Neurocomputation: Theory and brain-grounded neurocomputational modelling
We use neuroanatomically and neurophysiologically inspired network models to address questions about language and cognition. These networks mimic the structure and connectivity of the language areas and sensorimotor systems of the human cortex to model language learning and breakdown due to lesions. The networks help explain established knowledge about specific neurocognitive and –linguistic brain activations and offer new experimental predictions to be addressed in neuroimaging experiments or patient studies.Projects
New brain-constrained neural models to answer long-standing questions in language and cognitive sciences
Compared to our closest living relatives, who typically use fewer than 100 signs, humans can build huge vocabularies of tens of hundreds of thousands of words, combine these in multiple ways and use them for abundant social purposes. The EU-funded project MatCo will find out why, by spelling out the material basis of human-specific higher cognitive functions. Novel insights from human neurobiology will be used to constrain mathematically exact computational models of neural mechanisms so as to find new answers to long-standing questions in cognitive science, linguistics and philosophy. Relevant questions range from symbol acquisition, concept formation to action combination and semantic learning. One focus addresses how, i.e. by which material mechanisms, semantic meaning is implemented for gestures and words and, more specifically, for referential, categorical and abstract terms. To identify human cognitive capacities, MatCo will develop models replicating structural differences between human and non-human primate brains, so as to find out which exact neurobiological changes enabled and gave rise to specific aspects of human cognition. Performance of biologically constrained networks will be validated against human behavior and brain activation.
This project is funded by the European Research Council (ERC).
Why are humans capable of acquiring a large number of symbols, and what are the neural material features critical for this extraordinary ability? Here we will explore the material basis of symbols in the human brain by means of brain-inspired neural models constrained by cortical neuroanatomy, function and obeying well-established neuroscience principles (see figure). The subproject »exploring the material basis of symbols« aims to explain the cognitive activity (language, symbols, thought) by the structural features of the underlying neural matter present in the human brain, with the ultimate goal to generate brain simulator mimicking essential symbol processing functions with its application to specific healthy and patient brains.
Over and above neurocomputational simulations, neuroimaging and neuropsychological studies (using EEG, TMS, and/or fMRI) will be carried out to determine the neurobiological correlates of linguistic-conceptual and visual-gestalt-like symbolic processes in the human brain. Results will be compared with predictions generated by the neurocomputational model.
More information here.
This project is funded by DFG.
Previous research in behavioural and cognitive neuroscience demonstrates a close link between the brain systems for language, action and perception. These parallel developments in behavioural and computational neuroscience, as well as in cognitive robotics and in neuromorphic engineering constitute a timely opportunity to synergistically integrate the interdisciplinary methods and approaches from these fields with the aim of furthering the scientific and technological progress in language pro-cessing in natural and artificial cognitive systems. The project proposes the interdisciplinary integration of new brain imaging experiments, of neuro-anatomical computational and neuromorphic studies, and of humanoid robotics experiments in order to characterise the brain mechanisms supporting language learning in an embodied and pragmatic (situated) context, and to design and test novel brain-inspired neural technologies for action and language learning experiments with interactive intelligent systems such as humanoid robots.
This project was funded by Engineering and Physical Sciences Research Council (EPSRC), UK and Biotechnology and Biological Sciences Research Council (BBSRC)Publications
Tomasello, R., Garagnani, M., Wennekers, T., Pulvermüller, F., 2017. Brain connections of words, perceptions and actions: A neurobiological model of spatio-temporal semantic activation in the human cortex. Neuropsychologia 98, 111–129, doi: 10.1016/j.neuropsychologia.2016.07.004
Garagnani, M., Lucchese, G., Tomasello, R., Wennekers, T., Pulvermüller, F., 2017. A Spiking Neurocomputational Model of High-Frequency Oscillatory Brain Responses to Words and Pseudowords. Front. Comput. Neurosci. 10, doi: 10.3389/fncom.2016.00145
- Schomers, MR, Garagnani M, Pulvermüller F (2017) Neurocomputational consequences of evolutionary connectivity changes in perisylvian language cortex. Journal of Neuroscience 37(11), 3045-3055, doi: 10.1523/jneurosci.2693-16.2017
- Garagnani, M & Pulvermüller, F. (2016). Conceptual grounding of language in action and perception:a neurocomputational model of the emergence of category specificity and semantic hubs. European Journal of Neuroscience, in press. doi: 10.1111/ejn.13145
- Pulvermüller, F, Garagnani, M, and Wennekers, T (2014). Thinking in circuits: Towards neurobiological explanation in cognitive neuroscience. Biological Cybernetics, June 2014.
- Garagnani, M, Wennekers, T, and Pulvermüller, F (2008) A neuroanatomically-grounded Hebbian learning model of attention-language interactions in the human brain. European Journal of Neuroscience 27, 492-513.