Related Experiment Video
Updated: Jul 27, 2026
13:40
Examining Local Network Processing using Multi-contact Laminar Electrode Recording
Published on: September 8, 2011
A comparison of methods used to detect changes in neuronal discharge patterns
P R Churchward1, E G Butler, D I Finkelstein
1Neurosciences Department, Monash Medical Centre, Melbourne, Australia.
Journal of Neuroscience Methods
|November 14, 1997
Summary
Visual inspection reliably detects changes in neuronal discharge patterns during voluntary movements. Artificial Neural Networks confirm this consistency, outperforming cumulative sums and maximum likelihood methods.
Area of Science:
- Neuroscience
- Computational Neuroscience
- Motor Control
Background:
- Understanding neuronal activity during voluntary movement is crucial for motor control research.
- Accurate analysis of neuronal discharge patterns is essential for interpreting neural signals.
- Existing methods for analyzing neuronal firing rates have limitations in real-time application.
Purpose of the Study:
- To compare the effectiveness of visual inspection, artificial Neural Networks (NNs), cumulative sums (CUSUMs), and maximum likelihood methods in analyzing neuronal discharge patterns.
- To assess the reliability and consistency of human observers in classifying neuronal activity.
- To evaluate the efficiency of NNs in detecting changes in neuronal firing rates.
Main Methods:
- Recorded discharge patterns of two thalamic neurons in a conscious monkey during wrist movements.
- Classified neuronal discharge states (background, 'on', 'off') by three researchers to create a 'standard output'.
- Modeled the standard output using a back-propagation NN and compared its performance with CUSUMs and maximum likelihood analyses.
Main Results:
- High correlation (r > 0.99) between the NN and the standard output, indicating NN's accuracy in modeling human classification.
- Lower correlations for CUSUMs (r = 0.06) and maximum likelihood (r = 0.36) compared to the standard output.
- NNs demonstrated high efficiency (r > 0.99) in detecting changes in neuronal activity states and required minimal training data (12 trials).
Conclusions:
- Visual inspection is a reliable and superior method for detecting timing and state changes in single-trial neuronal discharge compared to CUSUMs and maximum likelihood.
- Artificial Neural Networks effectively confirm the consistency of visual inspection in analyzing neuronal discharge patterns.
- NNs show promise as a valuable tool for objective and efficient analysis of neural activity in neuroscience research.
