to perform the task In a yes/no matching task, SD patients saw and heard a category name paired with a picture and were asked to judge whether the picture was an example of the category (Rogers and Patterson, 2007). The category names represented three different levels of specificity. A picture of a robin, for example, can be categorised as an animal (general-level), bird (basic-level) or robin (specific-level). In both accuracy and response times, healthy controls show a robust basic-level advantage in this task, with performance somewhat lower/slower, and roughly equal, in the general and specific conditions. SD patients beyond the very mildest stage reveal a very different pattern: accuracy near controls at the general level, but impaired performance at the basic level and severely reduced accuracy at the specific level. 1.4. Minimal impact of cueing The profound anomia in SD is minimally assisted by phonological cueing. To summarise, this study asked whether the left temporal lesions of these two non-SD patients would be associated with a semantic impairment and, if so, whether that impairment would have any or all of these four characteristics.

Please cite this article as: Patterson, K., et al., Semantic memory: Which side are you on? Neuropsychologia (2014), http://dx.doi.org/ 10.1016/j.neuropsychologia.2014.11.024i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 Q8 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132

K. Patterson et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

3

Fig. 1. Lesion location in the two patients with unilateral left temporal lobe damage. Coronal slices from T1-weighted images are displayed. They are oriented from anterior to posterior, viewed from left to right. In addition, two axial T2-weighted images are shown at the level of the temporal lobes. On an MRI scan carried out in 2012, DJ showed extensive left temporal lobe damage, as the result of herpes simplex encephalitis, which included the superior, middle and inferior temporal gyri, the fusiform gyrus, the parahippocampal gyrus, and the hippocampus and rhinal cortices. The lesion extended posteriorly to the ventral occipito-temporal junction, medially to insular cortex and superiorly into the inferior frontal gyrus, including pars opercularis and pars triangularis. GF, following a middle cerebral artery infarction, had a lesion confined to the temporal lobe that spared the superior temporal gyrus and extended as far caudally as the ventral occipito-temporal junction. Although the medial temporal damage was less extensive than that in DJ, the left hippocampus was clearly atrophic.

1.5. Case reports DJ, a right-handed professional artist, aged 36 years, was admitted to hospital with an acute illness in 1990. Tests confirmed herpes simplex virus encephalitis. A study of DJ's memory and other cognitive abilities (Stanhope and Kopelman, 2000) reported an initially severe but resolving amnesic syndrome, plus considerable and persisting deficits in naming and reading. Verbal memory remained more severely affected than visual memory. His artistic skills were preserved. The assessments for the current study were carried out in 2011
2012. GF, a right-handed retired district company manager, aged 69 years, had a left middle cerebral artery territory infarct in 2000. His wife reported that he was initially mute, then began to say yes and no and then short phrases. At the time of testing, in 2008
2009, his speech was fairly fluent but anomic. MR imaging for both patients, in 2012 and 2011, respectively, is presented in Fig. 1. Control data were available from published results on agematched healthy participants. These are easy tests for healthy adults who achieve accuracy at or near ceiling. The results on DJ and GF are compared to published data obtained from patients with SD. We also refer to results from semantically impaired stroke patients, labelled semantic aphasics (SA) in the research of Jefferies et al. (2008), whose semantic deficits are very different to those in SD (Jefferies and Lambon Ralph, 2006; Jefferies et al., 2010).

2. Methods The two patients were administered a range of cognitive tests. We describe first the assessments of general cognitive status, and then those concerning semantic memory.

The general measures comprised the Mini-Mental State Examination (MMSE) (Folstein et al., 1975), the revised version of the Addenbrooke's Cognitive Examination (ACE-R) (Mioshi et al., 2006), forward and backward digit span (Wechsler, 1987) and the Test for the Reception of Grammar (TROG) (Bishop, 1989). Tests of general semantic ability included the following: 1. The Repeat and Point test (Hodges et al., 2008) consists of ten spoken multisyllabic words of relatively low familiarity and high phonological complexity. The experimenter presents each spoken word and asks the participant to repeat it. The target word is then spoken again, and the participant is asked to select the picture corresponding to the named object from an array of seven pictures, all from the same semantic class. 2. The Naming Test from the Cambridge semantic memory battery (Garrard et al., 2001) consists of 64 line drawings of familiar objects, half living and half manmade, each presented individually to be named. 3. The spoken Word-to-Picture Matching Test from the same battery uses the same 64 items. The spoken name of each target item is to be matched to a line drawing of the object from an array that includes nine foil pictures from the same semantic category. 4. In the picture version of the Camel and Cactus Test (Bozeat et al., 2000), the participant is required to select, from amongst four coloured picture alternatives in the same semantic class, the one that is most closely associated with a fifth, target picture. The same 64 items constitute the targets. GF also performed this test, at a different session, with the targets and response alternatives presented as written words, whereas DJ only did the picture version. Additional more specific semantic tests included the following: (1) A Three Alternative Forced Choice (3AFC) Synonym Judgement Test manipulating both the word frequency and the imageability of the target word and the three response alternatives

Please cite this article as: Patterson, K., et al., Semantic memory: Which side are you on? Neuropsychologia (2014), http://dx.doi.org/ 10.1016/j.neuropsychologia.2014.11.024i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132

K. Patterson et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

4

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

(2)

(3)

(4)

(5)

(6)

