In our preliminary study, conducted during ‘Neural 2023’, as part of the Brainhack initiative, we explored the theory that different networks underpin conceptual versus production-related deficits in praxis. This distinction is a cornerstone of traditional cognitive and behavioural models of apraxia (Leiguarda & Marsden 2000, Heilman & Rothi 2003, Rounis & Binkofski, 2023). The ‘indirect’ route to action utilises familiar gestures by connecting the present visual scene with the gesture’s semantic memory. In contrast, a ‘direct’ route allows gesture execution based solely on visual transformation, sidestepping semantic knowledge. By probing the effects of lesion location on white matter disconnection pathways, our study was able to identify disconnections specific to the meaningless gesture imitation task. Our findings underscore the importance of the splenium of the corpus callosum in mediating ‘direct’ route to action. Due to the limited number of patients in our study, we observed only a trend for the ‘indirect’ route, suggesting a potential implication of the left fornix (Supplemental Figure 1).
A study by Metzgar et al. (2022) previously delved into these behavioural manifestations using a comparable measure of brain disconnection. Their focus was on discerning which cluster of brain areas were disconnected, as opposed to our emphasis on white matter pathways. Out of a cohort of 29 left hemisphere stroke patients, they pinpointed two individuals with small lesions who exhibited pronounced deficits on pantomime of object use or meaningless gesture imitation tasks. Pantomiming familiar tool-use correlated with disconnections between left temporal and parietal regions. On the other hand, deficits in imitating meaningless gesture were linked with disconnection between the left inferior and superior parietal lobules, and the left middle and superior frontal gyri. However, unlike our current study, which identified white matter tracts using a deterministic approach, the Metzgar et al. (2022) study utilized a probabilistic ‘shortest path tractography’ approach. This approach investigated cortico-cortical disconnection overlooking large white matter tract pathways. Building upon their findings, our research offers insights into white matter tract disconnections. This unveiled specific dis-connectivity patterns corresponding to the two-system praxis model (Leiguarda & Marsden, 2000).
A compelling finding from our current results is the detection of a disconnection of left hemisphere’s extrastriate visual areas intrahemispherically through the splenium of the corpus callosum in patients exhibiting deficits in meaningless gesture imitation tasks. This suggests a pivotal role of transcallosal pathways from the right hemisphere in deficits underlying limb apraxia. Historical neuropsychology and lesion symptom mapping portrayed apraxia primarily as a ‘left hemisphere’ syndrome (Liepman, 1908, Leiguarda & Marsden, 2000, Pizzamiglio et al., 2019). Specifically, the left ventro-dorsal stream, which includes the inferior parietal lobe in the left hemisphere is believed to enable ‘blueprints’ of familiar gestures. They are also known as ‘affordance’ — actions triggered by visual representations of tools (Gibson J, 1979, Watson & Buxbaum, 2015, Rounis & Humphreys, 2015, Sakreida et al., 2016). In contrast, the dorso-dorsal streams bilaterally implement sensorimotor transformations required for a ‘direct’ route of meaningless gestures (Kalenine et al., 2010, Buxbaum et al., 2014). Moreover, there is evidence suggesting apraxia in the right hemisphere, predominantly affects meaningless gesture imitation, linked to visuospatial deficits causing neglect (Dressing et al. 2020). Our findings of a significant transcallosal disconnection of the splenium could support a mechanism underlying these deficits. A disconnection of the splenium is likely to impair the transfer of perceptual information relating to hand posture and body schema (Bracci et al. 2018, Buxbaum et al. 2014) for meaningless gesture imitation from the contralateral hemisphere.
The significance of the opposite hemisphere in the recovery of visuo-perceptual abilities after a stroke and its impact on motor recovery have been previously discussed (Karolis et al. 2019, Mattos et al. 2021). It is hypothesised to mediate some forms of ‘blindsight’ following brain damage, such as stroke, which is associated with the ability to respond to visual stimuli despite reporting no awareness of them (Ajina & Bridge, 2016, Celeghin et al., 2017, Danckert et al., 2021). This influence may have been under-appreciated in limb apraxia after stroke.
In addition, the disconnection identified in meaningless gesture imitation deficits extended interhemispherically through the inferior longitudinal fasciculus. This finding highlights the importance of extrastriate brain pathways in inferior temporal lobe (Mahon et al. 2007, Saygin 2007, Bracci et al. 2018) and their role in integrating perceptual information to guide action goals. We conjecture that this interhemispheric disconnection may affect the contribution of extrastriate visual areas in integrating perceptual information to an action plan leading to ‘ideomotor’ deficits in apraxia (Watson et al. 2015, Zhang et al. 2021, Rounis & Binkofski, 2023).
