Computational Learning and Memory Group Welcome Trust Investigator Award

home | members | publications | events | teaching | vacancies | directions

Dendritic computation in neural circuits
a Cosyne 2013 workshop

This workshop will concentrate on the role that dendritic processing plays in circuit computations. While there is now plenty of evidence, both experimental and computational, that dendrites of single neurons are capable of local nonlinear processing, the role this type of processing plays in behaviorally relevant computations at the network level has only recently begun to be elucidated. We aim to bring together leading experimentalists and theorists in the field to discuss and consolidate current ideas, focusing in particular on increased interaction between theory and experiment, as well as to discuss important future directions of research. We believe this workshop will provide a valuable forum for integrating research across multiple levels - from subcellular to behavioral - for both theorists and experimentalists.

Date and Location

Tuesday, March 5th, 2013
Information about the exact location will apear here.
for information about room bookings see the related conference website


morning session
8.20 welcome notes, introduction
8.30 Bartlett Mel, University of Southern California


9.00 Jeff Magee, Janelia Farm Research Campus


9.30 coffee break
10.00 Michiel Remme, Humboldt University Berlin


10.30 Albert Lee, Janelia Farm Research Campus


afternoon session
16.30 Tiago Branco, MRC Laboratory of Molecular Biology, Cambridge, UK


17.00 Walter Senn, University of Bern

Neurons displaying dendritic spikes can be compared to a 2-layer neural network. In terms of learning, this analogy offers a biological implementation of error-backpropagation. In fact, only if the presynaptic, dendritic and somatic spikes jointly modulate plasticity can the representational power of dendritic processing be exploited. A learning rule which maximizes the expected reward suggests how the timing among these spikes and the postsynaptic voltage optimally determines synaptic plasticity.

17.30 coffee break
18.00 Balázs Ujfalussy, University of Cambridge

Nonlinear dendritic phenomena, such as dendritic spikes, are fundamental to the integration of synaptic inputs in cortical neurons and depend on the detailed spatio-temporal pattern of incoming action potentials. This dependence has traditionally been characterized in experiments using a few but well-controlled inputs. However, it is unclear what computational roles nonlinear dendritic phenomena play in the face of ongoing cortical activity when each cortical neuron receives synaptic inputs from thousands of other neurons. Here, we derive optimal dendritic nonlinearities for a neuron to integrate its many presynaptic inputs. This theory predicts a close match between the long-run statistical patterns and correlations characterizing the population activity of a set of presynaptic neurons and the dendritic nonlinearity in the postsynaptic neuron. We tested and confirmed these predictions in previously published and novel experimental data. Our analysis offers a unique link between cellular- and systems-level properties of neural circuits.

18.30 Spencer Smith, UNC,



Walter Senn: Why should one dendritic branch modulate processing in another branch? -- A purely feedforward dendritic network is already an universal function approximator!
Questions proposed by panelists/speakers

19.30 end of workshop

Talks are 20 mins long + 10 mins discussion.

Panel discussion:

For the panel discussion, we would also like to have one question in advance from each speaker. These will be made public here. During the discussion, we will go through the questions one-by-one, with each question moderated by the person who proposed it. Of course, people can change questions in light of the workshop, but we would like to have a set of questions in advance to get people thinking and to fall back on.


DJ Strouse (primary contact), Balázs B Ujfalussy, Tiago Branco, Máté Lengyel
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback