 |
|||||
|
|||||
|
|||||
|
|||||
|
|||||
Electronic Letters to:
- BehavioralSystemsCognitive:
Vincent Hok, Pierre-Pascal Lenck-Santini, Sébastien Roux, Etienne Save, Robert U. Muller, and Bruno Poucet- Goal-Related Activity in Hippocampal Place Cells
J. Neurosci. 2007; 27: 472-482[Abstract] [Full text] [PDF]
eLetters: Submit a response to this article
Electronic letters published:
For goal scoring, the right place and the right time are matters of context
- Giorgio A. Ascoli, Alexei V. Samsonovich (8 February 2007)
For goal scoring, the right place and the right time are matters of context 8 February 2007 
Giorgio A. Ascoli,
Professor
Krasnow Inst. for Advanced Study, Psycology Dept., & Neuroscience Program, George Mason Univ.,
Alexei V. SamsonovichSend letter to journal:
Re: For goal scoring, the right place and the right time are matters of context
ascoli{at}gmu.edu Giorgio A. Ascoli, et al.
The relation between the hippocampal spatial code and internal representations of goals has been a topic of discussions since the discovery of place cells. Today most neuroscientists agree that the firing of hippocampal place cells in an alert animal encodes for the generalized context in which the subject is situated rather than the spatial location only. This view is consistent with the so-called remapping phenomena: the hippocampus may exhibit very different spatial codes for the same environment in behavioral episodes that differ by their paradigms, and even in different phases of the same paradigm performed repeatedly. In particular, the generalized context encoded by place cells could include the prospect of reward.
The evidence presented by Hok et al. supports this position. When the firing patterns are decomposed into different phases of the paradigm (Fig. 1D), two distinct spatial maps become apparent: the foraging map (Fig. 1D, right panel) and the goal-zone map (Fig. 1D, small circle in the middle panel), rather than one persistent map with excess firing at the goal location (Fig. 1D, left panel). In fact, the foraging map shows no surplus activity at the goal-zone. We surmise that the goal-zone map only exists for a short temporal window as a global hippocampal state, when the rat expects a reward in the goal-zone, but not during the larger time interval labeled by the authors as "navigation". Given this assumption, this map would be limited in space simply because it is limited in time. The goal- zone map is only defined and makes sense for the small circle, exactly as the foraging map is defined and makes sense for the large circle. The fact that most cells show up on the goal-zone map follows from the general observation that most place cells active on one map typically fire on another map in the same environment.
This parsimonious explanation invites the hypothesis that the excessive goal-zone firing could play a functional role in learning how to navigate to the goal-zone. In the model of hippocampal navigation we recently proposed (Samsonovich & Ascoli, 2005), any CA1 cell with a place field in a familiar environment is potentially available as a goal cell. The model predicts that place cells that fire on the way to the (actual or potential) goal get reactivated when the goal location is reached, enabling reinforcement learning. Available data indicate that this necessary reactivation occurs during sharp waves. The finding of Hok et al. suggests that reactivation could happen during theta, possibly under circumstances specific to this paradigm. Interestingly, the authors noticed switching from theta-I to theta-II in the goal-zone, which in agreement with the observed synchrony of firing might be associated with sharp waves.
As a key component of the spatial memory system, the hippocampus is frequently implicated in spatial navigation, and yet direct evidence of this function is missing. What mechanism (if any) connects the hippocampal place code with the ability to navigate towards a goal? In our model, the same neural mechanism, based on reinforcement learning, linked spatial and mnemonic functions by implementing episodic memory retrieval as contextual pathfinding (Samsonovich & Ascoli 2005). The results of Hok et al. could thus extend to human episodic memory, implying that most hippocampal pointers, rather than only one or few, should activate when subjects actively engage in context retrieval.
Samsonovich AV, Ascoli GA ( 2005) A simple neural network model of the hippocampus suggesting its pathfinding role in episodic memory retrieval. Learning & Memory 12(2):193–208.
|
|
|
|
 |
|