Rats were made to 'walk' left while moving right (by putting them on a small treadmill mounted on a model train), while recording place cells. Does theta phase continue to correlate with absolute position within the place field? No. Phase precession is 'reversed'. Likewise, theta sequences start from where the rat came from, and extend to where he is going, regardless of the direction he faces while running. Interestingly contradicts most network models of theta sequences.
Neat approach with convincing evidence that head direction cells actual fire based on head direction (as the name would suggest) rather than direction of travel. Left a lot of room for further studied regarding theta phase, which I think could have been addressed more clearly (see criticisms).
hsw28 (3) 3 years ago
This is a creative paper with what seems to be a pretty straightforward, interpretable finding. I especially like that they spent the discussion directly comparing their results to large number of theoretical models, which clearly motivated the study, in order to constrain that space a bit. I thought the experimental design was simple and well motivated. It certainly had flaws, but quite elegantly decoupled sensory and motor streams that are almost always strongly coupled during normal animal behavior. In the sensory space, by running the train backward, while the animal ran 'forward' on the treadmill, propreoceptive (internal reference frame) sensory input provided competing, dissonant information compared to visual, auditory, olfactory (external reference frame) input. Further, by turning the treadmill off while the animal was pulled around by the train, the influence of motor output could also be decoupled from the animal's velocity/location in space, and prevented potential feedback onto hippocampal activity from motor circuit activation.
jpn (6) 3 years ago
Theta sequences flips while head direction signal remains forward - very nice demonstration that theta sequence direction isn't a head-direction biased attractor network sequence.
Paper left quite a few unanswered questions that perhaps could have been addressed without too much extra work: 1) I would have liked to have seen results for treadmill running without train movement, as there were some differences in results between "free running" and treadmill-on-train running (for example, firing fields undergo random remapping between forward travel and free running but NOT between forward and backward travel) so a direct comparison would have been useful. 2) Similarly, I would have liked to have seen all the results for backwards travel when the animal was just being moved by the train but not running on the treadmill. In Supp fig 5, there are theta phase results for this situation, but no results for composition of place fields, head direction cells, etc. 3) It would have been interesting to see if theta phase or place cell fields during backwards movement shifted with familiarity with the track, and/or with exposure to external room cues that could be maintained during both backwards and forwards travel. It would also be cool (since it's an area of interest to me) to see what replay looks like after backwards travel, but that is a bit outside the scope of this paper.
hsw28 (3) 3 years ago
Still true: Earliest spikes happen at latest phases. Spike burst timing is slightly faster than theta freq. So, is reverse theta sequence in backward-running rat a trivial finding?
I have two criticisms, one major and one minor. Starting small, the use of rainbow color maps is highly misleading due to the non-uniform sensitivity of human vision across the visible spectrum, the non linear progression of luminescence across colors, and artificial 'bands' that it introduces into data. This document outlines the problem and solutions: http://root.cern.ch/drupal/content/rainbow-color-map). The major criticisms i have is that the the trained moved at a constant velocity regardless of the effort that the rat put forth on the treadmill. I think that coupling the velocity of the treadmill to the velocity of the train (via a direct or inverted mapping) would be a major improvement because it would link the animal's motor commands to its physical location in terms of the 'allocentric' frame of reference. Perhaps hippocampal circuitry is 'disregarding' the allocentric frame of reference because it 'realizes' that the mutual information between motor commands and allocentrially-referenced position is ~0. If the train's movement was coupled to motor output, forward or backward, this would not be the case.
jpn (6) 3 years ago