Machine psychology: autonomous behavior, perceptual categorization and conditioning in a brain-based device

Area of Science:

• Neuroscience
• Artificial Intelligence
• Robotics

Background:

• Simultaneously analyzing all brain activity levels, from molecular to motor, in living animals is challenging.
• Synthetic neural modeling offers a method to study interactions between different levels of neural control.

Purpose of the Study:

• To design and evaluate a brain-based real-world device (Darwin VII) for synthetic neural modeling.
• To investigate neural mechanisms underlying complex behaviors like object recognition and conditioning.

Main Methods:

• Developed Darwin VII, a device with computer-simulated cortical and subcortical structures.
• Observed and analyzed neural activity across all levels of Darwin VII's architecture during real-world behavior.
• Examined behaviors including perceptual categorization and conditioned responses.

Main Results:

• Darwin VII enabled simultaneous analysis of all neural architecture levels during behavior.
• The device exhibited mechanisms for invariant object recognition and experience-dependent categorization.
• Analysis revealed insights into first- and second-order conditioning and learning rate effects.

Conclusions:

• Brain-based devices like Darwin VII allow exhaustive analysis of neuroanatomy and neural dynamics.
• These devices provide heuristics for understanding cortical interactions.
• They lay the groundwork for intelligent machines built on neurobiological principles.