CompNeuro_Synapses
Dr. Cian O’Donnell
cian.odonnell@bristol.ac.uk
Synapses
COMS30127: Computational Neuroscience
mailto:cian.odonnell@bristol.ac.uk
Dr. [your name here]?
• Would you like to do a PhD in computational neuroscience?
• If so there are many good options in the UK and abroad. It’s more or less a
buyers market (you’re buying).
• Here’s 3 nice options:
1. EPSRC DTP with one of Bristol’s Computational Neuroscience Unit: Conor
Houghton, Nathan Lepora, Naoki Masuda, Rosalyn Moran, Cian O’Donnell.
Check out our research webpages and contact us directly if interested.
2. Bristol’s Neural Dynamics PhD program (joint experimental/computational).
Webpage: http://www.bristol.ac.uk/neural-dynamics/
3. Rafal Bogacz at Oxford is looking for a PhD student to start Oct 2017.
“The topic of research could range from developing a formal theory of
information processing in the brain to applying machine learning to the
data recorded from the brains of patients.”
Rafal’s page: https://www.ndcn.ox.ac.uk/team/rafal-bogacz
PhD advert: https://www.findaphd.com/search/ProjectDetails.aspx?
PJID=83868&LID=1239
http://www.bristol.ac.uk/neural-dynamics/
https://www.ndcn.ox.ac.uk/team/rafal-bogacz
https://www.findaphd.com/search/ProjectDetails.aspx?PJID=83868&LID=1239
https://www.findaphd.com/search/ProjectDetails.aspx?PJID=83868&LID=1239
https://www.findaphd.com/search/ProjectDetails.aspx?PJID=83868&LID=1239
What we will cover today
• What is a synapse?
• How do synapses work?
• How can we computationally model synapses?
What is a synapse?
• Synapses are the connections between neurons.
• They convert the action potential from one neuron’s axon into a ‘post-
synaptic-potential’ in the dendrite of another neuron.
• Although both a synapse’s input and output signals are electrical, the
most common type of synapse converts the signal into chemical form at
an intermediate stage.
• There are also purely electrical synapses (called ‘gap junctions’) but in
this course we will focus on chemical synapses.
• From a functional point of view, synapses are interesting for two reasons:
1. they are nonlinear, so can perform computations.
2. they are plastic, so can store information (memories).
What is a synapse?
Presynaptic neuron Postsynaptic neuron
Synapse
Action potential
Markram et al, J Physiol (1997)
Zooming in on synapses
Animal (~10 cm)
Brain (~1 cm)
Brain region (~1 mm) Neuron (~100 μm)
Synapses (~1 μm)
What is a synapse?
Blender demo & Youtube video (https://www.youtube.com/watch?v=FZT6c0V8fW4&t)
Electron microscopy data from Graham Knott (EPFL)
Dendrite and axon from mouse somatosensory cortex.
How do synapses work?
Axon
(presynaptic action potential)
Dendrite
(postsynaptic potential)
Chemical
signalling
Image from Wikipedia (modified by C Houghton)
Synapses are probabilistic
• When an action potential arrives at a synapse, it may or may
not lead to release of neurotransmitter.
• The ‘release probability’ p is often quantified experimentally,
0