Transformation of the head-direction signal into a spatial code

Abstract

Animals integrate multiple sensory inputs to successfully navigate in their environments. Head direction (HD), boundary vector, grid and place cells in the entorhinal-hippocampal network form the brain's navigational system that allows to identify the animal's current location, but how the functions of these specialized neuron types are acquired remain to be understood. Here we report that activity of HD neurons is influenced by the ambulatory constraints imposed upon the animal by the boundaries of the explored environment, leading to spurious spatial information. However, in the post-subiculum, the main cortical stage of HD signal processing, HD neurons convey true spatial information in the form of border modulated activity through the integration of additional sensory modalities relative to egocentric position, unlike their driving thalamic inputs. These findings demonstrate how the combination of HD and egocentric information can be transduced into a spatial code.

Fig. 1

Spatial correlates of example HD neurons. a Two example HD neurons recorded simultaneously in the PoSub (top) and in the ADn (bottom). Polar plots indicate average firing rate in function of animal’s orientation. b Top left: position of the animal (gray) superimposed with location of the animal when the neuron spiked (red dots); top right, spatial tuning of the neuron where average firing rate at each location is represented as a colormap. Contrast displays the overall occupancy at each location. Bottom, place fields of the neuron during exploration along the borders when the walls are located on the left of the animal (left panel) or on the right (right panel). Inset show animal’s position and spike location in the two configuration during which a neuron governed only by its HD tuning curve is expected to fire. c HD, spatial and unbiased information per spike of the neuron shown in a (see Methods and Supplementary Fig. 1). d Top, regression coefficients of the generalized linear model applied to the binned spike train of the neuron displayed in a, consisting of walls (East, North, West or South) or relative wall position (Ego), both regression included expected instantaneous firing rate from HD tuning curve (see Supplementary Fig. 4). Bottom, schemas depicting the two types of behavioral variables used to regress the neuronal data on: left, specific border sensitivity (East, North, West or South) independently of animal’s head-direction; right, animal’s position relative to the closest wall (‘all on the right’ or ‘wall on the left’) within a ± 60° range of head direction. b′–d′ Same as b–d for a ADn HD neuron. e–f Additional examples of two wall-modulated HD neurons in the PoSub. Brain displayed in a © 2015 Allen Institute for Brain Science. Brain Explorer. http://mouse.brain-map.org/static/brainexplorer

Fig. 2

Distribution of behavioral correlates in ADn and PoSub cell populations. a Bottom, the amount of head-direction vs. unbiased spatial information is shown for all the HD neurons in the ADn (left) and HD and non-HD putative pyramidal neurons in the PoSub (right). HD cells are shown in green, border-modulated neurons as filled markers and HD-by-border as red filled markers. Top histograms display the marginal distribution of unbiased spatial information within each category: green for non-border HD neurons, red for HD-by-border neurons, dashed blue line for non-border, non-HD pyramidal neurons, plain blue for pyramidal border, non-HD neurons. b Distribution of border modulation coefficients for HD cells in the ADn (gray bars) and the PoSub (red line). The vertical dashed line indicates the threshold that was chosen to distinguish between border and non-border modulated neurons. c Proportion of border-modulated neurons for all cell categories (black bars). Numbers indicate the total number of neurons in each category (‘non’ stands for ‘non-HD’). d Average (±s.e.m.) unbiased information in ADn HD neurons and PoSub HD and non-HD pyramidal neurons. e Same as d for egocentric modulation of neurons. **p < 0.01, ***p < 0.001

Fig. 3

HD neurons in the PoSub, but not in the ADn, integrate both allocentric (vestibular) and egocentric (self-centered) information to build an unbiased spatial code. a Egocentric modulation of HD neurons in the PoSub correlate with border modulation in for PoSub HD neurons. b Same as a for HD cells of the ADn. c Correlation between egocentric modulation of HD neurons in the PoSub and the amount of unbiased spatial information. d Correlation of the maximal border modulation coefficient of each neuron with the coefficient associated with the opposite wall (left) or with the adjacent walls (right; the average of the two border coefficients is considered in this case). e Schematic diagram of the transformation of an allocentric HD signal into a spatial code for border by a combination with an egocentric signal

Fig. 4

Functional connectivity of spatially related cells is not rigid. a Left, pairwise neuronal correlation during non-REM vs. wake for PoSub cell pairs with low unbiased spatial information (bottom 33% percentile); right, same for cell pairs conveying high unbiased spatial information (top 33% percentile). b Left, cross-brain states correlations (non-REM vs. wake and REM vs. wake) in the PoSub for neuron pairs conveying low (blue) and high (orange) unbiased spatial information; right, same for ADn cell pairs. **p < 0.01; error bars indicate 95% confidence interval

Fig. 5

Anatomical organization of neural information relevant to navigation. Schematic diagram of HD signal transformation to spatial information in a hierarchical (left) or distributed (right) fashion