Taken together, our results corroborate recent findings suggesting the importance of integrating high-order perceptual gesture plans into the sensorimotor transformations implemented within each hemisphere, which likely involve temporal lobe areas (Buxbaum et al. 2014, Pizzamiglio et al. 2019, Sperbert et al. 2019).
Our analyses also revealed a noteworthy but non-significant trend for disconnection involving the left arm of the fornix in the context of gesture production task. Importantly this lack of significance might stem from the small sample size of our study (29 patients). Nevertheless, if these findings are corroborated in future, larger scale studies, they could offer some scientific insight. Specifically they would concur with a previous conjecture that deficits underlying ‘ideational’ apraxia might arise from patients’ difficulties in retrieving actions ‘from memory’. This potential ‘indirect’ pathway to action could be mediated through the fornix, resulting in a disconnection of visual input from hippocampus— a crucial area underpinning flexible cognitive behaviours within the limbic lobe. The latter has been implicated in apraxia given presence of the disorder in stroke (see review on ventral stream regions above) and its implication has also been highlighted in several neuro-degenerative disorders including Alzheimer’s disease and fronto-temporal dementias (Crutch et al., 2007, Bozeat et al., 2002).
Mapping behavioural deficits using disconnection approaches
Despite clear distinction in the errors patients make on apraxia tasks, the evidence of separate networks underlying the two ‘routes to action’ – namely ‘ideational’ and ‘ideomotor’ apraxia, has been lacking (Leiguarda & Marsden, 2000) and has not matched the similarly termed ‘dual stream’ hypothesis used to describe visuomotor pathways (Goodale & Milner, 1992). Reasons for not identifying differences include limitations in our imaging and behavioural approaches to studying apraxia. As mentioned in the introduction, conventional imaging methods have historically relied on identifying lesion locations to map clinical neuropsychological manifestations onto the brain. However, it is essential to recognise that disconnection and diaschisis do not exclusively affect the lesioned area but instead extend their effects across broader and more distant networks, encompassing both structurally and functionally connected regions. Although, diffusion tractography studies have enabled the in vivo exploration of white matter connections, their application can encounter limitations due to technical challenges affecting interpretation and analyses. For instance, lesioned brains caused by injury or stroke may cause distortions, making probabilistic seed-based tractography difficult to interpret. one way to circumvent such issue has been to integrate lesions onto large cohorts such as those from the Human Connectome Project (HCP, https://www.humanconnectome.org/) acquired using high-resolution imaging and advanced deterministic tractography techniques (Foulon et al., 2018, Talozzi et al., 2023). This innovative approach allows for a deeper understanding, as the acquisition of large healthy control datasets, like the HCP, enables the development of templates that can now facilitate the estimation of lesion locations and disconnections in a more accurate and comprehensive manner.
Furthermore, several challenges hamper the examination of the neuropsychological consequences of stroke. While some studies have sought to establish double dissociations (Metzgar et al., 2022), it is important to note that various deficits may co-occur. For instance, patients may exhibit deficits in both pantomiming tool use and imitation (Buxbaum & Randerath, 2018). Besides the possibility that big lesions can impair both systems jointly, this co-occurrence could be attributed to the presence of shared cognitive processes that underlie these deficits, either in a ‘domain-general’ sense, such as the visual interpretation of gesture locations, or in a more ‘domain-specific’ manner such as those related to tool use or manipulation. Similar challenges have encountered in the study of aphasia and- neglect, where limited test selection sometimes fail to assess the full spectrum of heterogeneous deficits underlying these neuropsychological disorders. To address this, a promising approach involves employing data-driven techniques like principal component analyses to unveil underlying, latent, patterns (Lambon Ralph et al., 2003, Butler et al., 2014). Notably, a prior study delved into apraxic deficits using principal component analyses, focusing on a set of eight apraxia tasks (Rounis et al., 2021). This analysis unveiled three core components that contributes to the disorder, encompassing aspects like posture selection, semantic control and sequencing deficits. Further efforts to identify white matter disconnections based on these components, particularly if validated with larger sample sizes hold the potential to provide even more comprehensive insight into the mechanisms underlying limb apraxias, in the future.
In conclusion, our study addresses a long-standing challenge of differentiating between conceptual and production related deficits in in apraxia, shedding light onto its neural pathways and challenging historical assumptions about hemispheric dominance in apraxia. Future research in larger sample using advanced neuroimaging hold the promise to further clarify the intricate mechanisms underlying apraxia deficits.