(Jefferies et al., 2009). The words were shown in printed form and were also read aloud by the experimenter. A test of matching a spoken name of an object to a coloured photograph of its referent (in an array of seven photographs), as described by Rogers (2003); see also Rogers et al., this volume. Over three different sessions, the same target word (e.g. kingfisher) occurred three different times with three different levels of foil pictures: (a) very close semantic distractors (e.g. other small colourful birds); (b) close semantic distractors (e.g. an ostrich, a penguin); (c) vaguely related distractors (e.g. a fish, a gorilla). The other variable manipulated in this test is familiarity: for example, as mentioned above, kingfishers and robins are both small colourful birds but of lower and higher familiarity, respectively. A 2AFC colour decision test (Rogers et al., 2007), in which each trial consists of two pictures of the same object differing only in colour, one the correct colour (e.g. an orange carrot) and the other not (a green carrot). On half the trials, the correct colour is more typical of the object's category than the foil and, for other half, the typicality relationship is the other way round. A 2AFC written word lexical decision test (Rogers et al., 2004a, 2004b) with a similar design to the colour test. Each trial presented two letter strings, one real word and one pseudoword that can be pronounced in the same way as the real word. In half of the trials, the real word has a less typical orthographic spelling pattern than the pseudoword (e.g. fruit vs. frute); in the other half of matched trials, the orthographic typicality relationship was reversed (e.g. flute vs. fluit). Oral reading of the ‘surface list’ (Patterson and Hodges, 1992; Woollams et al., 2007) in which the words to be read aloud vary in frequency and typicality of spelling-sound correspondences. A naming test comprising 45 line drawings of objects (from Roberts, 2013) in which, if the patient was unable to name a picture, he was then given phonological cues in an incremental fashion until either he succeeded in naming it or the whole name had been presented. The benefit of such cueing for naming is minimal in SD but substantial in SA (semantic aphasia) (Jefferies et al., 2008).

3. Results Table 1 provides the data on the general cognitive tests for both patients, and Table 2 shows the results of the general assessments of semantic memory. The results of the general assessments are highly similar between GF and DJ and also similar to SD patients (Table 1). Table 1 Scores for the two patients on general cognitive assessments. Test

Max

MMSE

30

ACE-R Attention/orientation Memory Fluency Picture naming Visuospatial

100 18 26 14 2,10 16

Digit span forwards Digit span backwards TROG blocks TROG items

– – 20 80

GF

DJ

24

24

56 17 7 6 2, 0 15

60 16 8 9 2, 0 13

7 3 19 79

7 4 17 74

Table 2 Number and proportion correct for the two patients on semantic assessments. Test

Max.

GF: score (prop.)

DJ: score (prop.)

Repeat, Point

10,10

9, 4

10, 8

Picture naming Living Manmade

64 32 32

28 (0.44) 5 (0.16) 23 (0.72)

30 (0.47) 9 (0.28) 21 (0.66)

Word-picture match Living Manmade

64 32 32

51 (0.80) 22 (0.69) 29 (0.91)

60 (0.94) 28 (0.88) 32 (1.00)

Camel and cactus pictures Living Manmade

64 32 32

46 (0.72) 18 (0.56) 28 (0.88)

58 (0.91) 27 (0.84) 31 (0.97)

Camel and cactus words Living Manmade

64 32 32

42 (0.66) 20 (0.63) 22 (0.69)






The sub-components of the ACE-R on which these two patients performed relatively well—attention, orientation and visuospatial processing—are also those on which SD patients do well until very late in the course of the disease. The sub-components on which both patients performed poorly, which include memory (mainly verbal), are also where SD patients fail. The naming pattern is particularly reminiscent of SD: both patients correctly named the two highly familiar objects, but failed to name any of the ten less familiar ones. Their forward digit spans were entirely normal, and their backward spans were borderline normal. This is also characteristic of SD, though not of patients with SA following stroke (for example, an average of 3.7 digits forwards and 1.8 backwards in Jefferies and Lambon Ralph (2006)). Finally, syntactic comprehension on the TROG was almost perfect (GF) or good (DJ), as is observed in moderately impaired patients with SD. We now turn to the general semantic tests in Table 2. On Repeat-and-Point, the patients were perfect (GF) or nearly so (DJ) at repeating phonologically complex multisyllabic words, but made many errors (GF) or at least a few (DJ) in distinguishing the pictorial referents of these words from semantically similar category coordinates. A significant advantage for ‘repeat’ relative to ‘point’ is seen in SD (Hodges et al., 2008). On the 64-item picture naming test, both patients revealed significant anomia, and both were substantially better at naming the manmade artefacts than the living things, a phenomenon that will be addressed further below. As for naming errors, GF and DJ were similar to each other and also to SD patients in that they only made three different types of error: omissions (“don’t know”), single-word semantic errors (such as cherry-“apple”) and circumlocutions that were mostly relevant and informative (e.g. squirrel-“they go up and down trees”) though occasionally more empty (spanner-“I have been using one of those recently”). Unlike most anomic stroke patients but exactly like SD patients, neither GF nor DJ ever made either a phonological naming error or the kind of associative error reported for SA patients (such as squirrel-“nuts”). They did differ a little from each other in their proportions of the 3 error types, with GF producing more circumlocutions and DJ more “don’t know” responses; this could either reflect a difference in the amount of information that each patient derived from the pictures or a difference in response criterion. On the two semantic comprehension tests from the Cambridge Battery (word-picture matching and Camel and Cactus), the two patients were once again similar to each other, although DJ was more accurate than GF. Patients with SD, even early on in their progression, rarely if ever achieve scores like DJ's on these tests, and GF's scores were like those of mild SD.

Please cite this article as: Patterson, K., et al., Semantic memory: Which side are you on? Neuropsychologia (2014), http://dx.doi.org/ 10.1016/j.neuropsychologia.2014.11.024i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132

K. Patterson et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

5

Fig. 4. Comparison of the effect of object familiarity on proportion of correct responses in naming vs. comprehension for GF, DJ and SD patients on the 64-item Cambridge Picture Naming and Spoken Word-to-Picture matching tests. SD data are from 18 patients who completed both assessments within four months. Error bars represent 7 95% confidence intervals based on t-distribution. Fig. 2. Effect of object familiarity on percentage correct in naming by GF, DJ and SD patients on the 64-item Cambridge Picture Naming test. SD data are from 20 patients with a comparable average accuracy to that of GF and DJ (45%). Error bars represent 7 95% confidence intervals based on t-distribution.

As reviewed in the Introduction, two of the major determinants of performance by SD patients on any task requiring semantic memory are the degree of familiarity and typicality of the stimulus items. Unlike some of the tests yet to be reported, the Cambridge Semantic battery was not designed specifically to assess these factors, but the 64 items can be split into higher and lower halves with regard to either of these factors. Figs. 2 and 3 display the results of such a split on naming scores, for GF and DJ and a subset of 20 patients with SD from Woollams et al. (2008) whose mean naming performance matched that of GF and DJ; that is,  45% correct. Fig. 2 suggests that naming success for both of the patients was, if anything, even more sensitive to familiarity of the pictured items than is the case for patients with SD with an approximately equal degree of anomia; and Fig. 3 reveals an impact of object typicality for the two patients that is indistinguishable from that for the cases with SD. Note that significant effects of familiarity and typicality are not characteristic of patients with SA (Jefferies and Lambon Ralph, 2006; Jefferies et al., 2010).

