What do you say? An investigation of how lesions cause language deficits in post-stroke

Lead Research Organisation: University College London
Department Name: Institute of Neurology


A stroke can happen at any time to anyone, including new-borns and children. Stroke occurs when blood supply to an area in the brain is compromised, causing brain cells to die. Loss of brain cells can lead to deficits in functions for which an area specialised. For example, approximately a third of stroke survivors have deficits in speech comprehension and/or production, a condition referred to as aphasia. Aphasia is, thought to be, caused by lesions in dorsal and/or ventral language pathways post-stroke (Weiller, et al. 2011; Hickok and Poeppel, 2007; Hickok and Poeppel, 2000; Eggert, 1997). Aphasia has devastating consequences for the affected individual, due to inability to communicate and engage with loved ones and the wider community, alongside considerable societal cost.

We know that the ability of a patient to recover their language capabilities post-stroke depends on the total proportion of damage to the brain, whether areas that support language have been damaged and the intensity of the initial symptoms (Price, et al. 2010). Additionally, using data-driven methods, research shows that post-stroke recovery is much more systematic and predictable than was previously thought (Hope, et al. 2018; Hope, et al. 2017; Seghier, et al. 2014; Hope et al. 2013; Seghier, et al. 2012; Saur, et al. 2010; Blasi, et al. 2002; Musso, et al. 1999). However, without a mechanistic, process-level account of how language skills are implemented in a healthy brain, we cannot understand what has gone wrong (and therefore, consider how best to resolve it) when those skills are impaired by stroke. This understanding is crucial for giving patients an accurate indication of recovery.
Thus, the project aims to investigate the process by which lesion damage causes aphasic symptoms post-stroke using computational modelling. Computational modelling is an appropriate framework to study this paradigm since it will allow us to model both normal behaviour and then lesion the model to forge a better understanding if Bayesian. This will explain how and why lesion damage can cause different types and severity of aphasia.


10 25 50
publication icon
Da Costa L (2020) Active inference on discrete state-spaces: A synthesis. in Journal of mathematical psychology

publication icon
Friston K (2020) Generative models, linguistic communication and active inference in Neuroscience & Biobehavioral Reviews

publication icon
Friston KJ (2021) Active listening. in Hearing research

publication icon
Parr T (2020) Modules or Mean-Fields? in Entropy (Basel, Switzerland)

publication icon
Parr T (2021) Generative Models for Active Vision. in Frontiers in neurorobotics

publication icon
Sajid N (2021) Cancer Niches and Their Kikuchi Free Energy. in Entropy (Basel, Switzerland)

publication icon
Sajid N (2021) Active Inference: Demystified and Compared. in Neural computation

publication icon
Sajid N (2020) Degeneracy and Redundancy in Active Inference. in Cerebral cortex (New York, N.Y. : 1991)

Studentship Projects

Project Reference Relationship Related To Start End Student Name
MR/S502522/1 30/09/2018 12/07/2023
2088828 Studentship MR/S502522/1 30/09/2018 12/07/2023