Anika Löwe
Fellow, IMPRS COMP2PSYCH Berlin
Max Planck Institute for Human Development, MPRG Neurocode
Anika Löwe is a third-year PhD student working with Nico Schuck at the MPI for Human Development and also a fellow of the COMP2PSYCH research school. Her background lies in Psychology (BSc.) and Cognitive Neuroscience (MSc.).
Her core interests lie in the neural and computational basis of learning dynamics and representation formations (as well as changes thereof) in biological as well as artificial agents.
In her first PhD project she investigated how and when insight moments arise in humans and neural networks. Half of human volunteers performing a two-choice task, where previously learned input representations can be relearned to improve efficiency, showed insight-like learning about the newly relevant features. A simple linear neural network with regularised gate modulation on the input nodes showed abrupt learning dynamics resembling insight-like behaviour. These results suggest that insight phenomena can arise from regularised gradual learning mechanisms and shed light on representation formation in intelligent agents more generally.
Anika’s second project extends the experimental setup described above by investigating a potentially beneficial effect of sleep on rule discovery and representational changes. A third project will be concerned with the coordination between hippocampal replay and prediction error as well as value-related neural signals in the human brain. The goal is to test quantitative predictions regarding the content and order of replay at reward and planning stages based on theories from the animal literature and the computational replay account based on Mattar & Daw’s (2018) normative framework.