Fig. 3. Effect of object typicality on percentage correct in naming by GF, DJ and SD patients on the 64-item Cambridge Picture Naming test. SD data are from 20 patients with a comparable average accuracy to that of GF and DJ (45%). Error bars represent 7 95% confidence intervals based on t-distribution.

We next determined whether familiarity of the items affected these two patients’ comprehension as well as production. Fig. 4 demonstrates these effects for the 64-item spoken Word-Picture Matching and naming tests, for GF and DJ and 18 cases with SD who performed these two tests in close time proximity. The figure plots the proportion of correct responses that come from the higher and lower halves of the familiarity distribution (proportion of correct was used to counter any differences in overall accuracy). Whereas GF, DJ and the SD group all reveal a powerful impact of familiarity on the ability to produce correct object names, only the SD patients show a parallel effect for comprehension: GF and DJ were no more likely to comprehend high than low familiarity items in word-picture matching. It is true that GF's and especially DJ's high levels of performance on this comprehension test leave relatively little room for any factor to have a strong effect; nevertheless, even the SD patients with the top eight Word-Picture Matching scores (Z0.70 overall) revealed an 11% advantage for higher4lower familiarity targets, whereas it was 2% for GF and 0 for DJ. As noted above and demonstrated in Table 2, GF and DJ had consistently superior performance for manmade stimulus items relative to natural kinds or living things, especially in naming. There is some debate in the literature as to whether SD patients are characterised by a semantic advantage for manmade things. Libon et al. (2013) reported this category effect for a sample of 15 SD patients, in a task where the patients were asked to judge the semantic categories of pictures or words. In a large analysis of picture naming (on the same 64-item naming test employed in the current study) by 78 SD patients of varying severity (Woollams et al., 2008), there was a small but statistically significant (8%) advantage for naming manmade as compared to living items. For the 20-case subset of this SD group mentioned above, with levels of naming accuracy more-or-less identical to GF and DJ, the average percent advantage for artefacts in the SD cases was about 11%, compared to 56% for GF and 38% for DJ. On the same naming test, stroke patients with SA do not display any category discrepancy (Jefferies, personal communication). As DJ was so much more successful at receptive semantic processing of the 64 pictures than at naming them, his advantage for manmade 4living things was also smaller on the comprehension tests (a 12% difference for Word-Picture Matching, 13% for the Camel and Cactus Test) whereas this effect was more robust for GF's comprehension (22% and 32%, respectively). By contrast, this category effect was minimal (6%) when GF was assessed on the word version of the Camel and Cactus Test, which might suggest that the effect is at least partly due to the high visual/structural similarity amongst pictures of animals and fruits.

Please cite this article as: Patterson, K., et al., Semantic memory: Which side are you on? Neuropsychologia (2014), http://dx.doi.org/ 10.1016/j.neuropsychologia.2014.11.024i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132

K. Patterson et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

6

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

Table 3 Performance by GF and DJ germane to the contrast with semantic dementia. Test

GF

DJ

Synonym judgement 3AFC Hi-Im,Hi-F Hi-Im,Lo-F Med-Im,Hi-F Med-Im,Lo-F Lo-Im,Hi-F Lo-Im,Lo-F

0.94 0.94 1.00 1.00 0.81 0.56

1.00 1.00 1.00 1.00 0.81 0.88

SD (N ¼ 11)a 0.86 0.67 0.86 0.46 0.62 0.33

0.86




Milder SD (N ¼ 5)b 0.82

Severe SD (N ¼ 6)b 0.48

0.77




0.60

0.30

0.77 0.82 0.55 0.55

1.00 1.00 0.81 0.64

0.75 0.55 0.74 0.46

0.44 0.17 0.35 0.16

Word-picture match 7AFC Vaguely-related foils, HiFam Vaguely-related foils, LoFam Close-ish foils, Hi-Fam Close-ish foils, Lo-Fam Very close foils, Hi-Fam Very close foils, Lo-Fam Colour decision 2AFC

Lexical decision 2AFC Hi-F, word more typical Hi-F, nonword more typical Lo-F, word more typical Lo-F, nonword more typical Reading aloud Hi-F, Lo-F, Hi-F, Lo-F,

3.2. Word-picture matching varying semantic distance of foils This test includes not only a manipulation of stimulus familiarity but also of specificity: the more similar the foils are to the target item in semantic category, the more specific the knowledge that is required to discriminate the target from the other pictures in the response array. GF and DJ were both influenced by this factor of specificity. Their performance was most affected by very close foils, but improved substantially thereafter. With one exception (DJ's success with very close foils), the performance of the two patients was largely unaffected by familiarity. Scores for 11 patients with SD, divided into milder and more severe cases (from Rogers, 2003; Rogers et al., this volume), are included in Table 3, and demonstrate the standard large advantage for high- relative to low-familiarity stimuli, as well as substantial effects of semantic distance of the foils. Because this is a 7AFC test, the more severe cases with SD were scoring essentially at chance level on low familiarity items whatever the semantic distance.

Milder SD (N ¼ 5)c 0.83 0.89 0.91 0.78 0.94 0.56

Real more typical Non-real more typical

typical typical atypical atypical

Milder SD (N ¼ 13)d 0.78 0.61 0.83 0.93 0.83 0.83

Severe SD (N ¼ 13)d 0.85 0.63

0.93 0.61 0.88 0.78 0.72 0.56

0.85 0.34

Milder SD (N ¼ 7)e 0.98 1.00 0.97 0.93 0.92 0.87 0.98 0.90 0.93 0.64 0.51 0.55

Severe SD (N ¼ 7)e 0.88 0.68 0.63 0.31

3.3. 2AFC colour decision This test was designed to assess patients’ sensitivity to typicality of the stimuli. One stimulus example consists of two pictures of a cucumber (one green and one orange); the participant is asked “which is the real one?”. For this item, the correct choice is also the more typical choice, because many vegetables are green. On another trial, however, the orange and green objects are carrots; here the incorrect choice is more typical. Table 3 indicates that this manipulation had no impact on GF's or DJ's success, whereas mildmoderate SD patients (from Rogers et al. (2007)) were dramatically better on this same test when typicality favoured the correct choice. More impaired SD cases, also tested by Rogerset al., were at or near the 50% chance level in both typicality conditions.

