Spatial Organization of Neuronal Population Responses in

1
Spatial Organization of Neuronal Population
Responses in Layer 2/3 of Rat Barrel Cortex
Jason N. D. Kerr, Christiaan P. J. de Kock, David
S. Greenberg, Randy M. Bruno, Bert Sakmann, and
Fritjof Helmchen
Take Home Points
1. Sparse spiking, no precise patterns.
2. Spatially organized probalistic spiking
patterns.
3. Position ?Not related? direction sensitivity
(unlike afferents)
4. Population coding Each feature ? many
neurons. Each neuron ? several features.
May facilitate integration of multiple whiskers.
2
Mutual Information

How much information two things share.
. . . . ?
A measure of how knowledge about one thing
reduces your uncertainty about another thing.
a more sophisticated correlation.
3
Rationale
Large single cell variability.
Sparse and short-lived patterns
BUT could have an unambiguous pattern, but
requires many neurons. (Think sample size)
Methodologically Development of spatial maps
of neural activity. normally impossible with
extracellular recording.
4
Rat Barrel Cortex

• Rodent somatosensory cortex.
• Single whisker ? discrete structures (whisker
  barrels) separated by regions called septa.
• Same geometric order as whiskers.
• Model system for cortical columns.
• Organized into layers. Layer IV (L4) individual
  neurons -- consistent trial-to-trial, strong
  directional tuning.
• However, L2/L3 layer does not.

1 Woolsey and Van der Loos, 1970
5
Methods
Calcium Indictors
Location of all cells Single-cell and
single-spike resolution.
Two-Photon Microscopy
Skull exposed, optical imaging while stimulating
whisker.
Patch-clamp recordings visually targeted.
Random whisker deflected for 500ms. Interstim
3-6 sec
Cortex sectioned and area of WB determined.
.. Then a significant variety of analysis.
6
Layout Identification
Deflection ? transients similar to spontaneous
ones.
Electrical and microscopy produced similar
results.
7
Question Spatial organization?
Stimulated whisker
Septa near whisker.
Conclusion (1) Depends on distance from
BCC (2) Highly variable
Nearby whiskers
(3) For both onset offset
(4) Highly significant topology. (5) Offset
onset, but smaller. (6) Spontaneous all similar.
8
Tuning amount varied by individual.
Little individual direction tuning.
Tuning corrected for Spiking rate.
No spatial organization for directional tuning.
9
Question Sparse/Dense responses?
Conclusions
(1) Varies trial-to-trial
(2) Varies greatly between cells.
(3) Onset-Offset active cells may vary.
10
Fraction active by location Stimulus
different. Spontaneous similar.
Assuming independence does not match the data. ?
Correlated Activity
Subsets activated not consistent trial-to-trial.
11
Correlations present in spontaneous, but
increased with stimulus.
Distance between neurons ? means
little. Distance from BCC ? significant
meaning. For both spontaneous and stimulus.
During sensory stimulation, neighboring neurons
may be bound together by common inputs.
Question Is effect of distance to BCC on
correlation a result of pairs in the BCC being
packed closely?
Conclusion No.
lt 40µm pair distance
12
Stimulus detection always gt false positive.
Classification accuracy improves with population
size considered.
small errors increased with size. large
errors decreased with size.
13
Summary
Spatially organized but probalistically
Correlated spiking, but variant
No discrete subpopulations observed.