@jhawkins describes his latest thinking about the potential for reference frame transformations occurring in the thalamus.
Could it be that the parvocellular inputs are needed to drive the reference frame transformations for the magnocelluar inputs? By that I mean that the detailed information received on the parvocellular pathway could provide the orientation input necessary to properly decode the flow information arriving on the magnocellular pathway. I can tilt my head to the side, and still walk in a straight line. My interpretation of the flow data and sense of spatial orientation remain correlated to my actual movement.
I really enjoyed watching this research meeting! Very interesting ideas.
It makes me look for more info about thalamocortical circuitry and I found this comprehensive paper with a great illustration (not yet peer reviewed):
Probably worth scanning/reading even if it mainly focuses on the anatomy & connectivity rather than the dynamics of the thalamus during active sensing that Jeff & Subutai (and me!) are looking for.
Toward the end of this meeting there was a discussion between Jeff and Marcus about the transformation of sensory input and how individual columns may perform transformations that would seem incoherent when we imagine the reassembly of those transformations. This reminded me of earlier discussions in a research meeting where the problem of dealing with non-linear transformations (homeomorphisms or “rubber-sheet geometry”) came up. The ability of individual columns to recognize relations between locally transformed sensory patches might also relate to the system dealing with the sorts of non-linear transformations that seemed to upset the apple cart in earlier discussions.