Home - Research Teams - Neuronal mechanisms of spatial cognition

Neuronal mechanisms of spatial cognition

Our research

Where are we? Where do we want to go and how do we get there? Our research group is interested in how spatial information is represented in the brain, stored and used for efficient navigation. How do external (from the outside world) and internal (linked to our own movements) information combine to generate a cognitive map of our environment? How is our position and direction coded and constantly updated to navigate this map?What is the role of synaptic inputs and intrinsic neural properties in this process? We address these questions at the behavioral, circuit and single cell levels.


More info +More info -

Research interests

Our research group is interested in spatial navigation and memory. We use rodents as an animal model. Like humans, rodents can flexibly navigate in their environment, taking shortcuts or making detours if needed to reach their goal. This ability is thought to rely on an internal cognitive map. The activity of principal cells in the hippocampal formation is strikingly modulated by the position and orientation of an animal in its environment. Place cells in the hippocampus fire when an animal is in specific positions in its environment, grid cells in the entorhinal cortex fire when animals are at the crossings of a hexagonal grid covering the all environment and head direction cells fire when the animal’s head point towards a specific direction. Altogether, these cells could provide a neuronal substrate for the cognitive map.
Despite decades of research, which sensory information is important to activate place cells and whether all places are coded at a similar spatial resolution is unclear. To address this question we recently developed multi-shank silicon probe recordings in head-fixed mice navigating virtual reality environments (Bourboulou, Marti et al., 2019). Virtual reality allows a good control of the sensory cues available to an animal to orient itself while multi-shank silicon probe allow the recording of hundreds of cells simultaneously.
Not all cells are place cells in a given environment and the majority are in fact silent. Understanding the intracellular determinants of place and silent cell activation (or silence) in a given environment is fundamental to understand how specific internal representations of space are formed and stored in long-term memory. To address this question we use whole-cell patch-clamp recordings in vivo in animals navigating real and virtual environments (Lee, Epsztein and Brecht 2009; Epsztein, Lee et al., 2010; Epsztein et al., 2011). Of particular interest to us is how intrinsic cellular properties determine the involvement of single cells in spatial information coding (Morgan et al., 2019).
Altogether, we hope that our research will help us understand how cells and circuits for spatial navigation work and the origin of spatial orientation deficits observed in several pathologies such as temporal lobe epilepsy, attention deficits or Alzheimer’s disease.


Activity of different place cells while a rat forage in an linear environment. The path of the animal is depicted with a grey line. Different place fields are illustrated with colored oval zones. (image R. Bourboulou)

Specific projects

  • aRelative influence of internal and external sensory cues on spatial coding resolution and memory
  • bGrid cells contribution to distance coding and update of cognitive maps
  • cIntracellular determinants of place cells activation in new and familiar environments

Relative influence of internal and external sensory cues on spatial coding resolution and memory

PI: Jérôme Epsztein & Julie Koenig with Romain Bourboulou and Geoffrey Marti in collaboration with the team of Hervé Rouault

Every location in an environment is represented by the selective firing of an ensemble of place cells active there, among a larger population of silent neurons. This representation can be more or less precise depending on the size of this ensemble (with a higher proportion of place cells associated with a higher spatial coding resolution akin the number of pixels on a picture), the size of their place field (akin the size of the pixels) but also their spatial and temporal stability. The factors controlling spatial coding resolution are poorly understood. In this project we address this question taking advantage of recently developed virtual reality systems for rodents combined with high-density extracellular recordings (silicon probes). Virtual reality allows us a better control of the sensory cues that are available to animals for self-location. Our recent results (Bourboulou, Marti et al., 2019) suggest that spatial coding resolution could be modulated locally within the same environment. This could be relevant for navigating large scale/complex environments for rodents or memory space in humans.

Top and first person's view of a virtual reality environment (from Bourboulou, Marti et al., 2019)
Extracellular recording in the CA1pyramidal cell layer of a mouse during virtual reality navigation using a multi-shank silicon probe (8 shanks, 64 channels)
Example of two place cells with good (left) and less good (right) spatial coding resolutions. For each cell, left is the lap per lap raster plot and right is the time averaged mean tuning curve. The bold part of the tuning curve depicts the location of the place field. SI: Spatial Information, Out/In: Out of field versus In field firing Ratio.

