Real-time human-robot interaction underlying neurorobotic trust and intent recognition

Area of Science:

• Neuroscience
• Computational Biology
• Robotics

Background:

• Trust is crucial in social interactions and has neurological underpinnings.
• Oxytocin, a neurotransmitter, influences trust and fear by inhibiting the amygdala.
• The precise neural circuits mediating oxytocin's effect on trust remain unclear.

Purpose of the Study:

• To propose the first biologically realistic computational model of trust.
• To simulate trust mechanisms using spiking neurons in a real-time human-robot interaction scenario.
• To investigate the role of oxytocin's cellular structure in modulating trust.

Main Methods:

• Developed a biologically realistic model of oxytocin neurons with characteristic triple apical dendrites.
• Simulated nearly 100,000 spiking neurons in real-time within a human-robot interaction framework.
• Incorporated a Gabor mechanism for visual processing and modeled amygdala inhibition.

Main Results:

• The model demonstrated that oxytocin cell architecture directly inhibited amygdala firing as trust was established.
• This inhibition led to a simulated willingness for object exchange between a virtual neurorobot and a human.
• The system accurately trusted or distrusted human actors based on movement imitation.

Conclusions:

• The study presents a novel computational model for trust, linking neurophysiology to behavior.
• The findings elucidate a potential cerebral microcircuitry for oxytocin-mediated trust.
• The model provides a foundation for further research into the neural basis of social cognition and human-robot interaction.