Pantomime of tool use: looking beyond apraxia

Abstract Pantomime has a long tradition in clinical neuropsychology of apraxia. It has been much more used by researchers and clinicians to assess tool-use disorders than real tool use. Nevertheless, it remains incompletely understood and has given rise to controversies, such as the involvement of the left inferior parietal lobe or the nature of the underlying cognitive processes. The present article offers a comprehensive framework, with the aim of specifying the neural and cognitive bases of pantomime. To do so, we conducted a series of meta-analyses of brain-lesion, neuroimaging and behavioural studies about pantomime and other related tasks (i.e. real tool use, imitation of meaningless postures and semantic knowledge). The first key finding is that the area PF (Area PF complex) within the left inferior parietal lobe is crucially involved in both pantomime and real tool use as well as in the kinematics component of pantomime. The second key finding is the absence of a well-defined neural substrate for the posture component of pantomime (both grip errors and body-part-as-tool responses). The third key finding is the role played by the intraparietal sulcus in both pantomime and imitation of meaningless postures. The fourth key finding is that the left angular gyrus seems to be critical in the production of motor actions directed towards the body. The fifth key finding is that performance on pantomime is strongly correlated with the severity of semantic deficits. Taken together, these findings invite us to offer a neurocognitive model of pantomime, which provides an integrated alternative to the two hypotheses that dominate the field: The gesture-engram hypothesis and the communicative hypothesis. More specifically, this model assumes that technical reasoning (notably the left area PF), the motor-control system (notably the intraparietal sulcus), body structural description (notably the left angular gyrus), semantic knowledge (notably the polar temporal lobes) and potentially theory of mind (notably the middle prefrontal cortex) work in concert to produce pantomime. The original features of this model open new avenues for understanding the neurocognitive bases of pantomime, emphasizing that pantomime is a communicative task that nevertheless originates in specific tool-use (not motor-related) cognitive processes.


Brain lesion studies
For the selection of brain lesions studies, we used the terms 'apraxia' AND 'brain lesions' to identify studies of interest for each of our analyses (i.e., real tool use, pantomime of tool use, and imitation of meaningless gestures). We used the search engine "PubMed".
This search returned 773 results at the date of 22/12/2020. We restricted our selection for each analysis to studies that met a series of selection criteria: 1. Only patients presenting exclusively left brain-damaged patients were considered.
2. Single case studies were not included. 3. Only studies using lesion mapping method were included (e.g., lesionsubtraction analysis or voxel-based lesion symptom mapping). 4. Reports had to provide lesion peaks data or at least overlay lesion plots associated with selective disturbances.
When coordinates lesions were available, we depicted the location of the reported lesion sites on flat-map representations of a left hemisphere (PALS-B12; Population-Average, Surface and Landmark-based human cortical atlas 1 ). If only lesion plots were reported in the study, we made several transformations before depicting the lesion sites on the flatmap representation, by following a three-step method. This three-step method is completely described in Lesourd et al. 2 (see also 3 ). Finally, we used the parcellation of Glasser et al. 4 to identify the brain areas of interest and considered a brain area as a brain area of interest when a brain area was reported in all the "modalities" (e.g., modality for pantomime of tool use or type of posture for imitation of meaningless postures) or in at least 50% of the conditions included for a given "modality" (e.g., visual modality for pantomime of tool use).

Neuroimaging studies
For the selection of neuroimaging studies, we used the terms 'pantomime' AND 'neuroimaging' to identify studies of interest for our analysis on pantomime of tool use, and 'imitation' AND 'neuroimaging' to identify studies of interest for our analysis on imitation of meaningless gestures. We used the search engine "PubMed". This search returned 90 results at the date of 22/12/2020 for 'pantomime' AND 'neuroimaging' and 557 results at the date of 22/12/2020 for 'imitation' AND 'neuroimaging'. We also conducted an additional search with the terms 'gesture' AND 'neuroimaging', which returned 448 results at the date of 22/12/2020. We restricted our selection for each analysis to studies that met a series of selection criteria: 1. Reviews were excluded.
2. Studies had to use functional magnetic resonance imaging (fMRI) or positron emission tomography (PET) as imaging modality.
3. Only neurologically healthy adults were included. 4. Neuroimaging results had to be based on whole-brain scanning. Regions of interest analyses were therefore excluded from our selection. 5. The complete list of activation peaks (i.e., foci) of main effects with their coordinates had to be reported in a stereotactic space.
The meta-analysis of neuroimaging studies was conducted using the revised version 5,6 of the Activation Likelihood Estimation 7 (ALE) method, as implemented by the GingerALE 2.3 software (http://www.brainmap.org/ale/). The methods employed here is the same as the one described by Reynaud et al. 8 (see also 2,9,10 ). Finally, the resulting thresholded ALE maps were visualized on flatmap representations of a standardized brain atlas (PALS-B12) using Caret, version 5.65 1 (http://brainmap.wustl.edu/caret.html).

Studies of patients with selective semantic deficits
For the selection of studies of patients with selective semantic deficits, we did not use the search engine "PubMed" because of the difficulty to use the appropriate terms.
Indeed, in such studies, the assessment of apraxia can be very "peripheral" and difficult to be identified by a search engine. Instead, we conducted a bibliographical search from the key papers on the topic [11][12][13] . We reported only the papers in which the data on semantic knowledge tasks and pantomime/single tool use were available for each patient. Note that we did not find any paper in which a correlation coefficient was mentioned between semantic knowledge tasks and pantomime/single tool use without the individual data for each patient. Said differently, the data reported in the present review correspond to all the data we have obtained on this topic.

Supplementary Information
All the references mentioned in the tables are cited in the main text.  Fig. 2A N, Number of left brain-damaged patients included in the study. *, This study reports a significant cluster of 81981 voxels in the internal capsule (white matter) in Table 3 (MNI coordinates: x = -24, y = -24, z =12). Given the size of the cluster, it appeared inappropriate to report only one coordinate. This explains why we used the three-step method based on the lesion plots depicted in the Figure 5 of this study. **, The score reported here corresponds to a composite score "pantomime + symbolic gestures". # , In these three studies, a picture of the object is shown along with the verbal command. We considered that the verbal modality is privileged over the visual modality and assigned these three studies to the "verbal" category.