Picture naming with/without cues No cue 0.38 0.44 1
2 Phoneme cue 0.55 0.53 42 Phoneme cue 0.76 0.75 Abbreviations: familiarity.

Hi ¼ high;

Lo¼ low;

Im¼ imageability;

within any imageability band, on the more frequent words, with an average advantage of 30% overall. Their performance was also significantly graded by imageability. As this is a 3AFC test, the scores of these SD patients were exactly at chance for words low in both frequency and imageability.

F ¼ frequency;

Fam ¼

a

Jefferies et al. (2009). Rogers (2003). c Rogers et al. (2007). d Rogers et al. (2004a, 2004b). e Patterson et al. (2006).

3.4. 2AFC lexical decision

b

Finally, Table 3 indicates the performance of the two patients on the six tests more specifically designed for SD patients. 3.1. Synonym judgement This is the first of the tests given to the two patients in which a manipulation of familiarity was part of the test design. Despite performing well on this demanding semantic task, both were affected by the factor of imageability, with essentially perfect scores for the high and medium bands of this variable but a decline for low imageability words. DJ showed no impact of word frequency, whereas for GF the low imageability words proved more challenging when they were also of low frequency. Table 3 provides scores on this test from a group of 11 patients with SD reported by Jefferies et al. (2009). The exact level of success on this and any other semantic test in SD will depend upon stage of disease but, in contrast to GF and DJ, the SD cases had consistently higher scores,

Here each trial presents two choices, in this case a real word and a phonologically similar pseudoword. In half of the trials, the word has a more typical orthographic pattern than the pseudoword (e.g., booth/beuth), with the reverse for the other half (sleuth/ slooth); orthogonal to the typicality manipulation, the word is of either high or low frequency/familiarity. Table 3 indicates no consistent effect of either of these variables for GF or DJ. In data for SD patients from Rogers et al. (2004a, 2004b) on the identical test, summarised in Table 3, milder cases displayed a frequency-bytypicality interaction, with performance only substantially impaired when the correct word was low on both variables (although of course the patients were impaired relative to controls in all conditions). More severe cases with SD also displayed this interaction, but were yet more affected by typicality such that selection of even higher frequency real words was disturbed by the presence of a non-word with a more typical orthographic pattern. For this patient subgroup, the condition where the words had low familiarity and lower typicality than the pseudowords yielded performance below chance; that is, the patients were not choosing at random, but actually preferred a typically spelled nonword like slooth to an atypically-spelled low-familiarity real word like sleuth.

Please cite this article as: Patterson, K., et al., Semantic memory: Which side are you on? Neuropsychologia (2014), http://dx.doi.org/ 10.1016/j.neuropsychologia.2014.11.024i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132

K. Patterson et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 Q4 54 55 56 57 58 59 60 61 62 63 64 65 66

7

3.5. Reading aloud Deficits in the representation of semantic knowledge are almost invariably associated with a surface dyslexic pattern of reading: that is, success in reading words aloud is characterised by an interaction between word frequency and typicality of the spelling-sound correspondences of the words. This is very evident in patients with SD (see Woollams et al., 2007), to such an extent that surface dyslexia has recently been proposed as one of the classifying features for SD (or semantic variant primary progressive aphasia: Gorno-Tempini et al., 2011). Table 3 displays reading scores on the four conditions of the surface list frequently used in research on surface dyslexia. Reading performance by both GF and DJ was virtually identical to that of patients with mild SD reported by Patterson et al. (2006). Unlike GF, DJ and patients with SD, the reading performance of patients with post-stroke semantic aphasia, though impaired, is characterised by mild or absent effects of typicality and frequency and no interaction between these variables (Jefferies et al., 2010). Furthermore, whereas the vast majority of the errors in reading aloud for GF, DJ and patients with SD are regularisations (such as sew to rhyme with “few” and pint to rhyme with “hint”), most errors in reading by SA patients are not phonologically plausible pronunciations. 3.6. Naming with and without incremental phonological cueing The results of this test reveal that the two patients’ naming was facilitated but not resolved by cueing. Fig. 5 displays the pattern of naming performance, as a function of this incremental cueing procedure, for GF, DJ, a group of SD patients and a group of SA stroke patients (Jefferies et al., 2008). The starting points of the four curves are not identical (% correct without cueing: SA4DJ 4GF4SD) but they are surprisingly close; and in any case, the important point is not the absolute level of the un-cued baseline but rather (a) the degree of improvement as progressive cues are given, and (b) the level reached when a substantial chunk of the name has been provided to the patient. Performance by the SA patients leapt up with even the first one or two phonemes of the words and was near perfect with more than two. Performance by the SD patients crept up and did not quite reach even 60% correct when the patients were given three or more phonemes. Both GF and DJ fell almost precisely midway between SA and SD, reaching a maximum of only  75% success after three or more phonemes. 3.7. Other aspects of performance suggesting a deficit in semantic representation rather than “control” We did not assess GF or DJ on the kinds of tests, such as rapid and/or repeated presentation of small sets of related items, which are very challenging for some patients with semantic aphasia, especially those characterised as having “refractory semantic access” deficits (Crutch and Warrington, 2005; Jefferies et al., 2007). We also cannot do the type of cross-task correlation performed by Jefferies and Lambon Ralph (2006; see their Fig. 3e and f, p. 2140), for example between overall scores on the Camel-and-Cactus Test and those on the word-picture matching test employed here. This analysis, which revealed a highly significant correlation (r ¼0.96!) in SD patients but none in SA patients (r ¼0!), requires a case series of patients whereas we only have two. As a further indication that the semantic deficit in GF and DJ is better characterised as disrupted semantic representation rather than access/control, however, we can report two small further findings. DJ was tested on one of our naming tasks (the one used to investigate the effect of phonological cueing) on two different and well separated occasions without cues. His item-specific consistency was extremely

Fig. 5. The impact of phonemic cueing on naming for GF, DJ, SD and SA patients. SD data are from 8 patients and SA data are from 6 patients reported by Jefferies et al. (2008).

high: of the 45 items in the test, his response was either correct on both occasions or incorrect on both occasions for 39/45: only six items yielded discrepant outcomes across the two testing occasions. Secondly, GF—the only one of the two cases who made enough errors on the word-picture matching task to enable a sensible comparison with another harder test (naming) on the same items—demonstrated a high degree of item consistency across these two tests. Because naming is so much more difficult than multiple-choice matching, there were unsurprisingly a largeish number of items (25/64) for which he chose correctly in wordpicture matching but failed to produce the correct name. Of the remaining 39/64 items, however, 37/39 were either correct on both tests or wrong on both tests. Both of these types of consistency are more indicative of deficits in semantic representation than access/control.

