Data Assimilation Meetings at Reading
Date | Meeting type | Speakers |
19 February 2014 | Internal speaker |
Ingo Bojak (University of Reading) The cake is a lie - Limits to biological realism in (neural population) models. Computational modelling of brain activity is increasingly expected to match experimental data quantitatively, not just qualitatively. Yet biology is messy and complicated, hence realistic models typically are nonlinear and have a multitude of weakly constrained parameters. Meanwhile advances in non-invasive neuroimaging, like simultaneous EEG and fMRI BOLD measurements, have renewed the interest in neural population models (NPMs), since these models provide manageable descriptions at the experimentally available spatial resolution. After an introduction to NPMs I will consider some typical problems with increasing their realism, including implementing proper signal expression and connectivity. I will then present results from ongoing research concerning the modelling of general anaesthetic agents, and finish with some conceptual ideas why biologically plausible model dynamics tends to be "brittle" and distributed over high-dimensional parameter spaces. This could lead to novel computational approaches to model fitting, analysis of brain activity and ultimately provide the kind of robust modelling that clinical applications require. |