Representational Similarity Analysis for Tracking Neural Correlates of Haptic Learning on a Multimodal Device.

Area of Science:

• Neuroscience
• Human-Computer Interaction
• Sensory Perception

Background:

• Wearable haptic devices aim to enable communication via cross-modal associative learning.
• Neural correlates offer objective measures for haptic perception and learning assessment.
• Representational Similarity Analysis (RSA) can analyze neural representations of sensory information.

Purpose of the Study:

• To evaluate how neural representations of multimodal haptic cues change during association training using electroencephalography (EEG) and RSA.
• To investigate the initial phase of cross-modal associative learning: the perception of multimodal cues.
• To assess the efficacy of EEG-RSA in tracking shifts in haptic cue representation.

Main Methods:

• Trained a participant to map phonemes to multimodal haptic cues.
• Acquired EEG data before and after training to construct neural representational spaces.
• Compared neural spaces to theoretical perceptual and feature-based models using RSA.

Main Results:

• The perceptual model correlated better with pre-training neural data.
• The feature-based model showed stronger correlations with post-training neural data.
• These findings suggest training sharpens the sensory response to haptic cues.

Conclusions:

• EEG-RSA can capture shifts in neural representational spaces, indicating haptic learning.
• This approach shows promise for objectively tracking the learning process in haptic communication systems.
• Training appears to refine the neural processing of haptic cues.