Grid cells contribution to distance coding and update of cognitive maps

P.I.: Julie Koenig with Mathilde Nordlund

Navigation using an internal representation of the environment or a cognitive map requires allothetic information about the spatial arrangement of the landmarks and idiothetic (self-motion) information. Grid cells in the medial entorhinal cortex (MEC) exhibit a striking grid-like firing pattern that tesselate the environment and are supposed to be mainly driven by idiothetic cues in order to compute distances travelled in the environment. However, this point of view has been challenged by several studies showing that grid cells are highly sensitive to external visual cues suggesting that these cues might be important in the establishment of grid cells firing patterns. Finally, the grid cells population might be functionally more heterogeneous than previously thought as they are present in two populations of projecting cells in layer 2 of the MEC that are characterized by different morpho-functional properties: stellate and pyramidal cells.
Our general goal is to test how grid cells encode distance travelled and how they react to local change in the availability of environmental landmarks. To do so we take advantage of the recent development of virtual reality systems for rodents, an efficient tool to modify environments instantaneously and in a very controlled and reliable way.

Top: Examples of place cell (left) and grid cell (right) firing while an animal was randomly foraging for food in a square box for 20 min. For each cell: Left, the square is the box, the grey trace represents the trajectory of the animal during the recording period on which the locations of the animal for each spike fired by the recorded cell is reported by red dots. Right, Color coded firing rate maps. The box is divided in spatial pixels where the averaged firing rate for the recording period is calculated and normalized by the occupancy time. The colour codes for the firing rate: blue = 0 Hz; red = max firing rate. Place cells generally have single place fields while grid cells have multiple place fields located at the vertices of equilateral triangles covering the all environment. Adapted from Koenig et al., 2011. Bottom: different interconnected structures playing a role in spatial cognition.

Intracellular determinants of place cells activation in new and familiar environments

P.I.: Jerome Epsztein with Peter Morgan

At the cellular level, despite decades of study, the intracellular mechanisms responsible for the recruitment of a given cell into the assembly coding an environment are still poorly understood. Why are some cells active rather than silent in a given environment? Are these cells selected randomly or following specific rules? Slice experiments have shown that the input/output transformation in CA1 pyramidal cells is highly non-linear because CA1 pyramidal cells dendrites are endowed with voltage-gated conductances that can generate regenerative local or global dendritic spikes. Our previous intracellular recordings of CA1 pyramidal cells from freely moving rats indicated the presence of plateau potentials (a signature of dendritic regenerative events) and burst firing (a signature of increased intrinsic excitability) specifically in spatially coding cells (Epsztein et al., 2011). Based on the above results we propose that the inital level of intrinsic excitability of CA1 pyramidal cells is central to the spatial modulation of their firing rate and their recruitment into the cell assembly coding a new environment. We currently investigate the role of long-term plasticity of intrinsic excitability (Morgan et al., 2019) in the regulation of memory allocation.

A Top view of the maze. Animal position in the maze is determined by light emitting diodes. B. Trajectory of the animal in the environment (blue line) during the intracellular recording of a CA1 pyramidal cell. Red dots indicate animal position when the cell fires spikes. Most of the spikes occur when the animal is in the lower right part of the maze (grey shaded area). C. Membrane potential (black trace) and instantaneous speed of the animal (green trace) during three successive laps around the maze (green path in B). The cell fires at high frequency each time the animal crosses the grey shaded area (grey bars under the trace). This neuron is thus a place cell coding for the grey shaded area of this particular maze. D. Average membrane potential of the same cell over all laps around the maze in one direction plotted against the linearized position of the animal in the maze. We see a large bump or hill in the membrane potential at the location for the cell’s place field but not elsewhere in the maze. E. Same as in D for spiking activity.



Rouault lab

Robbe lab

Cossart lab

Crépel lab


Dr. Judith Makara,  Institute of Experimental Medicine of the Hungarian Academy of Sciences, Budapest, Hungary

Dr. Alex Roxin, Centre de Recerca Matemàtica, Barceona, Spain


Romain Bourboulou (PhD 2015-2019, now post-doc in the Barry lab, UCL, London)

Caroline Filippi (Engineer 2015-2019)

Geoffrey Marti (Post-doc 2015-2019)

François-Xavier Michon (PhD 2014-2018, now post-doc in the Lacaille lab, Montreal)

Peter Morgan (Post-doc 2014-2019)

David Ouedraogo (PhD 2009-2013, now working at Johnston & Johnston)




ERC starting grant FP7

Amidex "Rising star grant"


Human Brain Project



Nos publications

voir toutes les publications +