Bernstein Seminars May-June 2022

May 1 - June 30, 2022

Freiburg (Zoom) (Germany)


June 28th from 17h15 to 18h

Stefan Leutgeb: University of California, San Diego: TBA



June 14th from 17h15 to 18h

Janelle Pakan: from sensation to action: experience-dependent cortical plasticity during task engagement

We are constantly learning from our experiences as we engage with the world around us. During this active learning our senses work together to build an external reality, which is also influenced by our ongoing behaviours. This multimodal engagement across sensory and motor circuits results in experience-dependent plasticity and may lead to more efficient learning – when we are driving a car and are engaged in the task of navigating, we very often remember the route much better than as a passive passenger. A major challenge in modern neuroscience is to elucidate how sensory and behavioural systems integrate and influence each other across distributed brain networks. How does contextual sensory stimulation transform to behavioural output and how does our behavioural output subsequently feedback to affect fundamental sensory processing?

A vital step in understanding the functional principles of these neural circuits is to directly observe the activity of local circuit elements with high temporal and spatial resolution during sensation and action. To do this, we use advanced in vivo two-photon microscopy in combination with virtual environments to examine principles of cortical plasticity in sensory and association brain regions across different levels of task engagement in behaving mice. The overarching aim is to disentangle how sensory and motor systems function together in the dynamic complexity of the ‘real-world’ and how neuronal representations adapt across learning.



May 24th from 17h15 to 18h

Alessio Attardo: Aberrant synaptic dynamics precede disorganisation of activity patterns in the hippocampus upon repeated stress exposure

Stress exposure affects the structure and function of hippocampal CA1 leading to disruption of episodic memory. Drawing a connection between changes in structural connectivity and activity patterns upon repeated stress has been a major issue to this point. Even in rodents, the vast majority of previous experimental work focused either on structure or on function, resulting in separate streams of findings using different stress paradigms and models. To solve this issue, we took an integrative approach and combined wide field head-mounted miniaturized microscopes, longitudinal deep-brain two-photon optical imaging and a behavioral task to study the relationship between deficiencies of structural activity and connectivity in hippocampal CA1 pyramidal neurons and the ability to learn in mice undergoing repeated stress.

We found that repeated stress exposure led to immediate and sustained increase in neuronal activity and to a delayed disorganization of the temporal structure of this activity and loss of spatial coding content of dorsal hippocampal CA1 pyramidal neurons. In addition, thanks to the longitudinal nature of our in vivo two-photon imaging, we could track - for the first time - structural excitatory connectivity of dorsal hippocampal CA1 pyramidal neurons over one week of repeated stress. This enabled us to detect a significant decrease in excitatory connectivity occurring in two separate steps: an immediate but transient decrease in spinogenesis followed by a delayed increase in spine loss. Interestingly, spine loss in CA1 pyramidal neurons only became apparent after several days of hyperactivity, and disorganization of the temporal structure of activity and loss of spatial coding content was evident only after significant spine loss. Thus, suggesting that the effects of stress exposure on hippocampal temporal and spatial coding are mediated by loss of synaptic connectivity.



May 10th from 17h15 to 18h

Andreas Thum: from structure to function: what we can learn from the connectome of the drosophila larva

The Drosophila larva is a relatively simple, 10 000-neuron study case for learning and memory with enticing analytical power, combining genetic tractability, the availability of robust behavioral assays, the opportunity for single-cell transgenic manipulation, and an emerging synaptic connectome of its complete central nervous system.

Indeed, although the insect mushroom body is a much-studied memory network, the connectome revealed that more than half of the classes of connection within the mushroom body had escaped attention. The connectome also revealed circuitry that integrates, both within and across brain hemispheres, higher-order sensory input, intersecting valence signals, and output neurons that instruct behavior.

Further, it was found that activating individual dopaminergic mushroom body input neurons can have a rewarding or a punishing effect on olfactory stimuli associated with it, depending on the relative timing of this activation, and that larvae form molecularly dissociable short-term, long-term, and amnesia-resistant memories. Together, the larval mushroom body is a suitable study case to achieve a nuanced account of molecular function in a behaviorally meaningful memory network.



May 3rd from 17h15 to 18h

Joschka Boedecker: Inverse Q-learning as a tool to investigate behavior and its neural correlates

Inverse Reinforcement Learning offers the promise to recover the intention underlying an observed behavior, i.e. a reward signal that would explain the behavior of an agent, assuming it tries to maximize it in the long term. This has important applications for imitation learning in robotics, as well as behavior understanding in fields such as biology and neuroscience. Most approaches that allow to extract such a signal from observed data, however, need to solve a full reinforcement learning problem to convergence multiple times in an inner loop, making them computationally expensive and unsuitable for many application scenarios.

In this talk, I will present our work on novel inverse RL algorithms which exploit the structure of Q-Learning and result in algorithms that speed up learning of the reward signal by several orders of magnitude, while also providing more accurate value estimates than prior work. I will also illustrate how this enables the analysis of behavior and the activity of brain regions of different animals as specific examples.




Zoom Meeting. You can contact Fiona Siegfried for meeting ID and password.

Contact: This email address is being protected from spambots. You need JavaScript enabled to view it.