Poster: Organization of the neuronal assemblies in the anterior thalamus coding for head direction 326.14 (Mon Morning)
Authors: A. Peyrache, M. Lacorix, P. Petersen, G. Buzsaki; NYU
Head-direction cells are neurons that fire when ever a rat’s head is pointed in a particular direction. Discovered by Jim Ranck and first reported at SfN 29 years ago today’s findings are a major update, confirming and extending the cohesive properties of head-direction cell networks.
Clusters of Head-direction cells are found in post-subiculum, anterior thalamus and lateral mammillary nucleus. Several studies have proposed that the set of head-direction cells form an attractor, where the output of the set always points one head direction. That is, the output always points somewhere around the full-circle representation showing “head-angle with respect to environment”. Support of this is the consistent finding that pairs of head-direction cells appear to be rigidly offset from each other in their preferred direction.
Peyrache and colleagues are among the first to record from large sets of head direction cells. Using a wafer-probe electrode they are able to record up to 30 head-direction cells simultaneously from the anterior thalamic nucleus. Using Bayesian analysis of these relative large sets the heading direction of the attractor network can be inferred on a moment-by-moment basis. Peyrache et al’s results largely confirm the attractor dynamics of the set of head-direction cells.
In addition to demonstrating that the output of the network continually produces a clear and unique heading solution, there are two surprising findings. One is that the attractor dynamics is maintained during sleep. That is, even though the rat’s head is not moving, the output of the network continues to calculate a clear and ever-changing heading direction. During slow-wave sleep head-direction output varies continuously at speeds approximately 10x awake speeds. During REM sleep smooth transitions are maintained, but the speed of transition is similar to awake speed.
The second surprising finding is that pairs of simultaneously recorded anterior thalamic neurons had high cross correlations at very short (>50 msec) time scales. Peyrache and colleagues interpret this result as suggesting that attractor circuitry resides, in part, in the anterior thalamus. Although this is suggestive, the location of the attractor circuit remains uncertain, and may be distributed across several sites.
Gyorgi Buzsaki (left), Alex Peyrache and poster
Corona Radiata 
Title: “Assembies” ?
Is sleep variation just random jitter, or significant deviation from actual orientation?
Not random jitter. The head direction network is called a “circular attractor”. The behavior of the firing of members of the set of cells is organized in a circular pattern such that, during awake behavior, the firing of one cell is aways followed either by its neighbor on the right or neighbor on the left. So there are certain sequences for left head turning or right head turning (turning head to the left or right, not turning one of two heads). So, during awake state, sequences are highly constrained. During sleep could have seen “random” or unpredictable sequences. What they saw was the same restricted set of firing sequences seen in the awake state, even though the head was not turning. Clear? I should have been more explicit.
Also, the sleep behavior is deviant from actual orientation. the head is basically still, but the behavior of the attractor appears as if the head is moving. In fact, during slow-wave-sleep, the sequences are much more rapid than narwal. What they don’t know is if this swinging pattern is used by the rest of the brain. is it ‘noise’ or do other elements respond? If the animal is dreaming, is it dreaming that its head is turning?