Related Experiment Videos
On the optimal control of behaviour: a stochastic perspective
1Department of Ophthalmology, Great Ormond Street Hospital for Children NHS Trust, Institute of Child Health, University College London, UK. chris@vissci.ion.ucl.ac.uk
Journal of Neuroscience Methods
|October 9, 1998
Summary
Evolution optimizes behavior through nervous system control, minimizing survival risks. Goal-directed movements like saccades demonstrate this, with the cerebellum potentially using real-time optimization.
Area of Science:
- Neuroscience
- Evolutionary Biology
- Control Theory
Background:
- Evolutionary processes drive organisms towards optimal fitness.
- Nervous systems are hypothesized to evolve for optimal behavioral control.
- Minimizing future survival hazards is a key aspect of this optimization.
Purpose of the Study:
- To investigate the role of optimization in nervous system control of behavior.
- To explain goal-directed movements, such as saccades, through an optimization framework.
- To explore the mechanisms underlying real-time behavioral optimization in the cerebellum.
Main Methods:
- Analysis of goal-directed saccade trajectories.
- Comparison of different optimization strategies (e.g., minimizing flight-time vs. error).
- Modeling of intra-movement trajectories using control theory principles.
Main Results:
- Minimizing total movement flight-time better explains saccade behavior than minimizing primary movement error.
- Smooth, low-bandwidth velocity profiles are more consistent with observations than bang-bang control.
- Minimum-time behaviors suggest a cerebellum-based, unreferenced optimization process.
Conclusions:
- The nervous system optimizes behavior to minimize survival risks.
- The cerebellum may employ real-time, unreferenced optimization, possibly using stochastic gradient descent.
- Both explorative and exploitative behaviors are necessary for this optimization process.