4. Discussion This study examined two patients with profound damage of the left temporal lobe, including the rostral and ventral regions. The aetiology in both cases (stroke and HSVE) was acute or sub-acute. This same region becomes severely atrophic in SD patients, but over months and years. The study was designed to determine first of all whether these two patients would have a significant semantic impairment, and secondly, if so, whether it would have the features specifically characteristic of SD. The features explored were: the familiarity of the stimulus items; the typicality of the stimuli within the domain relevant to the task; the specificity of the semantic knowledge required to perform the task; and the degree of improvement of naming in response to phonological cues. The answer to the first question is yes: the two patients had a semantic deficit. Despite their large lesions, this impairment was moderate rather than severe, which probably should not be surprising: it is well known from both human and animal research that a unilateral lesion often results in restricted or even minimal cognitive impairment, whereas bilateral damage to the same region is more than twice as devastating (Heffner and Heffner, 1986; Scoville and Milner, 1957; Schapiro et al., 2013). With respect to the second issue, these two patients resembled SD in broad outline but differed in more detailed characteristics. Both patients were impaired on many of the tests that reveal deficits in SD, especially object naming and reading aloud which require verbal responses; and the nature of their errors in both of these tasks was also exactly the same as observed in SD. On receptive tasks like synonym judgements, word-picture matching or picture-picture associative matching, GF's accuracy usually resembled that of mild SD patients

Please cite this article as: Patterson, K., et al., Semantic memory: Which side are you on? Neuropsychologia (2014), http://dx.doi.org/ 10.1016/j.neuropsychologia.2014.11.024i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 Q5123 124 125 126 127 128 129 130 131 132

8

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

K. Patterson et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

on the same tests. DJ's scores on the receptive tests were mostly higher than GF's, and only revealed deficits in the most difficult experimental conditions, such as synonym judgements with lowfrequency abstract stimulus words and forced-choice word-picture matching with semantically very similar foils. The object-naming success of both patients was substantially affected by both the familiarity and typicality of the object concepts, as is true in SD (and not in SA). By contrast, whereas patients with SD also show substantial effects of these variables in the receptive tasks employed in the current study, neither GF nor DJ did. GF and DJ were both affected by the specificity of the knowledge required in the matching task varying semantic closeness of the foils, though less dramatically than is the case for SD patients. The other major feature addressed was the degree to which GF's and DJ's anomia could be resolved by phonological cueing. Incremental cueing yields minimal benefit in SD, indicating a deficit in semantic representations themselves rather than unreliable access to largely intact representations (Crutch and Warrington, 2005; Graham et al., 1995; Jefferies et al., 2008; Shallice, 1988). In contrast to SD, the cueing benefit in SA is impressive: even after a partial cue, success in naming approaches 80% correct, and after a larger cue it approaches 100%. GF and DJ were both intermediate in this regard: given maximal cues, they were both still anomic, naming 75% of the pictures, which is roughly equal to the level of success achieved by SA patients following minimal cues. It should be emphasised that maximal cues in this test provide almost all of the phonological information about the object's name, lacking only the terminal sound: if any reasonably precise semantic representation has been activated by the picture, the combination of that with this extensive phonological information from the cue should be sufficient to elicit the name. The fact that it often did not suggests that, though less profound than in SD, the semantic deficits in GF and DJ reflect genuine disruption of semantic representations rather than recalcitrant access to them. The aspects of GF's and DJ's performance on which we were able to assess intra-item consistency also suggest a deficit in semantic representation: These two patients revealed a high degree of predictability from one occasion to another on the same items in the same task or even across different tasks; such consistency is likewise high in SD but much lower or even absent in SA. A final task in which GF and DJ resembled mild SD (and not SA) is reading aloud: both of the patients had striking surface dyslexia, with many errors on lower frequency words with an atypical spelling-sound relationship. Reading aloud is a task requiring verbal output, even if the stimulus gives far more clues to the correct response than a picture does for naming. Furthermore, like SD patients, the great majority of GF's and DJ's errors in reading these atypical words were regularisations (e.g., sew-“sue”). The patients with SA described by Jefferies et al. (2010) were significantly, though mildly, impaired at reading aloud, but their errors to atypical words were rarely the precise regularisations characteristic of GF, DJ and patients with SD. The conclusion supported by this combination of features would seem to be that unilateral left temporal lobe damage, as long as it includes the rvTL region severely affected in SD, does produce a rather SD-like pattern of disruption to semantic representations. Unlike patients with semantic dementia, GF and DJ have an intact right temporal lobe; as a result, (a) their deficits on non-verbal semantic assessments are measurable but much milder than that typically observed in SD; (b) their deficits on any task that does not require verbal output are measurable but lack the sensitivity to familiarity and typicality that is ubiquitous in SD. This pattern of results is what we might expect to see if SD patients with predominantly left rvTL atrophy were assessed very early in the disease progression, as argued by Mion et al. (2010)

67 and as largely supported in studies by Graham et al. (1995) and 68 Mesulam et al. (2009). 69 As for more specific theoretical implications: there are two 70 different views about the role(s) of the left vs. right rvTL. One is 71 that they represent two separate brain networks, with the left 72 specialised for the processing and representation of verbal in73 formation and the right devoted to processing and representation 74 of objects and faces (Gainotti, 2012; Mesulam et al., 2013). The 75 second is that the left and right rvTLs represent one amodal se76 mantic network, sometimes referred to as a semantic hub (Lam77 bon Ralph et al., 2009). Despite their differences, both hypotheses 78 embrace the concept that, as neural processing moves from more 79 caudal modality-specific sensory regions (visual, auditory, somatosensory, etc) towards the rostral regions of the TL, the nature of 80 the representations becomes less feature-specific and more char81 acterised by higher-level feature conjunctions (Barense et al., 82 2007; Bussey and Saksida, 2002; Cowell et al., 2010; Clarke et al., 83 2011). Some models suggest that the semantic knowledge in the 84 rvTL may even have a form that does not contain any retrievable 85 features but instead represents concepts in a multidimensional 86 space reflecting the similarity of related concepts (Rogers et al., 87 2004a, 2004b). The results from the current study of GF and DJ do 88 not provide a basis for choosing between these views. The mi89 micking of the SD pattern but only for tasks requiring speech 90 output is of course easily encompassed by the proposal of separate 91 left and right ATL networks, with the left specialised for verbal 92 processing. It is also compatible with the single semantic hub 93 framework accounts, with reference to two principles. First, a great 94 deal of semantic processing involves communication from mod95 ality-specific input regions to the hub and from the hub to mod96 ality-specific output regions. Second, the majority of cortico-cor97 tical connections in the human brain are within hemisphere (Ril98 ling and Insel, 1999). Speech production is well known to be 99 strongly left-lateralised in right-handed individuals. Connections 100 from the hub to the regions involved in speech production will 101 therefore be significantly stronger in the left than the right 102 hemisphere. These principles were instantiated in a connectionist 103 model of object naming that included a single bilateral hub net104 work but with much stronger connections to speech from the 105 units on the left than from units on the right side of the hub. This 106 model was successful at reproducing the difference in degree of 107 anomia between SD-L and SD-R patients equated for degree of 108 comprehension deficit (Lambon Ralph et al., 2001). 109 In closing, a few other aspects of GF's and DJ's performance 110 deserve brief comment. Both patients had consistently better 111 scores for naming of manmade artefacts than living items. This 112 semantic variable also affected their performance on receptive 113 tests, to a lesser extent. More intensive testing would be required 114 to sort out the contributions to this effect of a modality-free se115 116 mantic category difference vs. the visual crowding or structural 117 similarity that is typical of the visual representations of natural 118 kinds but not of artefacts (e.g. Gale et al., 2001). Convincing de119 monstrations of the manmade 4natural advantage have mainly 120 been reported in patients with HSVE (e.g., Noppeney et al., 2007; 121 Warrington and Shallice, 1984; also DJ tested previously: Stanhope 122 and Kopelman, 2000), stroke (e.g., Caramazza and Shelton, 1998) 123 or head injury (Laiacona et al., 1993). There are a few reports of a significant category effect in SD cases (e.g., Barbarotto et al., 1995; Q6124 125 Lambon Ralph et al., 2003), but in larger case series, the effect— 126 even if significant—is usually mild (Libon et al., 2013; Woollams 127 et al., 2008). As far as we know, there are no reports of a category 128 advantage in SA patients. 129 A further observation was that GF, with extensive left-temporal 130 damage sparing only the superior temporal gyrus, had nearly 131 flawless performance on the TROG (Bishop, 1989); and DJ, with 132 almost complete destruction of the left temporal lobe extending

Please cite this article as: Patterson, K., et al., Semantic memory: Which side are you on? Neuropsychologia (2014), http://dx.doi.org/ 10.1016/j.neuropsychologia.2014.11.024i

K. Patterson et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

1 2 3 4 5 6 7 8 9 10 11 12 13 Q2 14 15 16 17 18 19 20 21 22 23 Q9 Q10 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

into the inferior frontal gyrus, also had a good if not perfect score on this test. Whatever brain network supports grammatical competence of this sort, it apparently does not depend heavily on these regions. Finally, the extent of damage to the left hemisphere was greater in DJ than in GF, and yet, with only a few exceptions, DJ's success on the tasks employed here was better than GF's. There is no way of knowing from the existing data whether GF's greater impairment relates to the age at onset, the timing of tests post insult, the aetiology, or some other factor.

Uncited references Corbett et al. (2009), Lambon Ralph and Patterson (2008), and Kopelman and Stanhope (1999).

Acknowledgements Written informed consent was obtained from the participants. The study was approved by the Regional Ethics Committee. This work was supported by the Wellcome trust.

