Hurrah! Hurrah! The ants going marching one by one, the
little one stopped to suck his thumb, and they all go marching down... in the
ground... to get out... of the rain...
Today we will discuss two papers that explore path
integration in the desert ant species, Cataglyphis. What did everyone think of
the 2006 Science paper? If you were
conducting related studies, what would you do next? Did anyone read the Wittlinger and Wolf, 2013 paper?
I thought it was fascinating to read about neural-coding of vector related information. I would have like to see an expanded experiment that mimicked the natural environment a little bit better. That is, a large flat sheet (the "landscape") that has the anthill on one corner, and a food item at some random location. After the ants found the food, modify their legs and see how the vector that they calculate to go back differs in direction. To account for visual effects, the ants could be blinded (similar to 2013 paper) or low/non-contrasting landscape used.
ReplyDeleteI thought that the 2013 paper was interesting in that it concluded that leg proprioceptive feedback is used for stride integration, even if large deviations from normal (i.e. amputation) exist. At the circuit level, this makes me wonder if this is another application of a sinusoidal encoding mechanism, which at least mathematically would be a plausible solution even with large "deviations."
That is an interesting point, John. Do you have an idea for a possible neurological circuit that could underlie/support the behavioral results?
DeleteI would propose a variation on Joe's idea. I picture a multi-level circuit arranged 360 degrees around a center neuron. The neuron, without any additional information, would signal "home". As the ant moves, the lowest level neuron fires information about direction, which is then encoded at a higher level neuron and potentiated. These higher level neurons surround the center neuron, and each synapses on the center. Thus, the center neuron can access and be influenced by information from each higher level neuron, and with an increasing number of neurons, increase the precision of vector directionality.
DeleteThe caveat to this circuit would be that there would have to be a "conscious" choice at some point to retrieve the information and act on it. That is, the center neuron is essentially "off" until a decided point, when it receives input from the other neurons and then outputs movement coordinates based on how all vector inputs average out amongst each other. This assumption may be plausible; the idea of a task-completion recognition mechanism is fairly common. For example, when I am on a hiking a mountain and reach the top I can recognize that I have completed my task and can turn around and head home; I can access my ability to turn around at any point along the way but I don't respond until I'm at the top. Likewise, recognition that an ant picked up a food item may allow him to access his coordinates and act on it.
I have been trying to visualize how a neural circuit could encode the "step integrator" and the best I can reason is some sort of central pattern generator that is potentiated during scavenging and then depotentiated as it runs home. Does anybody have any better ideas on how this could work circuit wise?
ReplyDeleteJoe,
DeleteI was wondering the exact same thing.
on an unrelated-to-ant-navigation note, (more on an ants-are-neat note) check these ants out: http://www.youtube.com/watch?v=ozkBd2p2piU
ReplyDeletei wonder how navigation works within ant hill tunnels, where both visual cues and polarized light cues are gone. which cues are the ants using there?
gravity?
DeleteOn a related note, I knew a professor at FSU that used to encase ant tunnels in plaster. He displayed them in the corridors of the Biology building and they were remarkable.
DeleteIn line with the unrelated note, here is a cool video of an excavation in the wild:
Deletehttps://www.youtube.com/watch?v=lFg21x2sj-M
i also wish more had been done from an experimental standpoint. for instance in the first paper, could they have achieved a "dose response" by manipulating the legs to variable lengths? chris
ReplyDeleteChris,
DeleteThat is an interesting idea. What additional information would the 'dose response' manipulation provide?
At the end of the Science paper, they suggest that the mechanism of the step integrator could be registering steps by proprioception or integration of activity via a walking pattern generator. The 2013 paper seems to refute the latter, as a walking pattern generator would presumably be disrupted by multiple amputation. Perhaps blocking proprioception in the ants could answer the former? Not sure how this could realistically be done though...
ReplyDeleteI like Joe's idea about increasing potentiation with each step and that on return to the nest the ant's neural circuitry switches to depotentiation.
I also thought it was somewhat interesting that the ants' vector cues didn't alter their path when moved out of the training box and into the test box. Unless they were very close to each, the x-y positioning of the ant would seemingly be off. Also interesting that they could remove the ants, do a physical manipulation, and put it back, and the ant still "remembers" how far to go. I wonder if they did some sort of mock/sham manipulation to the control ants...
Jared
Great points, Jared. In a more recent paper, the authors attempted to manipulate proprioception by shaving the hairs off the ants' body segment mechanoreceptors (ref.), with little to no effect on path integration. I added this paper to the Foraging and Feeding- Part 2 JMS folder on our course Dropbox.
DeleteTest 3
ReplyDelete