Answers are grounded in this paper and cite their sources.
• Patterns of (un-)touchings of hands, objects and ground can code action categories. • Robots can effectively use these patterns to recognize and respond to human actions. • Patterns relate to perceived action similarity in Inverse Multidimensional Scaling. • Left parietal area codes patterns as shown by Representational Similarity Analyses. • Touching-Untouching sequences organize parietal action representation. At an abstract temporospatial level, object-directed actions can be described as sequences of touchings and untouchings of objects, hands, and the ground. These sparse action codes can effectively guide automated systems like robots in recognizing and responding to human actions without the need for object identification. The aim of the current study was to investigate whether the neural processing of actions and their behavioral classification relies on the action categorization derived from the touching-untouching structure. Here we show, using a representational similarity analysis of functional MRI data from two experiments, that action representations in left anterior intraparietal sulcus (aIPS) are particularly associated with this categorization of touching-untouching structures. Within the examined action observation network, only the touching-untouching category model selectively correlated with the representational profile of the left aIPS. The behavioral results showed a significant relation between the touching-untouching structure and the observers’ judgments on the similarity of actions with weakly-informative objects. Extending prior research on touchings and untouchings as meaningful anchor points for explicit action segmentation, our findings suggest that touching-untouching sequences serve as an organizing principle in inferior parietal action representation.
No takes yet. Share an insight, caveat, or question.
Pomp et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: