The talk was conducted on 11 November 2020. Dr. Samar Husain from IIT Delhi, covered the following in his talk:

Abstract: “Processing of Subject-Object-Verb (SOV) languages has been argued to involve robust clause-final verbal prediction. Robust verbal prediction and its maintenance has been shown to lead to facilitation during sentence comprehension and has been attributed to the parser’s adaptability to certain typological features (e.g., word order) in such languages. In this talk I will argue that the parser’s adaptability for robust prediction in SOV languages is limited. To this effect, I will provide converging evidence from corpus-based studies, behavioural experiments as well as computation modelling. In particular, I will show that as the nature of preverbal linguistic context becomes complex, comprehension suffers in these languages. This suggests the overarching role of working-memory constraints during sentence comprehension.”

The talk was conducted on 23 October 2020. Dr. Sengupta is a Research Affiliate at Cambridge University and an MRI Research Fellow at the Royal Holloway, University of London. His research is centered on neuroimaging, specifically Functional Magnetic Resonance Imaging (fMRI). His main interests are in ultra-high field human fMRI and the application of machine learning and other computational modelling in understanding how visual and tactile information are represented in different parts of the brain.

The talk covered basics of functional MRI analysis and applications of machine learning for decoding the mind.

The talk was conducted on 15th October, 2020 where Dr. Britt Anderson from University of Waterloo discusses the following:

Abstract: “Informally, mental models are your conception of the world’s biases and tendencies; they are that which you rely on for choosing options and generating actions. But things change. How do you decide that a change has occurred, and when you do, how do you plot a path to the new model? Is there something general about this process that spans domains? I will address these questions through the lens of data where participants’ abilities to build simple mental models allow them to exploit biases to improve performance. Tasks range from perceptual updating of small incremental changes in visual images to playing rock, paper, scissors against virtual opponents. By comparing performance across tasks and populations (controls, children, and patients with focal brain injury) we can begin to address whether “mental model” is more than a useful metaphor, and via fMRI explore the constellation of brain structures temporally correlated to updating events. Lastly, I will suggest that identifying a mental model as a point in a conceptual space leads one to consider some less widely used mathematical areas for cognitive and neural modelling.”