Do things look the way they feel?

Commentary: “Multimodal Theories of Recognition and Their Relation to Molyneux's Question”

“…Altieri is also correct in endorsing criticisms I have raised (Schwenkler, 2012 , 2013 ; Connolly, 2013 ) against the resolution of Molyneux's question attempted by Held et al ( 2011 ). In this study, newly sighted individuals could match seen objects with seen objects and felt with felt, but could not match seen objects with felt ones, leading the authors to conclude that “the answer to Molyneux's question is likely negative,” as any innate link between vision and touch “is insufficient for reconciling the identity of the separate sensory representations” (Held et al, 2011 ).…”

Multimodal theories of recognition and their relation to Molyneux's question

“…One potential shortcoming of Held et al's ( 2011 ) study was that the newly sighted may have poor visual representations of depth cues (cf. Connolly, 2013 ; Schwenkler, 2013 ). This led Connolly ( 2013 ) and Schwenkler ( 2013 ) to discuss the possibility of re-running the experiment using raised line-drawings rather than 3D objects.…”

“…Connolly, 2013 ; Schwenkler, 2013 ). This led Connolly ( 2013 ) and Schwenkler ( 2013 ) to discuss the possibility of re-running the experiment using raised line-drawings rather than 3D objects. I agree that this is a viable improvement, although argue below that other predictions for the newly sighted can likewise get to the theoretical core of Molyneux's Question.…”

On Exploiting Haptic Cues for Self-Supervised Learning of Depth-Based Robot Navigation Affordances

“…Humans are known to be extremely good in the ability to learn a↵ordances, particularly because they can e ciently correlate visual and haptic cues [20,37]. Following the same rationale, this article presents a method for the self-supervised learning of navigation a↵or-dances, given the outcome of haptic interactions between the robot and the environment.…”