References Acosta-Cabronero, J., Patterson, K., Fryer, T.D., Hodges, J.R., Pengas, G., Williams, G. B., et al., 2011. Atrophy, hypometabolism and white matter abnormalities in semantic dementia tell a coherent story. Brain 134, 2025–2035. Agosta, F., Henry, R.G., Migliaccio, R., Neuhaus, J., Miller, B.L., Dronkers, N.F., et al., 2010. Language networks in semantic dementia. Brain 133, 286–299. Barense, M.D., Gaffan, D., Graham, K.S., 2007. The human medial temporal lobe processes online representations of complex objects. Neuropsychologia 45, 2963–2974. Binney, R.J., Embleton, K.V., Jefferies, E., Parker, G.J., Lambon Ralph, M.A., 2010. The ventral and inferolateral aspects of the anterior temporal lobe are crucial in semantic memory: evidence from a novel direct comparison of distortioncorrected fMRI, rTMS and semantic dementia. Cereb. Cortex 20, 2728–2738. Bishop, D., 1989. Test for the Reception of Grammar: TROG. The Psychological Corporation, New York, NY. Bozeat, S., Lambon Ralph, M.A., Patterson, K., Garrard, P., Hodges, J.R., 2000. Nonverbal semantic impairment in semantic dementia. Neuropsychologia 38, 1207–1215. Bussey, T.J., Saksida, L.M., 2002. The organization of visual object representations: a connectionist model of effects of lesions in perirhinal cortex. Eur. J. Neurosci. 15, 355–364. Caramazza, A., Shelton, J.R., 1998. Domain-specific knowledge systems in the brain: the animate-inanimate distinction. J. Cogn. Neurosci. 10, 1–34. Clarke, A., Taylor, K.I., Tyler, L.K., 2011. The evolution of meaning: spatio-temporal dynamics of visual object recognition. J. Cogn. Neurosci. 23, 1887–1899. Corbett, F., Jefferies, E., Ehsan, S., Lambon Ralph, M.A., 2009. Different impairments of semantic cognition in semantic dementia and semantic aphasia: evidence from the non-verbal domain. Brain 132, 2593–2608. Cowell, R.A., Bussey, T.J., Saksida, L.M., 2010. Functional dissociations within the ventral object processing pathway: cognitive modules or a hierarchical continuum? J. Cogn. Neurosci. 22, 2460–2479. Crutch, S.J., Warrington, E.K., 2005. Gradients of semantic relatedness and their contrasting explanations in refractory access and storage semantic impairments. Cogn. Neuropsychol. 22, 851–876. Desgranges, B., Matuszewski, V., Piolino, P., Chetelat, G., Mezenge, F., Landeau, B., et al., 2007. Anatomical and functional alterations in semantic dementia: a voxel-based MRI and PET study. Neurobiol. Aging 28, 1904–1913. Folstein, M.F., Folstein, S.E., McHugh, P.R., 1975. “Mini-mental state”. A practical method for grading the mental state of patients for the clinician. J. Psychiatr. Res. 12, 189–198. Gainotti, G., 2012. The format of conceptual representations disrupted in semantic dementia: a position paper. Cortex 48, 521–529. Gale, T.M., Done, D.J., Frank, R.J., 2001. Visual crowding and category specific deficits for visual stimuli: a neural network model. Cogn. Neuropsychol. 18, 509–550. Garrard, P., Lambon Ralph, M.A., Watson, P., Powis, J., Patterson, K., Hodges, J.R., 2001. Longitudinal profiles of semantic impairment for living and nonliving concepts in dementia of Alzheimer's type. J. Cogn. Neurosci. 13, 892–909. Gorno-Tempini, M.L., Hillis, A.E., Weintraub, S., Kertesz, A., Mendez, M., Cappa, S.F., et al., 2011. Classification of primary progressive aphasia and its variants. Neurology 76, 1006–1014. Graham, K., Patterson, K., Hodges, J.R., 1995. Progressive pure anomia: insufficient activation of phonology by meaning. Neurocase 1, 25–38.

9

Hodges, J.R., Patterson, K., Oxbury, S., Funnell, E., 1992. Semantic dementia. Progressive fluent aphasia with temporal lobe atrophy. Brain 115, 1783–1806. Hodges, J.R., Martinos, M., Woollams, A.M., Patterson, K., Adlam, A.L., 2008. Repeat and point: differentiating semantic dementia from progressive non-fluent aphasia. Cortex 44, 1265–1270. Hodges, J.R., Patterson, K., 2007. Semantic dementia: a unique clinicopathological syndrome. Lancet Neurol. 6, 1004–1014. Jefferies, E., Lambon Ralph, M.A., 2006. Semantic impairment in stroke aphasia versus semantic dementia: a case-series comparison. Brain 129, 2132–2147. Jefferies, E., Patterson, K., Jones, R.W., Lambon Ralph, M.A., 2009. Comprehension of concrete and abstract words in semantic dementia. Neuropsychology 23, 492–499. Jefferies, E., Patterson, K., Lambon Ralph, M.A., 2008. Deficits of knowledge versus executive control in semantic cognition: insights from cued naming. Neuropsychologia 46, 649–658. Jefferies, E., Rogers, T.T., Hopper, S., Lambon Ralph, M.A., 2010. “Pre-semantic” cognition revisited: critical differences between semantic aphasia and semantic dementia. Neuropsychologia 48, 248–261. Kopelman, M.D., Stanhope, N., Kingsley, D.E.P., 1999. Retrograde amnesia in patients with diencephalic, temporal lobe or frontal lesions. Neuropsychologia 37, 939–958. Laiacona, M., Barbarotto, R., Capitani, E., 1993. Perceptual and associative knowledge in category specific impairment of semantic memory: a study of two cases. Cortex 29, 727–740. Lambon Ralph, M.A., Cipolotti, L., Manes, F., Patterson, K., 2010. Taking both sides: do unilateral anterior temporal lobe lesions disrupt semantic memory? Brain 133, 3243–3255. Lambon Ralph, M.A., Ehsan, S., Baker, G.A., Rogers, T.T., 2012. Semantic memory is impaired in patients with unilateral anterior temporal lobe resection for temporal lobe epilepsy. Brain 135, 242–258. Lambon Ralph, M.A., McClelland, J.L., Patterson, K., Galton, C.J., Hodges, J.R., 2001. No right to speak? The relationship between object naming and semantic impairment: neuropsychological abstract evidence and a computational model. J. Cogn. Neurosci. 13, 341–356. Lambon Ralph, M.A., Patterson, K., 2008. Generalisation and differentiation in semantic memory: insights from semantic dementiaAnn. NY Acad. Sci. 1124, 61–76. Lambon Ralph, M.A., Patterson, K., Garrard, P., Hodges, J.R., 2003. Semantic dementia with category specificity: a comparative case-series study. Cogn. Neuropsychol. 20, 307–326. Lambon Ralph, M.A., Pobric, G., Jefferies, E., 2009. Conceptual knowledge is underpinned by the temporal pole bilaterally: convergent evidence from rTMS. Cereb. Cortex 19, 832–838. Libon, D.J., Rascovsky, K., Powers, J., Irwin, D.J., Boller, A., Weinberg, D., et al., 2013. Comparative semantic profiles in semantic dementia and Alzheimer's disease. Brain 136, 2497–2509. Mesulam, M., Rogalski, E., Wieneke, C., Cobia, D., Rademaker, A., Thompson, C., et al., 2009. Neurology of anomia in the semantic variant of primary progressive aphasia. Brain 132, 2553–2565. Mesulam, M.-M., Wieneke, C., Hurley, R., Rademaker, A., Thompson, C.K., Weintraub, S., et al., 2013. Words and objects at the tip of the left temporal lobe in primary progressive aphasia. Brain 136, 601–618. Mion, M., Patterson, K., Acosta-Cabronero, J., Pengas, G., Izquierdo-Garcia, D., Hong, Y.T., et al., 2010. What the left and right fusiform gyri tell us about semantic memory. Brain 133, 3256–3268. Mioshi, E., Dawson, K., Mitchell, J., Arnold, R., Hodges, J.R., 2006. The Addenbrooke's Cognitive Examination Revised (ACE-R): a brief cognitive test battery for dementia screening. Int. J. Geriatr. Psychiatry 21, 1078–1085. Nestor, P.J., Fryer, T.D., Hodges, J.R., 2006. Declarative memory impairments in Alzheimer's disease and semantic dementia. Neuroimage 30, 1010–1020. Noppeney, U., Patterson, K., Tyler, L.K., Moss, H., Stamatakis, E., Bright, P., et al., 2007. Temporal lobe lesions and semantic impairment: a comparison of Herpes Simplex Virus Encephalitis and Semantic Dementia. Brain 130, 1138–1147. Patterson, K., Hodges, J.R., 1992. Deterioration of word-meaning: implications for reading. Neuropsychologia 30, 1025–1040. Patterson, K., Lambon Ralph, M.A., Jefferies, E., Woollams, A., Jones, R., Hodges, J.R., et al., 2006. “Presemantic” cognition in semantic dementia: six deficits in search of an explanation. J. Cogn. Neurosci. 18, 169–183. Patterson, K., Nestor, P.J., Rogers, T.T., 2007. Where do you know what you know? The representation of semantic knowledge in the human brain. Nat. Rev. Neurosci. 8, 976–987. Peelen, M., Caramazza, A., 2012. Conceptual object representations in human anterior temporal cortex. J. Neurosci. 32, 15728–15736. Pobric, G., Jefferies, E., Lambon Ralph, M.A., 2010. Amodal semantic representations depend on both anterior temporal lobes: evidence from repetitive transcranial magnetic stimulation. Neuropsychologia 48 (5), 1336–1342. Reed, L.J., Lasserson, D., Marsden, P., Bright, P., Stanhope, N., Kopelman, M.D., 2005. . Correlations of regional cerebral metabolism with memory performance and executive function in patients with herpes encephalitis or frontal lobe lesions. Neuropsychology 19, 555–565. Rilling, J.K., Insel, T.R., 1999. Differential expansion of neural projection systems in primate brain evolution. Neuroreport 10 (7), 1453–1459. Roberts, D.J., 2013. A Picture Paints a Thousand Words: Examining the Visual Impairment in “Pure” Alexia. University of Manchester (Doctoral thesis).

