Processing and Plasticity in Rodent Somatosensory Cortex Dan Feldman Dept. of Molecular & Cell Biology Helen Wills Neuroscience Institute UC Berkeley
Primary sensory neocortex Secondary and higher areas Integrative, abstract features; identification Primary cortex (A1, V1, S1) Local, low-level stimulus features Brainstem & thalamus Initial processing, gating Sensory neurons Detection and filtering Specialized circuits to process sensory stimuli Neural activity represents local stimulus features Sensory representations change with behavioral state and experience Questions How are natural sensory inputs represented, particularly in awake, behaving animals? What computations are implemented by cortical synapses and circuits? What are the synaptic mechanisms for sensory map plasticity?
The rodent whisker system
Active whisker sensation Whisker shaft Extrinsic Muscle Intrinsic Muscle slowed ~10x Berg et al., 2003 Knutsen.. Ahissar, J. Neurophysiol., 2005
Moveable sensors Active tactile sensation Form and texture
Active whisker sensation Where? Mehta et al., 2007 What? Carvell & Simons,1990 Texture A Texture B
Active whisker sensation Takeshi Morita
Active whisker sensation Texture discrimination
Sensory coding during active whisker sensation Surface Whisker kinetics Neural response in S1 cortex Physical Transformation Neural Transformation
Models for Texture Coding Whisker resonance model Place code for texture on the face
Sensory coding during active whisker sensation Image whisker vibrations Record S1 spikes 1.5 sec ~ 1 sec Jason Wolfe Shantanu Jadhav Nose poke Whisking on surface Leave Nose poke Drink
Sensory coding during active whisker sensation
Testing the whisker resonance model Wolfe et al., PLoS Biology 2008 Map of whisker resonance
Testing the whisker resonance model D3 whisker on P150 sandpaper
Testing the whisker resonance model Wolfe et al., PLoS Biology 2008 Whiskers resonate, but resonance does not encode texture
Whisker slips as a cue for surface features High-velocity, high-acceleration slip-stick events Whisker moving across rough sandpaper stick Protraction Time slip & ring Ring Slip Wolfe et al., PLoS Biology 2008
Whisker slips as a cue for surface features Slips 0.5 mm/ ms2 * * * * * *
Models for Texture Coding Whisker resonance model Place code for texture on the face Slip coding model (kinetic signature model) Statistics of slips encode texture
S1 coding during active whisker sensation Chronic tetrode recording n = 90 neurons, L4 and L5, 3 rats
Mean firing rate is low L4 L5 Median 6 Hz Jadhav et al., Nat. Neuro 2009
Whisker slips drive sparse, temporally precise spikes 14 ms jitter (temporal precision)
Whisker slips drive sparse, temporally precise spikes Slip responses are usually 1-2 spikes. Net P(spike) over background = 0.11
Whisker slips drive sparse, temporally precise spikes Low response probability is not due to tuning for stimulus features
Slips contribute to firing rate elevation on surfaces Mean 2.2-fold increase in firing rate on surfaces
Decoding a sparse population signal in S1 ROC analysis shows that slips are accurately encoded by synchronous firing (20 ms scale) in small neuronal populations (100 neurons, 97% accuracy) Thus, precise timing of responses enables decoding of the sparse population signal. This constitutes a transient synchrony code for slips. 20 ms
Decoding a sparse population signal in S1 Confirming this idea, slips transiently increase firing correlation for neuron pairs in vivo. 26
Slip-driven firing synchrony provides a useful code for surface roughness Air no large slips few large slips more large slips Firing rate Firing correlations (20 ms window) Firing correlations (100 ms window) Very different textures
Slip-driven firing synchrony provides a useful code for surface roughness Air no large slips few large slips more large slips Firing rate Firing correlations (20 ms window) Firing correlations (100 ms window) Four similar sandpapers (P150, 400, 800, 1200)
Whisker sensory coding in S1 Natural whisker stimuli are time series of discrete events (slips, contacts) Slips (and contacts) are encoded by sparse, precisely timed spikes. Sparseness is probabilistic, not based on narrow stimulus tuning. Precise spike timing allows efficient decoding from transient firing correlations.
People Lab Members David House Renna Stevens Joe Goldbeck Cara Allen Kevin Bender Vanessa Bender Jason Wolfe DJ Brasier Shantanu Jadhav Lu Li Toshio Miyashita Ray Shao Tansu Celikel Elisabeth Foeller Sowmya Venkataramani Patrick Drew Sharri Zamore Takeshi Morita Collaborators Beat Lutz (Munich) Daniel Shulz (CNRS) Ken Mackie (UW) David Kleinfeld (UCSD) Support NINDS NSF CAREER Award NSF Temporal Dynamics of Learning Center McKnight Endowment Klingenstein Foundation March of Dimes Alfred P. Sloan Foundation *results presented today
Independent firing of S1 neurons
Utility of a synchrony code for slips: coding of surface properties Slip-driven firing synchrony provides a useful code for surface roughness Air no large slips few large slips more large slips Firing rate Firing correlations (20 ms window) Firing correlations (100 ms window)
Whisking on surfaces modestly increases firing rate
Active whisker sensation Takeshi Morita
Sparse population coding of slip events
Sparse population coding of slip events 20 ms
Primary sensory cortex Secondary and higher areas Integrative, abstract features; identification Contains synapses and circuits that process sensory stimuli Learns in response to altered sensory experience and training Primary cortex (A1, V1, S1) Local, low-level stimulus features Brainstem & thalamus Initial processing, gating Sensory neurons Detection and filtering