Download presentation
Presentation is loading. Please wait.
Published byHilary Ward Modified over 9 years ago
1
2nd TAC Meeting Christian B. Mendl Tim Gollisch Lab Neuronal Coding in the Retina and Fixational Eye Movements Neuronal Coding in the Retina and Fixational Eye Movements April 22, 2010
2
Outline Review of last TAC meeting Informative spike response features Latency coding by cell pairs Modeling response features Outlook
3
Review of Last TAC Meeting Fixational eye movements, microsaccades Counteract visual perception fading Enhancement of spatial resolution Last TAC meeting: information theory: mutual information, synergy → use as screening tool To-do: – stimulus variation: grating instead of border – neuronal model building – decoding strategies
4
Informative Spike Response Features Observed spike responses of a single cell
5
Various Response Types a)b) d)c)
6
Informative Spike Features (cont) Observed spike responses of a single cell
7
Informative Spike Responses: Number of Spikes/Trial
8
Informative Spike Responses: Internal Structure ISI (inter-spike-interval)
9
Informative Spike Responses: Latency
10
Latency Coding by Cell Pairs Latency emerges as most informative spike response feature Timing reference? (Brain doesn’t know stimulus onset) → Need several cells
11
Cell Pairs: Experimental Data
12
Relative Latency time intervals accessible to readout by higher brain regions
13
Cell Pairs: Latency Scatter Plot K-means clustering: relative weight of off- diagonal elements: 19.1%
14
Global Drift Correction
15
Drift-Corrected Latency Scatter Plot K-means clustering: relative weight of off-diagonal elements: 9.6%
16
Latency Correlations Observation: global latency drift leads to (artificial) correlations, can correct for that Question: cells internally interacting on short-term scale? → Compare spikes shuffled by one trial
17
Latency Correlations (cont) shuffled version: no correlations
18
Latency Correlation Statistics
19
Conclusions Latency Coding Use latency instead of spike count and inter-spike- interval High information content in latency data from two cells Correlations might improve coding
20
Comparison with LN Models
21
LN Models (cont)
22
Conclusions Modeling Qualitative agreement But still much room for improvement, latency data on 10 ms scale not reproduced Gain control might be able to reproduce experimental spike histogram
23
Outlook Fixational eye movements have been reported in Salamander But precise quantification still missing → Search coil setup
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.