Our first COE seminar will take place at 3pm ET on Oct. 31st. The speaker is Prof. Ioannis Delis from University of Leeds. The title and abstract of his talk is listed below. Please join the seminar using Zoom link.
Neurocomputational characterisation of multi-sensory decision-making in the human brain
Perceptual decisions rely on the integration of evidence from the environment, which involves the combination of stimuli from different senses. Multisensory information has been shown to improve decision-making performance; however, the neural origins of this benefit remain largely unknown.
In this talk, I will address this question by introducing a neurally-informed modelling approach to the characterization of multisensory decision formation. Specifically, I will present a set of experiments where brain activity was recorded using electroencephalography (EEG) while human subjects performed different binary-choice reaction-time tasks involving the processing of sensory evidence from different modalities.
To identify components of brain activity underlying performance of these tasks, we implement single-trial multivariate discrimination analyses. This approach disentangles brain activity into components encoding the participants’ choice and quantifies any multisensory benefits at the neural level, i.e. neural gains. Then, to characterise the functional roles of these components in decision-making behaviour, we use them to inform a hierarchical drift diffusion model, a model of decision-making behaviour that decomposes task performance into its constituent processes, such as sensory encoding, evidence accumulation and motor response. Thus, we can associate the identified neural gains with the latent processes they underpin and characterize their mechanistic contribution to decision-making performance. Finally, we employ a novel information-theoretic methodology to quantify multisensory interactions in the human brain and assess their contribution to behavioural benefits.
Ultimately, we suggest that the proposed computational framework opens a window onto the neural representations of the processes implicated in multisensory decision formation.