Please cite this article as: Patterson, K., et al., Semantic memory: Which side are you on? Neuropsychologia (2014), http://dx.doi.org/ 10.1016/j.neuropsychologia.2014.11.024i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132

10

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

K. Patterson et al. / Neuropsychologia ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Rogers T.T. What accounts for patterns of sparing and impairment in patients with general semantic deficits? Paper Presented to the British Neuropsychological Society, London, November 2003. Rogers, T.T., Patterson, K., 2007. Object categorization: reversals and explanations of the basic-level advantage. J. Exp. Psychol.: Gen. 136, 451–469. Rogers, T.T., Patterson, K., Graham, K.S., 2007. Colour knowledge in semantic dementia: It's not all black and white. Neuropsychologia 45, 3285–3298. Rogers, T.T., Lambon Ralph, M.A., Garrard, P., Bozeat, S., McClelland, J.L., Hodges, J.R., et al., 2004a. Structure and deterioration of semantic memory: a neuropsychological and computational investigation. Psychol. Rev. 111, 205–235. Rogers, T.T., Lambon Ralph, M.A., Hodges, J.R., Patterson, K., 2004b. Natural selection: the impact of semantic impairment on lexical and object decision. Cogn. Neuropsychol. 21, 331–352. Schapiro, A.C., McClelland, J.L., Welbourne, S.R., Rogers, T.T., Lambon Ralph, M.A., 2013. Why bilateral damage is worse than unilateral damage to the brain. J. Cogn. Neurosci. 25, 2107–2123. Scoville, W.B., Milner, B., 1957. Loss of recent memory after bilateral hippocampal lesions. J. Neurol. Neurosurg. Psychiatry 20, 11–21. Shallice, T., 1988. From Neuropsychology to Mental Structure. CUP, Cambridge. Snowden, J.S., Goulding, P.J., Neary, D., 1989. Semantic dementia: a form of circumscribed cerebral atrophy. Behav. Neurol. 2, 167–182. Snowden, J.S., Goulding, P.J., Neary, D., Mann, D.M.A., 1996. Frontotemporal Lobar Degeneration: Frontotemporal Dementia, Progressive Aphasia, Semantic Dementia. Churchill Livingstone, London. Snowden, J.S., Thompson, J.C., Neary, D., 2004. Knowledge of famous faces and names in semantic dementia. Brain 127, 860–872. Stanhope, N., Kopelman, M.D., 2000. Art and memory: a 7-year follow-up of herpes encelphalitis in a professional artist. Neurocase 6, 99–110.

Thompson, S.A., Patterson, K., Hodges, J.R., 2003. Left/right asymmetry of atrophy in semantic dementia. Neurology 61, 1196–1203. Tulving, E., 1972. Episodic and semantic memory. In: Tulving, E., Donaldson, W. (Eds.), Organisation of Memory. Academic Press, New York and London382–403. Warrington, E.K., 1975. The selective impairment of semantic memory. Q. J. Exp. Psychol. 27, 635–657. Warrington, E.K., Shallice, T., 1984. Category specific semantic impairments. Brain 107, 829–854. Wechsler, D., 1987. Wechsler Memory Scale: Revised (WMS-R). Psychological Corporation, New York. Woollams, A.M., Cooper-Pye, E., Hodges, J.R., Patterson, K., 2008. Anomia: a doubly typical signature of semantic dementia. Neuropsychologia 46, 2503–2514. Woollams, A.M., Lambon Ralph, M.A., Plaut, D.C., Patterson, K., 2007. SD-squared: on the association between semantic dementia and surface dyslexia. Psychol. Rev. 114, 316–339. Yamamoto, R., Iseki, E., Higashi, S., Murayama, N., Minegishi, M., Sato, K., et al., 2009. Neuropathological investigation of regions responsible for semantic aphasia in frontotemporal lobar degeneration. Dement. Geriatr. Cogn. Disord. 27, 214–223.

Please cite this article as: Patterson, K., et al., Semantic memory: Which side are you on? Neuropsychologia (2014), http://dx.doi.org/ 10.1016/j.neuropsychologia.2014.11.024i

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132