Peter Thestrup Waade
Switzerland
Postdoctoral Researcher
ETH Zürich
Neuroscience/Cognitive Science, Machine Learning/Technical AI, Interdisciplinary
Switzerland Postdoctoral Researcher ETH Zürich Neuroscience/Cognitive Science, Machine Learning/Technical AI, Interdisciplinary
Project
Do embodied neural cultures learn predictive world models through active inference?
Project Overview
Embodied neural cultures, such as the Cortical Labs CL1 system, have been shown to be capable of adaptive behaviour and sensitive to sensory stimulation. It is, however, unclear to which degree this implies that these neural cultures can be considered comparable to other, conscious and ethically relevant organisms such as humans, other animals, and plants. Specifically, it is unclear whether the adaptive behaviour of the CL1 system is due to simple associative adaptation, or whether the cultures can be considered to embody a more complex model of their environment – a world model. In order to assess whether this is the case, and to better position the neural cultures in the larger space of cognitive and adaptive systems, we use methodologies from computational cognitive neuroscience to systematically investigate and model (in a predictive processing framework) neural cultures’ ability to perform behavioural tasks that require cognitive processes beyond associative learning. We begin with a simple, purely associative forced choice learning task, and progressively increase the complexity of the task to require contextual (re-)learning, short-term memory, inference about unobservable variables, and predictive planning into the future. Finally, we situate these cognitive capacities within the broader debate around world models and consciousness.
Mentor
About the Scholar
Peter Thestrup Waade is a Danish/Norwegian computational cognitive scientist working in computational psychiatry. Specialising in cognitive modelling - often but not exclusively in Bayesian mind and predictive processing paradigms - he develops Julia software for this methodology, and applies it to investigate phenomena such as hallucinations, fatigue, and the effect of therapeutic interventions. As an interdisciplinary-hearted researcher, he also studies embodies social interaction - such as partner dancing - and applies phenomenological interview methods alongside third-person experimental methods. He speaks Chinese fluently, translates poetry, is a dancer and martial artist, and in general believes in relationality, clarity and curiosity as the foundations of both his work and personal life.
Why AISS?
"Joining the AISS program is an incredibly opportunity to work with a mentor and on a project I have been eager to connect with for a long time, and to connect with a group of competent and excited researchers from different fields. It is not so often that interdisciplinary collaboration is supported this clearly, nor that early-career work is supported both structurally and financially. And what a wonderful thing, that is also happens in connection with this year's CIFAR winter school - and is hosted by an organisation that I respect and admire."
Links
