Related Experiment Video
Updated: Apr 5, 2026
07:34
A Simple Stimulatory Device for Evoking Point-like Tactile Stimuli: A Searchlight for LFP to Spike Transitions
Published on: March 25, 2014
10.3K
Learning touch preferences with a tactile robot using dopamine modulated STDP in a model of insular cortex
Ting-Shuo Chou1, Liam D Bucci2, Jeffrey L Krichmar3
1Department of Computer Sciences, University of California, Irvine Irvine, CA, USA.
Frontiers in Neurorobotics
|August 11, 2015
Summary
This study introduces CARL-SJR, a neurorobot that learns tactile preferences using a spiking neural network. The robot robustly learns associations between touch and color, mimicking animal behaviors and explaining hedonic touch mechanisms.
Area of Science:
- Neuroscience
- Robotics
- Artificial Intelligence
Background:
- Neurorobots facilitate studying neural mechanisms of behavior in real-world environments.
- Understanding somatosensory processing of noisy touch is crucial for human-robot interaction.
- Existing models often require heavily pre-processed sensory data.
Purpose of the Study:
- To investigate somatosensory system processing of noisy, real-world touch inputs using a novel neurorobot.
- To model the neural mechanisms underlying tactile preference and hedonic touch.
- To demonstrate robust learning in a neurorobot with a full-body tactile sensory area.
Main Methods:
- Introduction of CARL-SJR, a neurorobot with a full-body tactile sensory area designed for gentle touch interaction.
- Application of a spiking neural network (SNN) with neurobiologically inspired plasticity, modeling key brain regions (somatosensory cortex, prefrontal cortex, striatum, insular cortex).
- Utilizing dopamine-modulated Spike Timing Dependent Plasticity (STDP) and tuning the SNN for traveling waves of activity to process noisy, spatiotemporal data.
Main Results:
- CARL-SJR successfully learned associations between visual cues (conditioned stimuli) and touch patterns (unconditioned stimuli) despite noisy inputs.
- The model demonstrated robust learning, with insular cortex activity enabling control over touch preference without extensive pre-processing.
- Learned behaviors, including preferences for specific touch areas and directions, mirrored observed animal behaviors.
Conclusions:
- The developed neurorobot and SNN model provide insights into the neural mechanisms of tactile preference and hedonic touch.
- The study highlights the potential of SNNs with neurobiologically inspired plasticity for processing noisy sensory data in real-world robotic applications.
- CARL-SJR serves as a platform for future research into somatosensory processing and the development of more intuitive human-robot interactions.
