Even though most everyday decisions are among more than two alternative, the optimal policy for such decisions was unknown. Together with Satohiro Tajima, Nisheet Patel, and Alexandre Pouget we derived this optimal policy and showed that it is significantly more complex than a simple scaling-up of decisions among two alternatives. However, a simple network implementation with features inhibitory cross-talk among the accumulators that collect evidence for the different alternatives turns out to feature close-to-optimal performance. This network has several cortex-like features, like an...
Anna was trained as s physicist in Germany, where she received an M.Sc. from the University of Heidelberg. She completed her Ph.D. in computational neuroscience at the University of Zurich, where she worked with Prof. Dr. Jean-Pascal Pfister on non-linear filtering in general, its... Read more about Anna Kutschireiter, Ph.D., joins the lab as postdoc
Qiao holds an Ingénieur Polytechnicien Degree from the Ecole Polytechnique, France, and a Master of Finance from MIT. He completed his Ph.D. in computer science at INRIA (French Institue for Research in Computer Science and Automation), where he worked with Prof. Nicholas Ayache using deep learning... Read more about Qiao Zheng, Ph.D., joins the lab as postdoc
Location: Harvard Medical School, Boston, MA Start: April 2018 or after
We are looking for a highly creative and motivated postdoctoral fellow to work in the field of computational and systems neuroscience in the group of Jan Drugowitsch - Department of Neurobiology at Harvard Medical School (http://neuro.hms.harvard.edu/people/faculty/jan-drugowitsch). The group works on questions of decision-making and navigation under uncertainty, ranging from the level of neural populations to that of behavior. In the short term, the postdoctoral fellow is expected to design and analyze...
Our new pre-print investivates how to learn to make optimal decisions with uncertain evidence that needs to be accumulated over time and is encoded in neural populations. Diffusion models, which have been the go-to models for such decisions, commonly assume the evidence to be one-dimensional rather than to be distributed in a population, and the format of this evidence is usually known. We instead assume a distributed representation of this evidence whose format needs to be learned, and derive near-optimal learning rules for this, more natural, case. These learning rules turn out to include... Read more about New pre-print on how to make optimal decisions with confidence
Congratulations to Emma Krause, Harvard Program in Neuroscience graduate student and member of the Drugowitsch lab, for having been awarded the National Defense Science & Engineering Graduate (NDSEG) fellowship! Provided by the Department of Defense, this highly competitive fellowship will support her next years of graduate studies.
In a new paper that just appeared in Neuron, Valentin Wyart, Anne-Dominique Devauchelle, Etienne Koechlin and Jan Drugowitsch identify mental probabilistic inference as the main source of behavioral variability. This is in contrast to previous reports that have focused on sensory noise or stochastic action selection. Having identified inference as the main variability source allowed us to leverage this variability to gain further insight into the structure of the mechanisms implementing this inference.
Jan Drugowitsch (Department of Neurobiology, Harvard) and Sam Gershman (Department of Psychology, Harvard) are seeking a postdoctoral fellow to work on a project combining computational modeling, psychophysics, and clinical studies. The project focuses on the neural basis of visual structure discovery, using motion perception as a model system. Our goal is to understand how neural circuits represent and reason about complex combinatorial structures, and how these neural circuits break down in autism.
The Drugowitsch and Gershman labs are located in the Longwood medical area...