Related Experiment Videos
The information content of receptive fields
Thomas L Adelman1, William Bialek, Robert M Olberg
1Department of Molecular Biology, Princeton University, Princeton, NJ 08544, USA. tadelman@princeton.edu
Neuron
|November 19, 2003
Summary
Simple receptive field models capture most information processed by dragonfly target-selective descending neurons (TSDNs). This suggests these models are effective for understanding neural responses to visual stimuli.
Area of Science:
- Neuroscience
- Computational Neuroscience
- Insect Behavior
Background:
- Neural responses are often simplified using receptive fields or tuning curves.
- The accuracy of these simplified models in reflecting neural dimensionality reduction is debated.
- Target-selective descending neurons (TSDNs) in dragonflies control visually guided behaviors.
Purpose of the Study:
- To evaluate the completeness of receptive field descriptions for dragonfly TSDNs.
- To determine if simple receptive field models adequately capture stimulus information processed by TSDNs.
Main Methods:
- Extracellular multielectrode recordings were used to capture the activity of a population of TSDNs.
- The information content of single spikes was analyzed in relation to stimulus features.
- Independent instantaneous position and velocity receptive fields were modeled.
Main Results:
- Information from independent instantaneous position and velocity receptive fields accounted for 70%-90% of the total information in single spikes.
- This indicates that simple receptive field models capture a substantial portion of the stimulus-evoked neural response.
Conclusions:
- Simple receptive field models provide a near-complete description of the stimulus features driving TSDN responses.
- These findings support the utility of simplified models in understanding neural computations for behavior.