Op-brai140361 456..471

verbally-mediated’ representations, instead of ‘image-based’ ones (Noppeney and Price, 2004). Performance on the Wisconsin Card Sorting Test has been correlated with activation in the dorsolateral prefrontal cortex (Nagahama et al., 1996). Performance on the Delis-Kaplan Executive Function System has also been correlated with left frontal lobe volume (Fine et al., 2009), as has a total abstract-reasoning score based on similarities and proverb interpretation (Kramer and Quitania, 2007). The generation of inappropriate concrete responses in this latter task has been associated with lesions in the left lateral frontal lobe, whereas overall performance is significantly impaired in patients with lesions in the medial frontal cortex (Murphy et al., 2013). A recent lesion-mapping study has shown that lower performance in verbal comprehension tasks of the WAIS, including the similarities subtest, is related to lesions in the left inferior frontal gyrus (Gläscher et al., 2009), a frontal area that was also activated in a functional MRI study of taxonomic categorization (Sachs et al., 2008). Nevertheless, it is worth noting that other disorders in which direct cortical lesions are less pronounced, such as autism, have also been related to altered performance in tests of abstract thinking and concept formation with a bias towards concrete responses (Minshew et al., 2002; Frith, 2003; Ropar and Peebles, 2007). Furthermore, as numerous cognitive components seem to contribute to verbal concept formation (Reverberi et al., 2005; Fuster, 2008), it is likely that this process relies on a distributed network of brain areas, rather than a unique and circumscribed region. It is for this reason that we also assessed a group of patients presenting with the ‘classic’ SteeleRichardson form of progressive supranuclear palsy (PSPr), an atypical parkinsonian syndrome characterized by oculomotor palsy, gait disturbance and cognitive dysfunction (Williams and Lees, 2009). The subcortical lesions affecting cognitive and limbic prefrontal-basal-ganglion-prefrontalcortex circuits in PSPr are severe, and have resulted in making the frontal-like impairments seen in PSPr the prototype of ‘subcortical dementia’ (Albert et al., 1974), although the presence of direct cortical lesions predominantly involving the posterior portions of the frontal cortex is also well established (Verny et al., 1996; Kertesz et al., 2010). Furthermore, although PSPr and Neural bases of verbal concept formation BRAIN 2015: 138; 456–471 | 457 by gest on O cber 6, 2016 http://braiordjournals.org/ D ow nladed from behavioural variant FTD patients have distinct grey matter atrophy patterns when compared to controls, a direct comparison of these two groups did not reveal any difference that persisted after correction for multiple comparisons in a recent study (Lagarde et al., 2013b). Based on their comparable clinical phenotype with regard to cognitive functions, characterized by a demonstrated severe dysexecutive syndrome in both cases, it would appear that behavioural variant FTD and PSPr patients can be pooled together into a ‘frontal’ group. Nevertheless, our aforementioned study also reveals that in spite of their mostly comparable clinical phenotypes and cortical atrophy patterns, the dysexecutive ‘frontal-like’ syndromes of behavioural variant FTD and PSPr are associated with partially divergent neural circuits (Lagarde et al., 2013b). To summarize, the addition of patients with PSPr provides an opportunity to study the involvement of the frontal syndrome as a whole in poor concept formation abilities, regardless of the pathophysiological or topographic entity involved, and to look for possible quantitative or qualitative differences between patients with behavioural variant FTD and those with PSPr that could be instructive with respect to the neural bases of this cognitive process. In clinical practice, patients with frontal damage usually provide two main types of inappropriate responses when performing a verbal concept formation task such as the similarities task, which relies on the ability to detect similarities between items and group them into abstract categories. When asked ‘In what way are an orange and a banana alike?’ these patients do not always spontaneously answer that they are both fruits, and many are not able to indicate that they belong to the same taxonomic category. These patients either remain stuck in concreteness, stating for instance that an orange and a banana share some perceptual features (‘they are sweet’, ‘they have a peel’, ‘they can be eaten’. . .), or they emphasize the differences between them (e.g. ‘an orange is round and a banana long’) (Dubois et al., 2000). The precise neurological substrate of these two types of inappropriate answers has never been directly addressed, and they have often been attributed to a general executive dysfunction (Giovannetti et al., 2001). Could the unexpected answers (concrete link and discrimination) observed in frontal patients be explained by the disruption of a unique mechanism (e.g. a systematic bias towards concrete features or perceptual details instead of abstract and more global representations), leading to either the discrimination or the linking of items on a concrete basis depending on whether these perceptual features are either divergent or convergent, respectively? Contrarily, can this complex cognitive process be dissociated into distinct components, such as an ability to link items (i.e. to make and/ or select a convergent representation) and an ability to provide an abstract (i.e. taxonomic) representation? To study these issues, we designed a new experimental paradigm based on similarities, called the Verbal Concept Formation Task, for an optimized analysis of verbal concept formation. A test of similarities seemed to be the most appropriate methodology, as the answer consists of explicitly providing a link rather than choosing between alternatives. However, existing tests, such as the similarities subtest of the WAIS, are based on a small number of items, which are linked in some cases according to taxonomic category, but in other instances based on theme/ mode (e.g. dictionary and directory: notion of alphabetical order) or general knowledge (e.g. rubber and paper: are obtained from trees). Our new experimental task was aimed at homogenizing the material, improving quantitative analysis and yielding stronger inferences by increasing the number of items. Concept formation performance was studied in healthy participants and compared to those of patients with behavioural variant FTD, Alzheimer’s disease or amnestic mild cognitive impairment due to Alzheimer’s disease, and PSPr. Finally, to investigate the cognitive mechanisms and neural bases underlying verbal concept formation, a dissection of the types of responses provided in the Verbal Concept Formation Task as well as their correlations to other neuropsychological tests were performed. We also correlated the scores obtained with grey matter volume for all participants, by performing an exploratory whole-brain analysis, without prespecified anatomical regions of interest, because of the lack of robust and converging information on this subject in the literature. The aims of the present study are as follows: (i) to confirm the alteration of verbal concept formation in patients with behavioural variant FTD using abstract categorization, and to see if it is present to the same extent in Alzheimer’s disease or amnestic mild cognitive impairment due to Alzheimer’s disease, as has sometimes been stated; (ii) to verify if altered performance in concept formation depend on direct prefrontal lesions such as those present in behavioural variant FTD or if it could be explained by an indirect frontal syndrome (e.g. via dysfunctions of prefrontal-subcortical-prefrontal circuits); and (iii) to obtain new insights into the cognitive mechanisms and anatomical bases of verbal concept formation, and more precisely to verify the hypothesis that verbal concept formation relies on two distinct cognitive processes underlain by two, at least partially different, neural circuits. Materials and methods The ethics committee of the Salpêtrière Hospital (Paris, France) approved the study. All participants gave written informed consent.